| |
 |
Picture encoding method, picture encoding apparatus and picture recording medium |
| RE40679 |
Picture encoding method, picture encoding apparatus and picture recording medium
|
|
| Patent Drawings: | |
| Inventor: |
Oda |
| Date Issued: |
March 24, 2009 |
| Application: |
09/474,479 |
| Filed: |
December 29, 1999 |
| Inventors: |
Oda; Tsuyoshi (Chiba, JP)
|
| Assignee: |
Sony Corporation (Tokyo, JP) |
| Primary Examiner: |
Rao; Andy |
| Assistant Examiner: |
|
| Attorney Or Agent: |
Frommer Lawrence & Haug LLPFrommer; William S.Presson; Thomas F. |
| U.S. Class: |
375/240.13; 375/240.03; 375/240.04; 375/240.05; 375/240.06; 375/240.07; 375/240.14; 375/240.18 |
| Field Of Search: |
375/240.01 |
| International Class: |
H04N 7/18 |
| U.S Patent Documents: |
|
| Foreign Patent Documents: |
0 475 251; 0 493 130; A-0493130; A-0509576; A-05009576; 0 520 789; A-0535960; 0 613 306 |
| Other References: |
Grueneberg K etal: "Hardware Implementation of the Framestore and Data Rate Control for a Digital HDTV-VCR" Signal Processing of HDTV.Proceedings of the International Workshop on HDTV, vol. 4, Nov. 18, 1992, pp. 51-58, XP000199113. cited by other.
Viscito E et al: "A Video Compression Algorithm With Adaptive Bit Allocation and Quantization" Visual Communication and Image Processing '91: Visual Communication, Boston, Nov. 11-13, 1991, vol. Part 1, No. vol. 1605, Nov. 11, 1991, pp. 58-72,XP000479218 Kou-Hu Tzou; Toshio Koga. cited by other.
Netra Vali A et al: "The Digital Spectrum-Compatible HDTV System" Signal Processing, Image Communication, Elsevier Science Publishers, Amsterdam, NL, vol. 4, No. 4 / 5, Aug. 1, 1992, pp. 293-305, XP000293750 ISSN: 0923-5965. cited by other.
Abbas Razavi et al: "VLSI Implementation of an Image Compression Algorithm With a New Bit Rate Control Capability" Digital Signal Processing 2, Estimation, VLSI, San Francisco, Mar. 23-26, 1992, vol. 5, No. Conf. 17, Mar. 23, 1992, pp. 669-672,XP000341509 Institute of Electrical and Electronics Engineers. cited by other.
Wen-Hsiung Chen et al: "Adaptive coding of monochrome and color images" IEEE Transactions on Communications, Nov. 1977, USA, vol. COM-25, No. 11, 1977, pp. 1285-1292, XP002085968 ISSN: 0090-6778. cited by other.
Cheng-Tie Chen et al: "Hybrid Extended MPEG Video Coding Algorithm for General Video Applications" Signal Processing. Image Communication, Elsevier Science Publishers, Amsterdam, NL, vol. 5, No. 1 / 2, Feb. 1, 1993, pp. 21-37, XP000345611 ISSN:0923-5965. cited by other.
Proceedings of the Picture Coding Symposium (PCS), Lausanne, Mar. 17-19, 1993, No., Mar. 17, 1993 Swiss Federal Institute of Technology, pp. 19.3/A-19.3/B, XP 000346426 Nicoulin A. et al `Feed-Back Free Rate Control for Digital Video Coding` * p.19.3.A, right column, last paragraph -p. 19.3.B. left column, paragraph 1*. cited by examiner. |
|
| Abstract: |
This invention employs a scheme to allow an input video signal to undergo encoding, e.g., predictive encoding, DCT processing, quantization at fixed quantization step size and variable length encoding to generate first encoded data to determine (calculate) allocated code quantity every frame or every GOP on the basis of data quantity every predetermined time, e.g., every frame or every GOP of the first encoded data and total quantity of usable data to encode the input video signal every predetermined time on the basis of the allocated code quantity to generate second encoded data. Thus, variable rate encoding such that encoding rate changes every predetermined time is realized. As a result, even if pictures (frames) of complicated are successive, there is no possibility that quantization step size is caused to be large with respect to these pictures as in the conventional apparatus. Thus, uniform high picture quality can be obtained through the entirety. Further, since second encoded data obtained in a manner as described above has variable rate, in the case where such encoded data is recorded onto picture recording media, limited memory capacity can be effectively used, and recording time of picture recording media can be prolonged. In addition, picture data of high picture quality uniform over the entirety can be reproduced from the picture recording media. |
| Claim: |
What is claimed is:
1. An encoding method, comprising the steps of: receiving an input video signal; selectively encoding at least a portion of said input video signal by intra-frame encodingor predictive encoding to generate first encoded data, said predictive encoding including forward predictive encoding and backward predictive encoding; transform encoding said first encoded data to generate first coefficient data; quantizing said firstcoefficient data by a fixed step size; variable length encoding said first quantized data to generate a first bit stream; determining an encoding rate of said first bit stream every GOP as a function of a data quantity of intra-frame andforward-predictive encoded pictures only in the GOP; selectively encoding said input video signal by intra-frame or predictive encoding to generate second encoded data; transform encoding said second encoded data to generate second coefficient data; setting a variable step size according to the encoding rate determined every GOP; quantizing said second coefficient data by said variable step size; variable length encoding said second quantized data to generate a second bit stream; and outputtingsaid second bit stream.
2. An encoding apparatus, comprising: means for receiving an input video signal; means for selectively encoding at least a portion of said input video signal by intra-frame encoding or predictive encoding to generate first encoded data, saidpredictive encoding including forward predictive encoding and backward predictive encoding; means for transform encoding said first encoded data to generate first coefficient data; means for quantizing said first coefficient data by a fixed step size; means for variable length encoding said first quantized data to generate a first bit stream; means for determining an encoding rate of said first bit stream every GOP as a function of a data quantity of intra-frame and forward-predictive encodedpictures only in the GOP; means for selectively encoding said input video signal by intra-frame or predictive encoding to generate second encoded data; means for transform encoding said second encoded data to generate second coefficient data; meansfor setting a variable step size according to the encoding rate determined every GOP; means for quantizing said second coefficient data by said variable step size; means for variable length encoding said second quantized data to generate a second bitstream; and means for outputting said second bit stream.
3. A recording medium on which there is recorded a second bit stream obtained by: receiving an input video signal; selectively encoding at least a portion of said input video signal by intra-frame encoding or predictive encoding to generatefirst encoded data, said predictive encoding including forward predictive encoding and backward predictive encoding; transform encoding said first encoded data to generate first coefficient data; quantizing said first coefficient data by a fixed stepsize; variable length encoding said first quantized data to generate a first bit stream; determining an encoding rate of said first bit stream every GOP as a function of a data quantity of intra-frame and forward-predictive encoded pictures only in theGOP; selectively encoding said input video signal by intra-frame or predictive encoding to generate second encoded data; transform encoding said second encoded data to generate second coefficient data; setting a variable step size according to theencoded data rate determined every GOP; quantizing said second coefficient data by said variable step size; variable length encoding said second quantized data to generate a second bit stream; and recording said second bit stream on said recordingmedium.
4. An encoding method, comprising the steps of: receiving an input video signal; selectively encoding at least a portion of said input video signal by inter-frame encoding or predictive encoding to generate first encoded data, said predictiveencoding including forward predictive encoding and backward predictive encoding; transform encoding said first encoded data to generate first coefficient data; quantizing said first coefficient data by a fixed step size; variable length encoding saidfirst quantized data to generate a first bit stream; determining a difficulty of encoding said first bit stream every GOP based on inter-frame and forward-predictive encoded pictures only in the GOP; calculating an encoding rate frame said difficultyof encoding determined every GOP; selectively encoding said input video signal by intra-frame or predictive encoding to generate second encoded data; transform encoding said second encoded data to generate second coefficient data; quantizing thesecond coefficient data by a step size set according to said calculated encoding rate; variable length encoding said second quantized data to generate a second bit stream; and outputting said second bit stream.
5. An encoding apparatus, comprising: means for receiving an input video signal; means for selectively encoding at least a portion of said input video signal by intra-frame encoding or predictive encoding to generate first encoded data, saidpredictive encoding including forward predictive encoding and backward predictive encoding; means for transform encoding said first encoded data to generate first coefficient data; means for quantizing said first coefficient data by a fixed step size; means for variable length encoding said first quantized data to generate a first bit stream; means for determining a difficulty of encoding said first bit stream every GOP based on intra-frame and forward-predictive encoded pictures only in the GOP; means for calculating an encoding rate from said difficulty of encoding determined every GOP; means for selectively encoding said input video signal by intra-frame or predictive encoding to generate second encoded data; means for transform encodingsaid second encoded data to generate second coefficient data; means for quantizing the second coefficient data by a step size set according to said calculated encoding rate; means for variable length encoding said second quantized data to generate asecond bit stream; and means for outputting said second bit stream.
6. A recording medium on which there is recorded a second bit stream obtained by: receiving an input video signal; selectively encoding at least a portion of said input video signal by intra-frame encoding or predictive encoding to generatefirst encoded data, said predictive encoding including forward predictive encoding and backward predictive encoding; transform encoding said first encoded data to generate first coefficient data; quantizing said first coefficient data by a fixed stepsize; variable length encoding said first quantized data to generate a first bit stream; determining a difficulty of encoding said first bit stream every GOP based on intra-frame and forward-predictive encoded pictures only in the GOP; calculating anencoding rate from said difficulty of encoding determined every GOP; selectively encoding said input video signal by intra-frame or predictive encoding to generate second encoded data; transform encoding said second encoded data to generate secondcoefficient data; quantizing the second coefficient data by a step size set according to said calculated encoding rate; variable length encoding said second quantized data to generate a second bit stream; and recording said second bit stream on saidrecording medium.
7. An encoding method, comprising the steps of: receiving an input video signal; selectively encoding at least a portion of said input video signal by intra-frame or predictive encoding to generate first encoded data representing intra-frameor predictive encoded pictures, respectively; transform encoding said first encoded data to generate first coefficient data; quantizing said first coefficient data by a fixed step size; variable length encoding said first quantized data to generate afirst bit stream; counting a data quantity of said first bit stream every predetermined time to indicate a difficulty of encoding; determining said difficulty of encoding said first bit stream based on intra-frame and forward-predictive encodedpictures only; calculating an allocated code quantity for each unit of predetermined time as a function of said difficulty of encoding so that said allocated code quantity is set to be larger for complex pictures and smaller for simple pictures; selectively encoding said input video signal by intra-frame or predictive encoding to generate second encoded data; transform encoding said second encoded data to generate second coefficient data; quantizing said second coefficient data at a step sizeset in response to said allocated code quantity; variable length encoding said second quantized data to generate a second bit stream; and outputting said second bit stream.
8. A method according to claim 7, wherein the fixed quantization step size equals one.
9. A method according to claim 7, further comprising the step of storing said second bit stream in a buffer before it is output, wherein said quantization step size is further based on a quantity of said second bit stream stored in said buffer.
10. A method according to claim 7, wherein said predetermined time is a frame.
11. An encoding apparatus, comprising: means for receiving an input video signal; means for selectively encoding at least a portion of said input video signal by intra-frame or predictive encoding to generate first encoded data representingintra-frame or predictive encoded pictures, respectively; means for transform encoding said first encoded data to generate first coefficient data; means for quantizing said first coefficient data by a fixed step size; means for variable lengthencoding said first quantized data to generate a first bit stream; means for counting a data quantity of said first bit stream every predetermined time to indicate a difficulty of encoding; means for determining said difficulty of encoding said firstbit stream based on intra-frame and forward-predictive encoded pictures only; means for calculating an allocated code quantity for each unit of predetermined time as a function of said difficulty of encoding so that said allocated code quantity is setto be larger for complicated pictures and smaller for simple pictures; means for selectively encoding said input video signal by intra-frame or predictive encoding to generate second encoded data; means for transform encoding said second encoded datato generate second coefficient data; means for quantizing said second coefficient data at a step size set in response to said allocated code quantity; means for variable length encoding said second quantized data to generate a second bit stream; andmeans for outputting said second bit stream.
12. An apparatus according to claim 11, wherein the fixed quantization step size equals one.
13. An apparatus according to claim 11, further comprising a buffer for storing said second bit stream before it is output, wherein said quantization step size is further based on a quantity of said second bit stream stored in said buffer.
14. An apparatus according to claim 11, wherein said predetermined time is a frame.
15. A recording medium on which there is recorded a second bit stream obtained by: receiving an input video signal; selectively encoding at least a portion of said input video signal by intra-frame or predictive encoding to generate firstencoded data representing intra-frame or predictive encoded pictures, respectively; transform encoding said first encoded data to generate first coefficient data; quantizing said first coefficient data by a fixed step size; variable length encodingsaid first quantized data to generate a first bit stream; counting a data quantity of said first bit stream every predetermined time to indicate a difficulty of encoding; determining said difficulty of encoding said first bit stream based onintra-frame and forward-predictive encoded pictures only; calculating an allocated code quantity for each unit of predetermined time as a function of said difficulty of encoding so that said allocated code quantity is set to be larger for complicatedpictures and smaller for simple pictures; selectively encoding said input video signal by intra-frame or predictive encoding to generate second encoded data; transform encoding said second encoded data to generate second coefficient data; quantizingsaid second coefficient data at a step size set in response to said allocated code quantity; variable length encoding said second quantized data to generate a second bit stream; and recording said second bit stream on said recording medium.
.Iadd.16. An encoding method for encoding source video data, the method comprises the steps of: encoding said source video data with a predetermined quantization step size to generate first encoded data; detecting a difficulty of the encodingprocess of source video data based on bit amount of said first encoded data; deciding an optimum quantization step size, said optimum quantization step size being varied depending on said difficulty so that said optimum quantization step size becomessmaller when said source video data is more complex and said optimum quantization step size becomes larger when source video data to be encoded is more simple; and encoding said source video data by using said optimum quantization step on encoding unitbasis, wherein the predetermined quantization step size has a fixed value and the optimum quantization step size has a non-fixed value, wherein said source video data is always encoded using said predetermined quantization step and said optimumquantization step in which the predetermined quantization step size is always different from the optimum quantization step size. .Iaddend.
.Iadd.17. An encoding method for encoding source video data, the method comprises the steps of: encoding said source video data with a predetermined quantization step size to generate first encoded data; detecting a difficulty of the encodingprocess of source video data based on bit amount of said first encoded data; calculating an allocated code quantity which is varied depending on said difficulty so that said allocated code quantity is more increased when said source video data is morecomplex and said allocated code quantity is more decreased when source video data is more simple; and encoding said source video data by an optimum quantization step size based on said allocated code quantity, wherein the optimum quantization step sizehas a non-fixed value, and wherein said source video data is always encoded using said predetermined quantization step and said optimum quantization step in which the predetermined quantization step size is always different from the optimum quantizationstep size. .Iaddend.
.Iadd.18. An encoding method for encoding source video data, the method comprises the steps of: detecting motion vector of a macro block of said source video data; encoding said macro block of said source video data by using a predeterminedquantization step size and said detected motion vector to generate first encoded data; detecting a difficulty of the encoding process of source video data based on amount of said first encoded data; deciding an optimum quantization step size, saidoptimum quantization step size being varied depending on said difficulty so that said optimum quantization step size becomes smaller when said source video data is more complex and said optimum quantization step size becomes larger when source video databe encoded is more simple; and encoding said macro block of said source video data by using said optimum quantization step and said detected motion vector, wherein the predetermined quantization step size has a fixed value and the optimum quantizationstep size has a non-fixed value, and wherein said source video data is always encoded using said predetermined quantization step and said optimum quantization step in which the predetermined quantization step size is always different from the optimumquantization step size. .Iaddend.
.Iadd.19. An encoding method for encoding source video data, the method comprises the steps of: selecting a predictive mode of a macro block of said source video data; encoding said macro block of said source video data by using apredetermined quantization step size and said selected predictive mode to generate first encoded data; detecting a difficulty of the encoding process of source video data based on amount of said first encoded data; deciding an optimum quantization stepsize, said optimum quantization step size being varied depending on said difficulty so that said optimum quantization step size becomes smaller when said source video data is more complex and said optimum quantization step size becomes larger when sourcevideo data to be encoded is more simple; and encoding said macro block of said source video data by using sad optimum quantization step and said selected predictive mode, wherein the predetermined quantization step size has a fixed value and the optimumquantization step size has a non-fixed value, and wherein said source video data is always encoded using said predetermined quantization step and said optimum quantization step in which the predetermined quantization step size is always different fromthe optimum quantization step size. .Iaddend.
.Iadd.20. An encoding apparatus for encoding source video data, the apparatus comprising: means for detecting motion vector of a macro block of said source video data; first encoding means for encoding said macro block of said source videodata by using a predetermined quantization step size and said detected motion vector to generate first encoded data; means for detecting a difficulty of the encoding process of source video data based on amount of said first encoded data; means fordeciding an optimum quantization step size, said optimum quantization step size being varied depending on said difficulty so that said optimum quantization step size becomes smaller when said source video data is more complex and said optimumquantization step size becomes larger when source video data to be encoded is more simple; and second encoding means for encoding said macro block of said source video data by using said optimum quantization step and said detected motion vector, whereinthe predetermined quantization step size has a fixed value and the optimum quantization step size has a non-fixed value, and wherein said source video data is always encoded using said predetermined quantization step and said optimum quantization step inwhich the predetermined quantization step size is always different from the optimum quantization step size. .Iaddend.
.Iadd.21. An encoding apparatus for encoding source video data, the apparatus comprising: means for selecting a predictive mode of a macro block of said source video data; first encoding means for encoding said macro block of said source videodata by using a predetermined quantization step size and said selected predictive mode to generate first encoded data; means for detecting a difficulty of the encoding process of source video data based on amount of said first encoded data; means fordeciding an optimum quantization step size, said optimum quantization step size being varied depending on said difficulty so that said optimum quantization step size becomes smaller when said source video data is more complex and said optimumquantization step size becomes larger when source video data to be encoded is more simple; and second encoding means for encoding said macro block of said source video data by using said optimum quantization step and said selected predictive mode,wherein the predetermined quantization step size has a fixed value and the optimum quantization step size has a non-fixed value, and wherein said source video data is always encoded using said predetermined quantization step and said optimum quantizationstep in which the predetermined quantization step size is always different from the optimum quantization step size. .Iaddend.
.Iadd.22. An encoding apparatus for encoding source video data, the apparatus comprising: first encoding means for encoding said source video data with a predetermined quantization step size to generate first encoded data; second encodingmeans for encoding said source video data based on supplied quantization step size to generate second encoded data; transmitting buffer for buffering said second encoded data; and control means for detecting a difficulty of the encoding process in saidfirst encoding means, and for deciding said quantization step size, said optimum quantization step size being varied depending on said difficulty so that said quantization step size becomes smaller when said source video data is more complex and saidquantization step size becomes larger when source video data to be encoded is more simple, and said quantization step size being dependent on a remaining capacity of said transmitting buffer so as to suppress overflow and underflow in said transmittingbuffer, wherein the predetermined quantization step size has a fixed value and the optimum quantization step size has a non-fixed value, and wherein said source video data is always encoded using said predetermined quantization step and said optimumquantization step in which the predetermined quantization step size is always different from the optimum quantization step size. .Iaddend. |
| Description: |
TECHNICAL FIELD
This invention relates to a picture encoding method, a picture encoding apparatus and a picture recording medium, and more particularly to a picture encoding method, a picture encoding apparatus and a picture recording medium which are used in asystem for implementing encoding for storage to a video signal of moving picture to record the coded signal onto a picture recording medium such as an optical disc, a magnetic disc, or a magnetic tape, etc., or a system for transmitting a video signal ofmoving picture through a transmission path.
BACKGROUND ART
Hitherto, in a system for transmitting a video signal of moving picture to remote place, for example, as in the television conference system, the television telephone system, etc., or a system for recording a video signal of moving picture onto apicture recording medium such as an optical disc, a magnetic disc or a magnetic tape, etc., or reproducing a recorded video signal of moving picture, there is adopted for the purpose of efficiently utilizing transmission path (or picture recordingmedium), a scheme to implement so called efficient encoding to a video signal by making use of correlation between lines or correlation between frames that video signal has to reduce redundancies in the spatial axis direction and the time axis directionto transmit only significant information, thus to improve transmission efficiency.
For example, in encoding processing in the spatial axis direction (hereinafter intra-frame coding processing), e.g., correlation between lines of a video signal is utilized as shown in FIG. 7A. In the case of attempting to transmit respectivepictures PC1, PC2, PC3 . . . constituting a moving picture at times t1, t2, t3 . . . , picture data to be transmission processed is caused to undergo one-dimensional coding, e.g., within the same scanning line, or a picture is divided into, e.g., aplurality of blocks to allow picture data of respective blocks to undergo two-dimensional coding to thereby carry out data compression, thus to improve transmission efficiency.
Moreover, in coding processing in the time axis direction (hereinafter referred to as inter-frame coding processing), inter-frame correlation of video signal is utilized to determine, by so called predictive coding, for example, picture dataPC12, PC23 . . . comprised of deficiencies (so called predictive errors) of picture data every corresponding pixels between adjacent pictures PC1 and PC2, PC2 and PC3 . . . in succession to transmit these picture data PC12, PC23, . . . to therebycarry out data compression, thus to improve transmission efficiency.
Thus, as compared to the case where all picture data of pictures PC1, PC2, PC3 . . . are transmitted, a video signal can be transmitted by extremely lesser data quantity.
Further, in the predictive coding in the above-described inter-frame coding processing, motion compensated prediction is used, e.g., in macro block units in order to further improve efficiency. Namely, e.g., in the case where a person at thecentral portion of picture moves, or the like, motion (movement) of an object moving in the picture is detected to correct position of picture data used for prediction in the former picture by that motion to carry out predictive coding, thereby making itpossible to improve coding efficiency. However, even when such motion compensated prediction is employed, many data must be transmitted with respect to the portion where an object moves and appears from behind. In view of this, not only motioncompensation in the above-described forward direction, but also motion compensation in backward direction or in both directions of forward and backward directions are carried out in combination, thereby making it possible to further improve codingefficiency.
In actual terms, as shown in FIG. 8A, in macro blocks of frame data F0, F1, F2, F3 of the 0th, first, second, third . . . frames of a video signal of moving picture to be transmitted, in the case where there took place changes of pictures asrespectively indicated by motion vectors x0, x1, x2, x3 . . . between frames in succession, device on the transmitter side designates frames at intervals of a predetermined number of frames (e.g., every other frame), i.e., second, fourth . . . framesas interpolation frames to implement so called predetermined interpolation frame processing to these interpolation frames as shown in FIG. 8B to thereby generate transmit interpolated frame data F2X, F4X . . . . Further, with respect tonon-interpolation frames, the device on the transmitting side implements a predetermined coding processing to frame data F1, F3 . . . to generate transmit non-interpolated frame data F1X, F3X . . . .
For example, difference SP2 (predictive error) between motion compensated frame data F3 and F2, difference SP3 between motion compensated frame data F1 and F2, and difference between frame data obtained by implementing interpolation processing tomotion compensated frame data F1, F3 and frame data F2 are respectively determined in macro block units to compare difference (data) SP1 of frame data F2 and those differences. Then, data having minimum data quantity generated of those dataSP1.about.SP4 is caused to be transmit interpolated data F2X in macro block units. Similarly, transmit interpolated F4X . . . with respect to respective interpolation frames are generated. Further, e.g., DCT processing and variable length codingprocessing, etc. are implemented to frame data F1, F3 . . . of non-interpolation frames to generate transmit non-interpolated frame data F1X, F3X . . . .
The transmit non-interpolated frame data F1X, F3X . . . and transmit interpolated frame data F2X, F4X . . . are transmitted to the device on the receiving side as transmit data along with motion vectors x0, x1, x3 . . . .
On the other hand, the device on the receiving side implements decoding processing corresponding to coding processing on the transmitting side to transmit data (transmit non-interpolated frame data F1X, F3X . . . , transmit interpolated framedata F2X, F4X . . . , data of motion vectors x0, x1, x3 . . . ), thus to reproduce frame data F0, F1, F2, F3 . . . . As a result, motion compensation is implemented not only in forward direction but also in backward direction or in forward andbackward directions, thereby making it possible to further improve coding efficiency.
Picture encoding apparatus and picture decoding apparatus having the above-described function will now be described.
This picture encoding apparatus comprises, as shown in FIG. 9, a pre-processing circuit 61 for separating an input video signal VD into luminance signal and color difference signal, analog/digital (hereinafter referred to as A/D) convertingcircuits 62a, 62b for respectively converting the luminance signal and the color difference signal from the pre-processing circuit 61 into digital signals, a frame memory group 63 for storing luminance data and color difference data (hereinafter referredto as picture data) from the A/D converting circuits 62a, 62b, a format converting circuit 64 for reading out picture data from the frame memory group 63 in accordance with block format, and an encoder 65 for implementing efficient coding to picture dataof block from the format converting circuit 64.
In operation, pro-processing circuit 61 separates input video signal VD into luminance signal and color difference signal. A/D converting circuits 62a, 62b respectively converts luminance signal and color difference signal into luminance dataand color difference data each comprised of 8 bits. Frame memory group 63 stores these luminance and color difference data.
Format converting circuit 64 reads out, in accordance with block format, picture data (luminance data, color difference data) stored in the frame memory group 63. Encoder 65 encodes the picture data thus read out by a predetermined efficienctcoding to output bit stream.
This bit stream is delivered to picture decoding apparatus 80 through transmission media 70 comprised of transmission path or picture recording media such as, optical disc, magnetic disc or magnetic tape, etc.
This picture encoding apparatus 80 comprises, as shown in the FIG. 9 mentioned above, decoder 81 corresponding to the encoder 65, format converting circuit 82 for converting picture data reproduced by the decoder 81 into frame format, framememory groups 83 for storing picture data from the format converting circuit 82, D/A converting circuits 84a, 84b for converting luminance data, color difference data which have been read out from the frame memory group 83 into analog signals, andpost-processing circuit 85 for mixing luminance signal, color difference signal from the D/A converting circuits 84a, 84b, thus to generate output video signal.
Decoder 81 decodes bit stream by decoding corresponding to efficienct coding of encoder 65 to reproduce picture data of block format. Format converting circuit 82 converts this picture data into frame format to store it into frame memory group83.
D/A converting circuits 84a, 84b respectively convert luminance data and color difference data which have been read out from frame memory group 83 into luminance signal and color difference signal. Post-processing circuit 81 mixes theseluminance signal and color difference signal, thus to generate output video signal.
In actual terms, pre-processing circuit 61 and A/D converting circuits 62a, 62b convert luminance signal and color difference signal into digital signal as described above to reduce quantity of data so that the numbers of pixels become equal toone half of those of luminance signal in upper and lower directions and in left and right directions with respect to the luminance signal thereafter to implement time axis multiplexing processing thereto to deliver luminance data and color differencedata thus obtained to frame memory group 63.
From frame memory group 63, luminance data and color difference data are read out in accordance with block format as described above. Namely, e.g., picture data of one frame is divided into N slices as shown in FIG. 10A. Each slice is caused toinclude M macro blocks as shown in FIG. 10B. Each macro block is composed of luminance data Y1, Y2, Y3, Y4 of four luminance blocks consisting of 8.times.8 pixels adjacent in upper and lower directions and in left and right directions and colordifference data Cb, Cr of color blocks consisting of 8.times.8 pixels in a range corresponding to these four luminance blocks. From frame memory group 63, luminance data and color difference data are read out so that picture data are successive in macroblock units within slice and are successive in order of Y1, Y2, Y3, Y4, Cb, Cr within macro block. Picture data which have been read out in accordance with block format in this way are delivered to encoder 65.
Encoder 65 comprises motion vector detecting circuit 101 as shown in FIG. 11. This motion vector detecting circuit 101 detects, in macro block units, motion vector of picture data delivered thereto in accordance with block format. Namely,motion vector detecting circuit 101 detects, in macro block units, motion vector of current reference picture by forward original picture and/or backward original picture stored in frame memory group 83. Here, detection of motion vector is carried outsuch that minimum one of absolute value sums of differences between frames in macro block units is caused to be corresponding motion vector. The motion vector thus detected is delivered to motion compensating circuit 113, etc., and intra-framedifferences in macro block units are delivered to intra-frame/forward/backward bidirectionally predictive judging circuit 103.
This intra-frame/forward/backward/bidirectionally predictive judging circuit 103 determines predictive mode of reference block on the basis of this value to control predictive coating circuit 104 so as to carry out switching ofintra-frame/forward/backward/bidirectional prediction in macro block units. Predictive coding circuit 104 comprises adding circuits 104a, 104b, 104c and selecting (changeover) switch 104d, and is operative so that when predictive coding mode isintra-frame coding mode, it selects input picture itself, and when predictive coding mode is forward/backward/bidirectionally predictive mode, it selects differences (hereinafter referred to as difference data) every pixels of input picture data withrespect to respective predictive pictures, thus to deliver the selected data to DCT circuit 105.
DCT circuit 105 implements DCT processing to input picture data or difference data in block units by making use of the two-dimensional correlation of video signal to deliver coefficient data thus obtained to quantizing circuit 106.
The quantizing circuit 108 quantizes coefficient data by using quantization step size (quantization scale) determined every macro block or slice to deliver quantized data thus obtained to variable length coding (hereinafter referred to as VLC)circuit 107 and inverse quantizing circuit 10B. Meanwhile, quantization step size used for this quantization is determined so as to take a value such that transmitting buffer memory 109 which will be described later does not break by providing feedbackof buffer residual of transmitting buffer 109. This quantization step size is also delivered to VLC circuit 107 and inverse quantizing circuit 10B.
VLC circuit 107 implements variable length coding to quantized data along with quantization step size, predictive mode and motion vector to deliver them to transmitting buffer memory 109 as transmit data.
The transmitting buffer memory 109 temporarily stores transmit data thereafter to read out it at a predetermined bit rate to thereby smooth transmit data to output it as bit stream, and to feed quantization control signal in macro block unitsback to quantizing circuit 108 in accordance with residual data quantity remaining in the memory to control quantization step size. Thus, transmitting buffer memory 109 adjusts data quantity generated as bit stream to maintain data of appropriateresidual (remaining capacity) (data quantity such that no overflow or underflow takes place) within the memory. For example, when data residual of transmitting buffer memory 109 increase to allowed upper limit, transmitting buffer memory 109 allowsquantization step size of quantizing circuit 108 to be large by quantization control signal, thus to reduce data quantity of quantized data. On the other hand, when data residual of transmit buffer memory 109 decrease down to allowed lower limit,transmitting buffer memory 109 allows quantization step size of quantizing circuit 106 to be small by quantization control signal to thereby increase data quantity.
In this way, bit stream outputted from buffer memory 109 is delivered to picture decoded unit 80 through transmission media 70 comprised of a transmission path or a picture recording medium such as optical disc, magnetic disc, or magnetic tapeetc. at a predetermined bit rate as described above.
On the other hand, inverse quantizing circuit 108 inverse-quantizes quantized data delivered from quantizing circuit 106 to reproduce coefficient data (quantization distortion is added) corresponding to output of the above-described DCT circuit105 to deliver the coefficient data to Inverse Discrete Cosine Transform (hereinafter referred to as IDCT) circuit 110.
The IDCT circuit 110 implements IDCT processing to the coefficient data to reproduce picture data corresponding to input picture data in the intra-frame coding mode, and to reproduce difference data corresponding to output of predictive codingcircuit 104 in the forward/backward/bidirectionally predictive modes, thus to deliver it to adding circuit 111.
When predictive coding mode is the forward/backward/bidirectionally predictive modes, the adding circuit 111 is supplied with motion-compensated predictive picture data from motion compensating circuit 113 which will be described later to add themotion-compensated predictive picture data and difference data to thereby reproduce picture data corresponding to input picture data.
The picture data reproduced in this way is stored into frame memory 112. Namely, inverse quantizing circuit 108.about.adding circuit 111 constitute a local decoding circuit to locally decode quantized data outputted from quantizing circuit 106to write decoded picture thus obtained into frame memory 112 as forward predictive picture or backward predictive picture. The frame memory 112 is comprised of a plurality of frame memories. Bank switching of the frame memory is carried out. Incorrespondence with picture to be encoded, single frame is outputted as forward predictive picture data, or is outputted as backward predictive picture data. Moreover, in the case of bidirectional prediction, forward predictive picture data and backwardpredictive picture data are, e.g., averaged. The averaged data thus obtained is outputted. These predictive picture data are entirely the same pictures as pictures reproduced by decoder 81 which will be described later. Picture to be processed next iscaused to undergo forward/backward/bidirectional predictive coding on the basis of this predictive picture.
Namely, picture data which has been read out from frame memory 112 is delivered to motion compensating circuit 113. This motion compensating circuit 113 implements motion compensation to predictive picture data on the basis of motion vector todeliver the motion-compensated predictive picture data to predictive encoding circuit 104 and adding circuit 111.
Decoder 81 will now be described.
To decoder 81, bit stream is inputted through transmission media 70 is inputted. This bit stream is inputted to Variable Length Decoding (Inverse Variable Length Coding) (hereinafter referred to as IVLC) through receiving buffer 201. The IVLCcircuit 202 reproduces quantized data, motion vector, predictive mode and quantization step size, etc. from bit stream. These quantized data and quantization step size are delivered to inverse quantizing circuit 203. Motion vector is delivered tomotion compensating circuit 207, and predictive mode is delivered to adding circuit 205.
The operation of inverse quantizing circuit 203.about.adding circuit 205 is the same as that of local decoding circuit of encoder 61, the operations of frame memory group 206, motion compensating circuit 207 are respectively the same as those offrame memory 112 and motion compensating circuit 113 of encoder 61. On the basis of quantized data, motion vector, predictive mode, quantization step size, decoding is carried out. As a result, reproduction picture data is outputted from adding circuit205.
As described above, in the conventional apparatus, coding bit rate of bit stream generated at encoder 65 is caused to be fixed in correspondence with transfer rate of transmission media 70. Under this limitation, quantity of data generated,i.e., quantization step size of quantizing circuit 106 in encoder 65 was controlled. In other words, for example, a control was conducted such that when pictures of complicated pattern are successive, quantization step size is caused to be larger tosuppress quantity of data generated, while when simple patterns are successive, quantization step size is caused to be smaller to increase quantity of data generated so that buffer memory 109 does not produce overflow or underflow, thus to maintain afixed rate.
Accordingly, in the conventional apparatus, when complicated pictures are succesive, quantization step size is caused to be larger, so picture quality is deteriorated, while when simple pictures are successive, quantization step size is caused tobe smaller. As a result, uniform picture quality cannot be obtained through the entirety.
In addition, in the case of recording bit stream onto a picture recording medium of a limited data capacity, in order to avoid extreme deterioration of picture quality with respect to pictures of complicated pattern, a fixed rate of high ratesuch that picture quality of such complicated picture is not injured must be applied to the entirety, resulting in decreased recording time.
DISCLOSURE OF THE INVENTION
In order to solve the above-described problem, a first picture encoding method according to this invention comprises the steps of encoding at least a portion of an input video signal to generate first encoded data, determining an encording rateevery predetermined time on the basis of data quantity every predetermined time of the first encoded data and total quantity of usable data, and encoding the input video signal every predetermined time on the basis of the encoding rate to generate secondencoded data.
A second picture encoding method according to this invention is characterized in that, in the first picture encoding method, at least a portion of the input video signal is quantized by a fixed quantization step size to generate the first encodeddata.
A third picture encoding method according to this invention is characterized in that, in the first picture encoding method, the total quantity of usable data is proportionally allocated in accordance with data quantity every predetermined time,thus to determine the encoding rate every predetermined time.
A fourth picture encoding method according to this invention comprises the steps of implementing a predetermined predictive encoding or a predetermined transform encoding to at least a portion of an input video signal to generate firstcoefficient data, quantizing the first coefficient data by a fixed step size to generate first quantized data, allowing the first quantized data to undergo variable length encoding to generate a first bit stream, determining an encoding rate everypredetermined time on the basis of data quantity of the first bit stream and total quantity of usable data, implementing the predetermined predictive encoding and/or the predetermined transform encoding to the input video signal to generate secondcoefficient data, quantizing the second coefficient data by a quantization step size based on the encoding rate every predetermined time to generate second quantized data, and allowing the second quantized data to undergo variable length encoding togenerate a second bit stream.
A fifth picture encoding method according to this invention is characterized in that, in the fourth picture encoding method, the encoding rate is determined every one frame on the basis of data quantity every one frame in the first bit stream andtotal quantity of usable data.
A sixth picture encoding method according to this invention is characterized in that, in the fourth picture-encoding method, the encoding rate is determined every GOP on the basis of data quantity of at least a potion every GOP consisting of aplurality of frames in the first bit stream and total quantity of usable data.
A seventh picture encoding method according to this invention is characterized in that, in the sixth picture encoding method, encoding rate every GOP is determined on the basis of quantity with respect to intra-frame encoded picture and forwardpredictive encoded picture in the GOP.
An eighth picture encoding method according to this invention is characterized in that, in the fourth picture encoding method, the total quantity of usable data is proportionally allocated in dependency upon data quantity of the first bit streamevery predetermined time to determine an encoding rate every predetermined time.
A ninth picture encoding method according to this invention comprises the steps of determining difficulty of encoding every predetermined picture unit of an input video signal, setting an encoding rate every predetermined picture unit on thebasis of the difficulty of encoding and total quantity of usable data, and implementing encoding to the input video signal so that encoding rates of respective picture units are in correspondence with the set encoding rate every picture unit.
A tenth picture encoding method according to this invention is characterized in that, in the ninth picture encoding method, the predetermined picture unit is frame.
An eleventh picture encoding method according to this invention is characterized in that, in the ninth picture encoding method, the predetermined picture unit is GOP consisting of a plurality of frames.
A twelfth picture encoding method according to this invention is characterized in that, in the ninth picture encoding method, a predetermined predictive encoding and/or a predetermined transform encoding is implemented to at least a portion ofthe input video signal to generate coefficient data to quantize the coefficient data by a fixed quantization step size to thereby determine difficulty of encoding.
A first picture encoding apparatus according to this invention comprises first encoding means for encoding at least a portion of an input video signal to generate first encoded data; encoding control means for determining an encoding rate everypredetermined time on the basis of data quantity every predetermined time of the first encoded data from the first encoding means and total quantity of usable data; and second encoding means for encoding the input video signal every predetermined time onthe basis of the encoding rate every predetermined time from encoding control means to generate second encoded data.
A second picture encoding apparatus according to this invention is characterized in that, in the first picture encoding apparatus, the first encoding means comprises quantizing means for quantizing at least a portion of the input video signal bya fixed quantization step size.
A third picture encoding apparatus according to this invention is characterized in that, in the first picture encoding apparatus, the encoding control means is operative to proportionally allocate the total quantity of usable data in dependencyupon data quantity every predetermined time, thus to determine an encoding rate every predetermined time.
A fourth picture encoding apparatus according to this invention comprises first encoding means for implementing a predetermined predictive encoding and/or a predetermined transform encoding to at least a portion of an input video signal togenerate first coefficient data; first quantizing means for quantizing the first coefficient data from the first encoding means by a fixed quantization step size to generate first quantized data; first variable length encoding means for allowing thequantized data from the first quantizing means to undergo variable length encoding to generate a first bit stream; encoding control means for determining an encoding rate every predetermined time on the basis of data quantity of the first bit stream fromthe first variable length encoding means and total quantity of usable data; second encoding means for implementing the predetermined predictive encoding and/or the predictive transform encoding to the input video signal to generate second coefficientdata; second quantizing means for quantizing the second coefficient data from the second encoding means by a quantization step size based on the encoding rate every predetermined time from the encoding control means to generate second quantized data; andsecond variable length encoding means for allowing the second quantized data from the second quantizing means to undergo variable length encoding to generate a second bit stream.
A fifth picture encoding apparatus according to this invention is characterized in that, in the fourth picture encoding apparatus, the encoding control means determines the encoding rate every one frame on the basis to data quantity every oneframe in the first bit stream and total quantity of usable data.
A sixth picture encoding apparatus according to this invention is characterized in that, in the fourth picture encoding apparatus, the encoding control means determines the encoding rate every GOP on the basis of data quantity of at least aportion every GOP consisting of a plurality of frames in the first bit stream and total quantity of usable data.
A seventh picture encoding apparatus according to this invention is characterized in that, in the sixth picture encoding apparatus, the encoding control means determines an encoding rate every GOP on the basis of data quantity with respect tointra-frame encoded picture and forward predictive encoded picture in the GOP.
An eighth picture encoding apparatus according to this invention is characterized in that, in the fourth picture encoding apparatus, the encoding control means proportionally allocate the total quantity of usable data in dependency upon dataquantity of the first bit stream every predetermined time, thus to determine the encoding rate every predetermined time.
A ninth picture encoding apparatus according to this invention comprises difficulty calculating means for calculating (determining) difficult of encoding every predetermined picture unit of an input video signal; encoding rate setting means forsetting an encoding rate every predetermined picture unit on the basis of the difficulty of encoding from the difficulty calculating means and total quantity of usable data; and encoding means for allowing the input video signal to undergo encoding sothat encoding rates of respective picture units are in correspondence with the encoding rate every picture unit set by the encoding rate setting means.
A tenth picture encoding apparatus according to this invention is characterized in that, in the ninth picture encoding apparatus, the difficulty calculating means determines difficulty of encoding every frame.
An eleventh picture encoding apparatus according to this invention is characterized in that, in the ninth picture encoding apparatus, the difficulty calculating means determines difficulty of encoding every GOP consisting of a plurality offrames.
A twelfth picture encoding apparatus according to this invention is characterized in that, in the ninth picture encoding apparatus, the difficulty calculating means implements a predetermined predictive encoding and/or a predetermined transformencoding to at least a portion of the input video signal to generate coefficient data, and to quantize the coefficient data by a fixed step size to thereby calculate (determine) difficulty of encoding.
A first picture recording medium according to this invention is characterized in that there is recorded a second bit stream obtained by encoding at least a portion of an input video signal to generate first encoded data to determine an encodingrate every predetermined time on the basis of data quantity every predetermined time of the first encoded data and total quantity of usable data to encode the input video signal every predetermined time on the basis of the encoding rate.
A second picture recording medium according to this invention is characterized in that there is recorded a second bit stream obtained by implementing a predetermined predictive encoding and/or a predetermined transform encoding to at least aportion of an input video signal to generate first coefficient data to quantize the first coefficient data by a fixed quantization step size to generate first quantized data to allow the first quantized data to undergo variable length encoding togenerate a first bit stream to determine an encoding rate every predetermined time on the basis of data quantity of the first bit stream and total quantity of usable data to implement the predetermined predictive encoding and/or the predeterminedtransform encoding to the input video signal to generate second coefficient data to quantize the second coefficient data by a quantization step size based on the encoding rate every predetermined time to generate second quantized data to allow the secondquantized data to undergo variable length encoding.
A third picture recording medium according to this invention is characterized in that there is recorded encoded data obtained by determining difficulty of encoding every predetermined picture unit of an input video signal to set encoding rateevery predetermined picture unit on the basis of the difficulty of encoding and total quantity of usable data to encode the input video signal so that encoding rates of respective picture units are in correspondence with the set encoding rate everypicture unit.
In accordance with the first picture encoding method according to this invention, encoding rate every predetermined time is determined on the basis of data quantity every predetermined time of first encoded data obtained by encoding at least aportion of an input video signal and total quantity of usable data to encode the input video signal every predetermined time on the basis of the encoded rate to generate second encoded data.
In accordance with the second picture encoding method according to this invention, in the first picture encoding method, at least a portion of input video signal is quantized by fixed quantization step size to thereby generate the first encodeddata to determine the encoding rate to encode the input video signal every predetermined time on the basis of the encoding rate to generate second encoded data.
In accordance with the third picture encoding method according to this invention, in the first picture encoding method, the total quantity of usable data is proportionally allocated in dependency upon data quantity every predetermined time todetermine encoding rate every predetermined time to encode the input video signal every predetermined time on the basis of the encoding rate to generate second encoded data.
In accordance with the fourth picture encoding method according to this invention, predetermined predictive encoding and/or predetermined transform encoding processing, and quantization processing and variable length encoding processing at fixedquantization step size are implemented to at least a portion of an input video signal to generate first bit stream to determine encoding rate every predetermined time on the basis of data quantity of the first bit stream and total quantity of usabledata. Then, predetermined predictive encoding and/or predetermined transform encoding processing, and quantization processing and variable length encoding processing by quantization step size based on encoding rate every predetermined time areimplemented to input video signal, thus to generate second bit stream.
In accordance with the fifth picture encoding method according to this invention, in the fourth picture encoding method, the encoding rate is determined every one frame on the basis of data quantity every one frame in the first bit stream andtotal quantity of usable data. Then, predetermined predictive encoding and/or predetermined transform encoding processing, and quantization processing and variable length encoding processing at quantization step size based on encoding rate every oneframe are implemented to input video signal, thus to generate second bit stream.
In accordance with the sixth picture encoding method according to this invention, in the fourth picture encoding method, the encoding rate is determined every GOP on the basis of data quantity of at least a portion every GOP consisting of aplurality of frames in the first bit stream and total quantity of usable data. Then, predetermined predictive encoding and/or predetermined transform encoding processing, and quantization processing and variable length encoding processing atquantization step size based on encoding rate every GOP are implemented to input video signal, thus to generate second bit stream.
In accordance with the seventh picture encoding method according to this invention, in the sixth picture encoding method, encoding rate every GOP is determined on the basis of data quantity with respect to intra-frame encoded picture and forwardpredictive encoded picture in the GOP. Then, predetermined predictive encoding and/or predetermined transform encoding processing, quantization processing at quantization step based on encoding rate every GOP and variable length encoding processing areimplemented to input video signal, thus to generate second bit stream.
In accordance with the eighth encoding method according to this invention, in the fourth picture encoding method, the total quantity of usable data is proportionally allocated in dependency upon data quantity of the first bit stream everypredetermined time to determine the encoding rate every predetermined time. Then, predetermined predictive encoding and/or predetermined transform encoding processing, quantization processing at quantization step size based on encoding rate everypredetermined time and variable length encoding processing are implemented to input video signal, thus to generate second bit stream.
In accordance with the ninth picture encoding method according to this invention, difficulty of encoding every predetermined picture unit of input video signal is determined to set encoding rate every predetermined picture unit on the basis ofthe difficulty of encoding and total quantity of usable data. Then, the input video signal is encoded so that the encoding rates of respective picture units are in correspondence with the set encoding rate every picture unit.
In accordance with the tenth picture encoding method according to this invention, in the ninth picture encoding method, difficulty of encoding is determined every frame of input video signal to determine the encoding rate every frame. Then, theinput video signal is encoded so that encoding rates of respective frames are in correspondence with the set encoding rate every frame.
In accordance with the eleventh picture encoding method, in the ninth picture encoding method, difficulty of encoding is determined every GOP of input video signal to determine the encoding rate every GOP. Then, the input video signal is encodedso that encoding rates of GOPs are in correspondence with the set encoding rate every GOP.
In accordance with the twelfth encoding method according to this invention, in the ninth picture encoding method, coefficient data obtained by implementing predetermined predictive encoding and/or predetermined transform encoding to at least aportion of input video signal is quantized at fixed quantization step size to thereby determine difficulty. Then, the input video signal is encoded so that encoding rates of respective picture units are in correspondence with the set encoding rate everypicture unit.
Further, in accordance with the first picture encoding apparatus according to this invention, encoding rate every predetermined time is determined on the basis of data quantity every predetermined time of first encoded data obtained by encodingat least a portion of input video signal and total quantity of usable data to encode the input video signal every predetermined time on the basis of the encoding rate, thus to generate second encoded data.
In accordance with the second picture encoding apparatus according to this invention, in the first picture encoding apparatus, at least a portion of input video signal is quantized at fixed quantization step size to thereby generate the firstencoded data to determine the encoding rate to encode the input video signal every predetermined time on the basis of the encoding rate, thus to generate second encoded data.
In accordance with the third picture encoding apparatus according to this invention, in the first picture encoding apparatus, the total quantity of usable data is proportionally allocated in dependency upon the data quantity every predeterminedtime to determine the encoding rate every predetermined time to encode the input video signal every predetermined time on the basis of the encoding rate, thus to generate second encoded data.
In accordance with the fourth picture encoding apparatus according to this invention, predetermined predictive encoding and/or predetermined transform encoding processing, and quantization processing at fixed quantization step size and variablelength encoding processing are implemented to at least a portion of input video signal to generate first bit stream to determine encoding rate every predetermined time on the basis of data quantity of the first bit stream and total quantity of usabledata. Then, predetermined predictive encoding and/or predetermined transform encoding processing, and quantization processing and variable length encoding processing at quantization step size based on encoding rate every predetermined time, thus togenerate second bit stream.
In accordance with the fifth picture encoding apparatus according to this invention, in the fourth picture encoding apparatus, the encoding rate is determined every one frame on the basis of data quantity every one frame in the first bit streamand total quantity of usable data. Then, predetermined predictive encoding and/or predictive transform processing, and quantization processing at quantization step size based on encoding rate every one frame and variable length encoding processing areimplemented to input video signal, thus to generate second bit stream.
In accordance with the sixth picture encoding apparatus according to this invention, in the fourth picture encoding apparatus, the encoding rate is determined every GOP on the basis of data quantity of at least a portion every GOP consisting of aplurality of frames in the first bit stream and total quantity of usable data. Then, predetermined predictive encoding and/or predetermined transform encoding processing, quantization processing at quantization step size based on encoding rate every GOPand variable length encoding processing are implemented to input video signal to generate second bit stream.
In accordance with the seventh picture encoding apparatus according to this invention, in the sixth picture encoding apparatus, the encoding rate every GOP is determined on the basis of data quantity with respect to intra-frame encoded pictureand forward predictive encoded picture in the GOP. Then, predetermined predictive encoding and/or predetermined transform encoding processing, quantization processing at quantization step size based on encoding rate every GOP and variable lengthencoding processing are implemented to input video signal, thus to generate second bit stream.
In accordance with the eighth picture encoding apparatus according to this invention, in the fourth picture encoding apparatus, the total quantity of usable data is proportionally allocated in dependency upon data quantity of the first bit streamevery predetermined time to determine the encoding rate every predetermined time. Then, predetermined predictive encoding and/or predetermined transform encoding processing, quantization processing at quantization step size based on encoding rate everypredetermined time and variable length encoding processing are implemented to input video signal, thus to generate second bit stream.
In accordance with the ninth picture encoding apparatus according to this invention, difficulty of encoding is determined every predetermined picture unit of input video signal to set encoding rate every predetermined picture unit on the basis ofthe difficulty of encoding and total quantity of usable data. Then, the input video signal is encoded so that encoding rates of respective picture units are in correspondence with the set encoding rate every picture unit.
In accordance with the tenth picture encoding apparatus according to this invention, in the ninth picture encoding apparatus, difficulty of encoding is determined every frame of input video signal to determine the encoding rate every frame. Then, the input video signal is encoded so that encoding rates of respective frames are in correspondence with the set encoding rate every frame.
In accordance with the eleventh picture encoding apparatus according to this invention, in the ninth picture encoding apparatus, difficulty of encoding is determined every GOP of input video signal to determine the encoding rate every GOP. Then,the input video signal is encoded so that encoding rates of respective GOPs are in correspondence with the set encoding rate every GOP.
In accordance with the twelfth picture encoding apparatus according to this invention, in the ninth picture encoding apparatus, coefficient data obtained by implementing predetermined predictive encoding and/or predetermined transform encoding toat least a portion of input video signal is quantized at fixed quantization step size to thereby the determine difficulty of encoding. Then, the input video signal is encoded so that encoding rates of respective picture units are in correspondence withthe set encoding rate every picture unit.
In addition, in accordance with the first picture recording medium according to this invention, there is recorded second bit stream obtained by determining encoding rate every predetermined time on the basis of data quantity every predeterminedtime of first encoded data obtained by encoding at least a portion of input video signal and total quantity of usable data to encode the input video signal every predetermined time on the basis of the encoding rate.
In accordance with the second picture recording medium according to this invention, there is recorded second bit stream obtained by implementing predetermined predictive encoding and/or predetermined transform encoding processing, andquantization processing and variable length encoding processing at fixed quantization step size to generate first bit stream to determine encoding rate every predetermined time on the basis of data quantity of the first bit stream and total quantity ofusable data to implement predetermined predictive encoding and/or predetermined transform encoding processing, and quantization processing and variable length encoding processing at quantization step size based on encoding rate every predetermined timeto the input video signal.
In accordance with the third picture recording medium according to this invention, there is recorded encoded data obtained by determining difficulty of encoding every predetermined picture unit of input video signal to set encoding rate everypredetermined picture unit on the basis of the difficulty of encoding and total quantity of usable data to encode the input video signal so that encoding rates of respective picture units are in correspondence with the set encoding rate every pictureunit.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram showing a circuit configuration of the essential part of a picture encoding apparatus to which this invention is applied.
FIG. 2 is a flowchart for explaining the operation of first encoding circuit constituting the above-mentioned picture encoding apparatus.
FIG. 3 is a flowchart for explaining the operation of second encoding circuit constituting the above-mentioned picture encoding apparatus.
FIG. 4 is a view showing respective pictures for explaining the configuration of GOP in MPEG.
FIG. 5 is a view showing respective pictures for explaining encoding control signal every GOP.
FIG. 6 is a flowchart for explaining the operation of the second encoding circuit constituting the above-mentioned picture encoding apparatus.
FIG. 7 is a view showing picture for explaining the principle of predictive encoding.
FIG. 8 is a view showing picture for explaining the principle of motion compensated predictive encoding.
FIG. 9 is a block diagram showing the configuration of picture encoding apparatus (unit) and picture decoding apparatus (unit) which are related to this invention.
FIG. 10 is a view showing the configuration of frame, macro block and slice.
FIG. 11 is a block diagram showing a circuit configuration of conventional encoder.
FIG. 12 is a block diagram showing a circuit configuration of conventional decoder.
BEST MODE FOR CARRYING OUT THE INVENTION
An embodiment of a picture encoding method, a picture encoding apparatus and a picture recording medium according to this invention will now be described with reference to the attached drawings.
A picture encoding apparatus to which this invention is applied comprises, as shown in FIG. 1, for example, a first encoding circuit 10 for encoding an input video signal to generate first encoded data, an encoding control circuit 30 fordetermining encoding rate every predetermined time on the basis of data quantity every predetermined time of the first encoded data from the first encoding circuit 10 and total quantity of usable data, and a second encoding circuit 40 for encoding theinput video signal every predetermined time on the basis of the encoding rate from the encoding control circuit 30 to generate second encoded data.
More particularly, the first encoding circuit 10 comprises, as shown in FIG. 1 mentioned above, a frame memory group 12 for storing input picture data which is input video signal, a motion vector detecting circuit 11 for detecting motion vectorof input picture data from input picture data on the basis of picture data stored in the frame memory group 12, a frame memory 22 for storing predictive picture data, a motion compensating circuit 23 for implementing motion compensation to predictivepicture data which has been read out from the frame memory 22 on the basis of motion vector from the motion vector detecting circuit 11, a predictive encoding circuit 14 for predictive-encoding input picture data on the basis of motion-compensatedpredictive picture data from the motion compensating circuit 23, a DCT circuit 15 for implementing encoding, e.g., Discrete Cosine Transform (hereinafter referred to as DCT) processing to differences, etc. which are predictive errors from the predictiveencoding circuit 14 to generate coefficient data, a quantizing circuit 18 for quantizing coefficient data from the DCT circuit 15 at a fixed quantization step size to generate quantized data, a Variable Length Code (hereinafter referred to as VLC)circuit 17 for allowing quantized data from the quantizing circuit 18 to undergo variable length encoding to output variable length encoded data, an inverse quantizing circuit 18 for inverse-quantizing quantized data from the quantizing circuit 16 toreproduce coefficient data, an Inverse Discrete Cosine Transform (hereinafter referred to as IDCT) circuit 20 for implementing decoding e.g., IDCT processing to coefficient data from the inverse quantizing circuit 18, and an adding circuit 21 for addingdifference from the IDCT circuit 20 and motion-compensated predictive picture data from the motion compensating circuit 23 to generate predictive picture data with respect to the next input picture data to deliver the predictive picture data to the framememory 22.
Moreover, the second encoding circuit 40 comprises, as shown in FIG. 1 mentioned above, a delay element 43 for delaying input picture data, a frame memory 52 for storing predictive picture data, a motion compensating circuit 53 for implementingmotion compensation to predictive picture data which has been read out from the frame memory 52 on the basis of motion vector from the motion vector detecting circuit 11, a predictive encoding circuit 44 for a predictive-encoding input picture datadelayed at the delay element 43 on the basis of the motion compensated predictive picture data from the motion compensating circuit 53, a DCT circuit 45 for implementing encoding, e.g., DCT processing to difference, etc. from the predictive encodingcircuit 44 to generate coefficient data, a quantization scale setting circuit 33 for setting quantization step size on the basis of encoding rate from the encoding control circuit 30, a quantizing circuit 46 for quantizing coefficient data from the DCTcircuit 45 at quantization step size form the quantization scale setting circuit 33, a VLC circuit for allowing quantized data from the quantizing circuit 46 to undergo variable length encoding to output variable length encoded data, a transmittingbuffer memory 49 for temporarily storing variable length encoded data from the VLC circuit 47 to output it at a fixed bit rate, an inverse quantizing circuit 48 for inverse-quantizing quantized data from the quantizing circuit 46 to reproduce coefficientdata, an IDCT circuit 50 for implementing decoding, e.g., IDCT processing to the coefficient data from the inverse quantizing circuit 48 to reproduce the difference, and an adding circuit 51 for adding difference from the IDCT circuit 50 and themotion-compensated predictive picture data from the motion compensating circuit 53 to generate predictive picture data with respect to the next input picture data to deliver the predictive picture data to the frame memory 52.
In this picture encoding apparatus, by the first encoding circuit 10, encoding processing, e.g., predictive encoding processing, DCT processing, quantization processing at fixed quantization step size and variable length encoding processing areimplemented to input picture data. By encoding control circuit 30, encoding bit rate is determined on the basis of data quantity every predetermined time of variable length encoded data which is the first bit stream obtained and data capacity of picturerecording medium 55, e.g., comprised of optical disc, magnetic disc or magnetic tape, etc. or total quantity of usable data determined by bit rate of transmission path (transfer rate). Thereafter, by second encoding circuit 40, predictive encodingprocessing, DCT processing, quantization processing and variable length encoding processing are implemented to input picture data for a second time. In generating variable length encoded data which is second bit stream, quantization is made byquantization step size based on encoding bit rate.
Namely, in this picture encoding apparatus, as shown in FIG. 2, for example, at step ST1, quantizing circuit 16 of first encoding circuit 10 sets quantization step size to, e.g., 1 to quantize coefficient data delivered from DCT circuit 15 togenerate quantized data. Counter 31 of encoding control circuit 30 counts data quantity of variable length encoded data (first bit stream) obtained by allowing the quantized data to undergo variable length encoding every predetermined time, e.g., oneframe to determine, every frame, quantity of code generated indicating difficulty of encoding.
At step ST2, bit rate calculating circuit 32 determines allocated code quantity allocated every frame on the basis of difficulty (quantity of code generated) every frame and total quantity of usable data.
At step ST3, quantizing circuit 46 of second encoding circuit 40 quantizes coefficient data delivered from DCT circuit 45 by quantization step size based on allocated code quantity to generate quantized data.
In actual terms, inputted picture data is temporarily stored into frame memory group 12. From the frame memory group 12, that picture data is read out in accordance with block format as described in the prior art.
Motion vector detecting circuit 11 reads out memory picture data from frame memory group 12 in macro block units described above to detect motion vector. Namely, motion vector detecting circuit 11 detects, in macro block units, motion vector ofcurrent reference picture by using forward original picture and/or backward original picture stored in frame memory group 12. Here, detection of motion vector is carried out such that a motion vector in which, e.g., absolute value sum of differencesbetween frames in macro block units becomes minimum is caused to be corresponding motion vector. The detected motion vector is delivered to motion compensating circuits 23, 53, etc., and absolute value sum of differences between frames in macro blockunits is delivered to intra-frame/forward/backward/bidirectionally predictive judging circuit 13.
The intra-frame/forward/backward/bidirectionally predictive judging circuit 13 determines predictive mode on the basis of the above value to control predictive encoding circuit 14 so as to carry out switching ofintra-frame/forward/backward/bidirectional prediction in block units.
The predictive encoding circuit 14 comprises, as shown in FIG. 1 mentioned above, adding circuits 14a, 14b, 14c and selecting (changeover) switch 14d. When predictive encoding mode is intra-frame encoding mode, the selecting switch 14d selectinput picture data itself, and when predictive encoding mode is forward/backward/bidirectionally predictive mode, it selects differences (hereinafter referred to as difference data) every pixels of input picture data with respect to respective predictivepictures. Then, the selecting switch 14d delivers selected data to DCT circuit 15.
The DCT circuit 15 implements, in block units, DCT processing to input picture data or difference data delivered from selecting switch 14d by making use of the two-dimensional correlation of video signal to deliver coefficient data thus obtainedto quantizing circuit 16.
The quantizing circuit 16 quantizes coefficient data delivered from DCT circuit 15 at a fixed quantization step size, e.g., with quantization step size being set to 1 to deliver quantized data thus obtained to VLC circuit 17 and inversequantizing circuit 18.
The VLC circuit 17 carries out variable length encoding of quantized data along with quantization step size, predictive mode, and motion vector, etc. to deliver variable length encoded data obtained to encoding control circuit 30 as a first bitstream.
The encoding control circuit 30 comprises, as shown in the FIG. 1 mentioned above, counter 31 for counting data quantity every predetermined time of variable length encoded data from the VLC circuit 17, and bit rate calculating circuit 32 forcalculating (determining) allocated code quantity per unit time on the basis of data quantity from the counter 31 and total quantity of usable data. The counter 31 counts data quantity of the first bit stream every predetermined time, e.g., every oneframe to determine (calculates) difficulty every frame to deliver this difficulty to bit rate calculating circuit 32.
The bit rate calculating circuit 32 calculates (determines) allocated code quantity allocated every frame, i.e., mean encoding rate every frame time on the basis of difficulty every frame and total quantity of usable data, and delivers thisallocated code quantity to quantization scale setting circuit 33 of second encoding circuit 40.
In actual terms, bit rate calculating circuit 32 performs the following calculation. Namely, assuming now that the number of all frames is N, total quantity of usable data is B, difficulty of the i (i=0, 1, 2, . . . N-1)-th frame is d.sub.i,and allocated code quantity with respect to the i-th frame is b.sub.i, when this allocated code quantity b.sub.i is caused to be proportional to difficulty d.sub.i as indicated by the following formula (1), data total quantity B can be calculated byadding allocated code quantities b.sub.i of all frames as indicated by the following formula (2). In the formula, a represents constant.
b.sub.i=a.times.d.sub.i (1) .times..times..times..times..times. ##EQU00001##
Accordingly, constant a can be calculated by the following formula (3). When substitution of this constant a into the formula (1) is made, allocated code quantity b.sub.i with respect to the i-th frame can be calculated by the following formula(4). .times..times..times. ##EQU00002##
Thus, bit rate calculating circuit 32 increases allocated code quantity b.sub.i with respect to frame of picture of complicated pattern, for example, and decreases allocated code quantity b.sub.i with respect to frame of simple pattern.
On the other hand, inverse quantizing circuit 18 inverse-quantizes quantized data delivered from quantizing circuit 16 at quantization step size caused to be set to 1 to reproduce coefficient data (quantization distortion is added) correspondingto output of DCT circuit 15 to deliver that coefficient data to IDCT circuit 20.
The IDCT circuit 20 implements IDCT processing to coefficient data to reproduce input picture data corresponding to output of predictive encoding circuit 14 in the intra-frame encoding mode, and reproduces difference data in theforward/backward/bidirectionally predictive mode, thus to deliver reproduced data to adding circuit 21.
The adding circuit 21 is supplied, when predictive encoding mode is forward/backward/bidirectionally predictive mode, with motion-compensated predictive picture data from motion compensating circuit 23. The adding circuit 21 adds this predictivepicture data and difference data delivered from IDCT circuit 20 to thereby reproduce picture data corresponding to input picture data.
The picture data reproduced in this way is stored into frame memory 22 as predictive picture data. Namely, inverse quantizing circuit 18.about.adding circuit 21 constitute a local decoding circuit to locally decode quantized data outputted fromquantizing circuit 18 on the basis of predictive mode to write decoded picture obtained into frame memory 22 as forward predictive picture or backward predictive picture. Frame memory 22 is composed of a plurality of frame memories. Back switching offrame memory is carried out. In dependency upon picture to be encoded, e.g., single frame is outputted as forward predictive picture data, or is outputted as backward predictive picture data. Further, in the case of forward/backward/bidirectionalprediction, forward predictive picture data and backward predictive picture data are, e.g., averaged and the averaged data is outputted. These predictive picture data are entirely the same picture data as picture data reproduced by picture decodingapparatus which will be described later. Picture to be processed next is caused to undergo forward/backward/bidirectionally encoding on the basis of this predictive picture.
The operation of second encoding circuit 40 will now be described. It is to be noted that since circuits except for quantization scale setting circuit 33, delay element 43, quantizing circuit 48, and transmitting buffer 49 constituting secondencoding circuit 40 perform the same operations as those of circuits constituting the above-described first encoding circuit 10, their explanation is omitted.
Delay element 43 delays input picture data, e.g., by time until encoding control signal is outputted from encoding control circuit 30. Then, at predictive encoding circuit 44 and DCT circuit 45, predictive encoding processing and DCT processingwhich are in accordance with predictive mode delivered from intra-frame/forward/backward/bidirectionally predictive judging circuit 13 are implemented to the delayed input picture data. Thus, coefficient data is generated.
Quantization scale setting circuit 33 determines (calculates) allocated code quantity every macro block (e.g., value obtained by dividing allocated code quantity every frame by the number of macro blocks in one frame) from delivered allocatedcode quantity every frame to carry out comparison between code quantity generated in a macro block which is detected from buffer feedback from transmitting buffer 49 and allocated code quantity every macro block. The quantization scale setting circuit33 operates as follows so as to allow encoding bit rate of respective frames to become close to set mean encoding bit rate every frame time. Namely, in the case where code quantity generated in corresponding macro block is greater than allocated codequantity every macro block, the circuit 33 sets quantization step size of next macro block to a greater value in order to suppress code quantity generated by the next macro block. In contrast, in the case where code quantity generated in correspondingmacro block is smaller than allocated code quantity every macro block, the circuit 33 sets quantization step size of next macro block to a smaller value so as to increase code quantity generated. It should be noted that quantization scale settingcircuit 33 is operative so that in the case where buffer feedback from transmitting buffer 49 indicates that transmitting buffer 49 is in a state close to overflow state, it allows quantization step size to be larger to suppress overflow irrespective ofcomparison result between the allocated code quantity and code quantity generated, while in the case where buffer feedback from the transmitting buffer 49 indicates that transmitting buffer 49 is in a state close to underflow state, it allowsquantization step to be smaller to suppress underflow irrespective of comparison result between the allocated code quantity and code quantity generated. While it has been described that comparison between code quantity generated and allocated codequantity is made every macro block to switch quantization step size every macro block, such switch may be carried out every slice. While it has been described that code quantity generated is detected from storage quantity of transmitting buffer 49, itmay be directly obtained from output of variable length encoding circuit 47. The quantization scale setting circuit 33 delivers quantization step size set in this way to quantizing circuit 46.
The quantizing circuit 46 quantizes coefficient data delivered from DCT circuit 45 by quantization step size delivered from the above-described quantization scale setting circuit 33 to generate quantized data.
VLC circuit 47 allows quantizing data delivered from quantizing circuit 46 to undergo variable length encoding along with quantization step size from quantization scale setting circuit 33, predictive mode fromintra-frame/forward/backward/bidirectionally predictive judging circuit 13, and motion vector from motion vector detecting circuit 11, etc. to deliver variable length encoded data obtained to transmitting buffer memory 49 as a second bit stream.
Namely, in this picture encoding apparatus, as shown in FIG. 3, for example, when picture data is inputted through delay element 43 at step ST1, quantization scale setting circuit 33 reads in, from encoding control circuit 30, allocated codequantity with respect to frame to be currently encoded at step ST2, then, the processing operation proceeds to step ST3.
At step ST3, predictive encoding circuit 44.about.VLC circuit 47 implement predictive encoding processing and DCT processing to picture data, and quantizes coefficient data by quantization step size based on allocated code quantity thereafter toallow it to undergo variable length encoding. Then, the processing operation proceeds to step ST4.
At the step ST4, whether or not encoding processing has been completed with respect to all frames (sequence) to which, e.g., the same picture size or the same transfer rate is applied is judged. If so, the processing is completed. In contrast,if not so, the processing operation returns to step ST1. Thus, variable rate encoding such that encoding rate changes in frame units is realized. Accordingly, even if pictures (frames) of complicated pattern are successive, there is no possibility thatquantization step size is caused to be large with respect to these pictures as in the conventional apparatus. Thus, uniform high picture quality can be obtained through the entirety.
The transmitting buffer memory 49 temporarily stores variable length encoded data thereafter to read out it at a fixed bit rate to thereby smooth the variable length encoded data to output it as bit stream. The bit stream which has beenoutputted from transmitting buffer 49 is multiplexed along with, e.g., encoded audio signal, synchronizing signal, etc. Further, code for error correction is added thereto, and a predetermined modulation suitable for transmission or recording is appliedthereto. Thereafter, bit stream thus processed is transmitted to picture decoding apparatus through, e.g., transmission path, or is recorded onto picture recording medium 55 comprised of optical disc, magnetic disc or magnetic tape, etc. as shown in theFIG. 1 mentioned above. Namely, since, in the second encoding circuit 40, there is carried out a variable rate encoding such that, e.g., allocated code quantity b.sub.i is increased in advance with respect to complicated picture and allocated codequantity b.sub.i is decreased with respect to simple picture, there is no necessity of applying a fixed rate of high rate through the entirety in order to avoid extreme deterioration of picture quality with respect to pictures of complicated pattern asin the case of conventional apparatus. Thus, recording time of picture recording medium 55 can be prolonged.
On the other hand, inverse quantizing circuit 48 inverse-quantizes quantized data delivered from quantizing circuit 46 by quantization step size used in the above-described quantizing circuit 46 to reproduce coefficient data (quantizationdistortion is added) corresponding to output of DCT circuit 45 to deliver this coefficient data to IDCT circuit 50. Namely, inverse quantizing circuit 48.about.adding circuit 51 constituting a local decoding circuit locally decode quantized dataoutputted from quantizing circuit 46 to write decoded picture obtained into frame memory 52 as forward predictive picture or backward predictive picture. Picture data stored in frame memory 52 is used as predictive picture for picture to be processednext.
Meanwhile, while, in the above-described embodiment, allocated code quantity per predetermined time, i.e., mean encoding rate per predetermined time is obtained every frame with frame being used as predetermined time, this invention is notlimited to such an implementation. For example, GOP (Group of Pictures in so called MPEG (Moving Picture Expert Group) may be used as a predetermined time. It should be noted that the above-described MPEG is general name of the moving picture encodingsystem being studied in WG (Working Group) 11 of SC (Sub Committee) 29 in JTC (Joint Technical Committee) of so called ISO (International Standardization Organization) and IEC (International Electrochemical Committee).
Namely, GOP in MPEG consists of at least one so called I picture, and a plurality of P pictures of B pictures (non-I picture). In a more practical sense, assuming that GOP consists of a single I picture, four P pictures having a period of 3pictures, and ten B pictures, encoding control circuit 30 determines allocated code quantity every GOP. Here, I picture is picture to be subjected to intra-field or intra-frame encoding. P picture is picture which can be predicted only from forwarddirection, and is subjected to interfield or intra-frame encoding. B picture is picture which can be predicted from forward direction, from back ward direction and from both directions and is subjected to interfield or intra-frame encoding.
When it is assumed, as shown in FIG. 5, for example, that successive arbitrary two pictures within GOP having the number of pictures consisting GOP as period are I picture and P picture, and quantization step size is, e.g., 1, the first encodingcircuit 10 implements predictive encoding processing, DCT processing, and variable length encoding processing to picture data of these I and P pictures to generate variable length encoded data to deliver the variable length encoded data to encodingcontrol circuit 30. The reason why two pictures are used as I picture, P picture is to examine complexity of pattern and correlation between frames. From code quantity generated of I picture, it is possible to recognize complexity of pattern. Fromcode quantity generated of P picture, it is possible to recognize correlation between frames. Since successive plural frames have similar pictorial images in general, it is possible to recognize tendency of pattern of GOP even from extracted twopictures.
Encoding control circuit 30 counts (calculates), every GOP, data quantity of I picture bit I.sub.j and data quantity of P picture bit P.sub.j, and determines, every GOP, difficulty (code quantity generated GOP d.sub.j (j=0, 1, 2 . . . ) on thebasis of these data quantities bit I.sub.j, bit P.sub.j and the number N of P pictures constituting GOP as indicated by the following formula (5), for example. GOPd.sub.i=bitI.sub.j+N.times.bitP.sub.j (5)
The encoding control circuit 30 determines allocated code quantity allocated every GOP on the basis of difficulty (code quantity generated) GOP d.sub.j every GOP and total quantity of usable data, and delivers this allocated code quantity tosecond encoding circuit 40.
In actual terms, when the number of GOPs is assumed to be M, total quantity of usable data is assumed to be B, allocated code quantity with respect to the j-th GOP is GOP b.sub.j, and this allocated code quantity GOP b.sub.j is caused to be inproportion to difficulty as indicated by the following formula (6), data total quantity B is determined by adding allocated code quantities GOP b.sub.j of all GOPs as indicated by the following formula (7). In the formula (6), a is constant.
GOPb.sub.j=a.times.GOPd.sub.j (6) .times. ##EQU00003##
Accordingly, constant a can be determined By the following formula (8). Substituting this constant a into the formula (6), .times..times..times..SIGMA..times. ##EQU00004## allocated code quantity GOPb.sub.j with respect to the j-th GOP can bedetermined by the following formula (9). .times..times..times. ##EQU00005##
Thus, encoding control circuit 30 increases allocated code quantity GOP b.sub.j with respect to, e.g., GOP in which pictures of complicated pattern are included or having low correlation between frames, and decreases allocated code quantity GOPb.sub.j with respect to GOP in which pictures of simple pattern are included or having high correlation between frames.
When second encoding circuit 40 is supplied with picture data through delay element 43 at step ST1, as shown in FIG. 6, for example, it judges, at step ST2, whether or not picture data being inputted is the leading picture of GOP. If so, theprocessing operation proceeds to step ST3. If not so, the processing operation proceeds to step ST4.
At step ST3, second encoding circuit 40 reads in allocated code quantity with respect to GOP currently subjected to encoding from encoding control circuit 30. Then, the processing operation proceeds to step ST4.
At the step ST4, second encoding circuit 40 implements predictive encoding processing and DCT processing, and quantizes coefficient data by quantization step size based on allocated code quantity thereafter to allow it to undergo variable lengthencoding. Then, the processing operation proceeds to step ST5.
Here, quantization scale setting circuit 33 sets allocated code quantity every frame from delivered allocated code quantity every GOP by taking into consideration picture type (I picture, P picture, B picture) in actual encoding, e.g., picturetype shown in FIG. 4. In a more practical sense, allocated code quantity with respect to I picture is increased, allocated code quantity with respect to B picture is decreased, and allocated code quantity with respect to P picture is caused to beintermediate therebetween.
At the subsequent step ST5, whether or not encoding processing has been completed with respect to all frames (sequence) to which the same picture size or the same transfer rate is applied is judged. If so, the processing is completed. Incontrast, if not so, the processing operation returns to step ST1. Thus, a variable rate encoding such that encoding rate changes every GOP is realized. Even if pictures (frames) of complicated pattern are successive, there is no possibility thatquantization step size is caused to be large with respect to these pictures as in the conventional apparatus. As a result, it is possible to obtain uniform high picture quality over the entirety. Further, since allocated code quantity every GOP isdetermined on the basis of two pictures in this embodiment, higher speed processing can be carried out as compared to the above-described embodiment. It is to be noted that it is a matter of course to determine allocated code quantities of respectiveGOPs on the basis of data quantities of all pictures within GOP.
It should be noted that this invention is not limited to the above-described embodiments. While, e.g., transform coding employs DCT in the above-described embodiments, so called Strato transform, Haar transform or Wavelet transform, etc. may beemployed.
Industrial Applicability
As is clear from the foregoing description, this invention employs a scheme to allow an input video signal to undergo encoding, e.g., predictive encoding, DCT processing, quantization at fixed quantization step size and variable length encodingto generate first encoded data to determine (calculate) allocated code quantity every frame or every GOP on the basis of data quantity every predetermined time, e.g., every frame or every GOP of the first encoded data and total quantity of usable data toencode the input video signal every predetermined time on the basis of the allocated code quantity to generate second encoded data. Thus, variable rate encoding such that encoding rate changes every predetermined time is realized. As a result, even ifpictures (frames) of complicated are successive, there is no possibility that quantization step size is caused to be large with respect to these pictures as in the conventional apparatus. Thus, uniform high picture quality can be obtained over theentirety.
Further, since second encoded data obtained in a manner as described above has variable rate, in the case where such encoded data is recorded onto picture recording media, limited memory capacity can be effectively used, and recording time ofpicture recording media can be prolonged. In addition, picture data of high picture quality uniform through the entirety can be reproduced from the picture recording media.
* * * * * |
|
|
|
