CA2034418A1 - Embedded coding system for video signals - Google Patents
Embedded coding system for video signalsInfo
- Publication number
- CA2034418A1 CA2034418A1 CA002034418A CA2034418A CA2034418A1 CA 2034418 A1 CA2034418 A1 CA 2034418A1 CA 002034418 A CA002034418 A CA 002034418A CA 2034418 A CA2034418 A CA 2034418A CA 2034418 A1 CA2034418 A1 CA 2034418A1
- Authority
- CA
- Canada
- Prior art keywords
- low frequency
- coefficients
- prediction
- video image
- signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Abandoned
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/30—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using hierarchical techniques, e.g. scalability
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/61—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding
- H04N19/619—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding the transform being operated outside the prediction loop
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
- Television Signal Processing For Recording (AREA)
Abstract
An Embedded Coding System for Video Signals Abstract A video processing system is disclosed that separates and separately encodes and decodes the low and high spatial frequency coefficients of images for transmission or storage. Each block of an image is transformed into the frequency domain. High frequency coefficients of the resulting transform matrix are separated from the low frequency coefficients. The low frequency coefficients are motion prediction compensated to derive motion vectors and a prediction error signal. The motion vectors, prediction error signal and high frequency coefficients are channel encoded for storage or transmission (FIG. 1). In a receiver, the motion vectors and prediction error signal are used to reconstruct a low frequency motion-compensated version of the image. The high frequency coefficients are inverse transformed into the pel domain and are combined with the reconstructed low frequency version of the image to reconstruct a version of the original image (FIG. 2).
Description
An Embedded Coding System for Video Signals Technical Field This invention relates to signal coding systems and, more particularly, to encoding and decoding video signals of moving images suitable for transmission or for storage.
Back~round of the Invention Prior video signal interframe coding, transmission and reproduction systems initially encode a base frame image. This base frarne image incorporatesboth the low and the high spatial frequency components of the image. The base 10 frame image is transmitted and thereafter only information representative of the difference between the base frame and each subsequent image is transmitted. Bothlow and high frequency spatial components are contained within the difference information. Using the base frame image and the subsequent difference information a high quality version, i.e., an accurate or error free representation, of any desired 15 subsequent image can be reconstructed. If difference information is lost during transmission, only a low quality version of the desired image, i.e., an inaccurate or error strewn representation, can be reconstructed. Such prior a~angements are inefficient because they lack the ability to rapidly reproduce high quality images in random access applications. Sin~ilarly, this prior arrangement requires a long time to 20 recover from loss of the difference inforrnation caused by transmission errors. The inefficiency and delays are caused by the leveraged encoding of each image beyond the base frarne since all of the information prior to the selected frame is required, i.e., all the information contained in the base frame and all the difference information relating to each sequentially subsequent image. However, it is known that 25 interframe coding tends to reduce the number of bits that are required to be transmitted.
In a prior attempt to reduce the reconstruction difficulty a new base frame is periodically incorporated into the bit stream. However, this technique drarnatically increases the average number of bits required to represent all the30 information in an average frarne because each base f:rarne is encoded using intrafirame coding and all of the information representing a base frame is used by each subsequent image. It is kno~wn that intraframe coding tends to require a larger number of bits than interframe coding. In intrafrarnç coding, characteristics ofneighboring picture elements (pels or pixels) are predicted based on the values of those characteristics of neighboring pels in the same frame, and the error or difference between the actual value and the predicted value is encoded. This type of encoding is illustrated in an article entitled "Adaptive Coding of Monochrome and Color Images", EEE Trans. Communications, ~rol. COM-25, pp. 1285-1292, Nov.
5 1977.
Summary of the Invention The problems with prior image coding and reproduction systems are overcome, in accordance with an aspect of the invention, by encoding and transmitdng a low spatial frequency representadon of an image using interframe 10 prediction techniques. High spatial frequency coefficients of the image are encoded directly for transmission.
In a specific embodiment of an encoder, a video signal is divided into blocks comprising an array of pels. Each block is transformed into the frequencydomain. The high and the low frequency coefficients of each block are separately15 extracted frorn the ~esulting transformed matli?c. A version of the original image represented by the low frequency coefficients of each block is modon compensatedto yield a prediction error signal and motion vectors. The prediction error signal, the motion vectors and the high frequency coefficients are encoded using a method suitable to the medium of transrmission or storage. The prediction error signal and 20 motion vectors are assigned a high priority. The high frequency coefficients are assigned a low priority and may be dropped by a network during congestion or notretrieved from storage.
A receiver decodes and separates the prediction error signal, the motion vectors and the high frequency coefficients. The prediction error signal and motion 25 vectors are used to recons~uct a motion compensated version of the low frequency image. The high frequency coefficients are inverse transforrned and the resulting reconstructed high frequency version of the image is combined with the reconstructed, motion compensated low frequency version of the image to produce a reconstructed version of the original image.
3û Brief Description of the Drawin~
FIG. 1 shows, in simplified block diagram form, an encoder embodying aspects of the invention; and FIG. 2 shows, in sitnplified block diagram form, a decoder embodying aspects of the invention.
1l r ~ 3 ~
_etailed Description There is some experimental evidence that for regions of high motion in a sequence of moving image firames the interframe co~relation of high frequency coefScients of a discrete cosine transform (DCT) of those regions is approximately zero (0). Therefore, the bit-rate to code these coefficients in an intraframe format is not much different than coding them in an inter~rame frame or Modon Compensated format. FIG. 1 shows an encoder that encodes the high frequency coefficients of a DCT of a sequence of image frames in an intraframe format. An additional DCI
beyond those used in prior motion compensated systems is used as a low-pass loop-lO filter to generate a low frequency prediction.
In an example encoder, an original video signal VIDlN is first delayedby delay 102 for reasons to be given later. The delayed signal 104 is then converted from the pel domain to the Discrete Cosine Transform domain by DCT 106. An example DCI` 106 groups input pels, i.e. picture elements, into 2- dimensional 15 blocks, e.g., 8x8 pels each. DCI' 106 produces a set of output frequency coefficients that are grouped into 2-dimensional blocks of the same size as the input bloclcs. The transform coefficients supplied as an output by DC~ 106 are then partidoned by separator 108 into two groups, namely, high spatial frequency coefficients and low spadal frequency coefficients. The high spatial frequency coefficients, HFREQ, are 20 selected based upon a predetermined threshold and are supplied to quandzer 111).
The remaining low spatdal frequency coefficients, LFREQ, are supplied to subtracter 112.
High spatial frequency coefficients HFREQ are quandzed by quantizer 110. A quandzer reduces the number of levels available for coefficients to assume.
2S The quantizing c,an optionally be made responsive to the quantity of inforrnation stored in elasdc store 114 so as to prevent elastic store 114 from overflowing. The output from quantizer llû, QHFREQ, is supplied to elastic store 114. Elastic store 114 subsequently supplies QHFREQ to channel coder 116 for encoding appropriate for transmission or storage, depending on the application. If elastic store 114 is not 30 present, QHFREQ is supplied direcdy to channel coder 116 for encoding. Low spatial frequency coefficients LFREQ are coded by interframe prediction techniques that are well known in the art and exploit the correlation normally present between image frames. For example, here we show simple Motion Compensation prediction.
(See Dig;ital Pictures Representa~on and Compression by Arun N. Netravali and 35 Barry G. Haskell, Plenum Press 1988 pp. 334-340.) Other methods might includeConditional Motion Compensated Interpolation (see co-pending application Serial ~ ' t ~ 7 Number 413,520 Filed on Septernber 27, 1989 and allowed on April 5, 1990).
Low frequency coefficient prediction signal PRED, supplied to subtracter 112, is subtracted from the low frequency coefficients LFREQ to yield an initial low frequency prediction error signal LPERR. Low frequency prediction error S signal LPERR is quantized by quandzer 118 and then supplied as low frequency prediction error signal QLPERR tO elastic store 114. In turn, QLPERR is subsequently encoded and supplied for transmission by channel coder 116. Data supplied to channel coder 116 from elastic store 114 is encoded using well knownreversible data compression methods. A serial bit-stream output that can be 10 transmitted or stored is supplied as an output from channel coder 116.
It should be noted that in some appl*ations quantizers 110 and 118 and elastic store 114 may be eliminated. If quantizer 118 is eliminated, low frequency prediction error signal LPERR is supplied directly to elastic store 114 and adder 120.
Additionally, a significance detector (not shown~ may be incorporated into quantiær 15 118. This detector would determine whether the values of the low frequency coefficients in a block of QLPERR were too small to warrant transn~ission.
Insignificant blocks would be rçplaced by zero (0) before being supplied as an outpu~.
After any significance thresholding is performed, the quantiæd 20 prediction error signal QLPERR is added to low frequency coefflcient prediction signal PRED by adder 120 to produce "reconstructed" low frequency coefficients RLFREQ. These reconstructed coefficients are inverse transformed by INVERSE
DCT 122 to form a low frequency pel domain signal LPEL. LPEL is passed into a frame memory 123 of motion compensation prediction unit 124. Motion 25 compensation plediction units are well known in the art. Original signal VIDIN is also supplied to motion compensation prediction unit 124. Delay 102 is required to insure that coefficients LFREQ correspond in time to predicted coefficients PREDdue to the processing delay in motion compensation plediction lmit 124. Motion compensadon prediction unit 124 compares each block of the original signal VIDIN30 with signal LPEL stored in its frarne memory 123 and calculates a motion vector for each block. Motion vectors are necessary to reconstruct a shift-matrix for the motion compensated low frequency version of the image. The motion vectors are supplied as output signal MOVE~T to elastic store 114 for subsequent transmission by channel coder 116. Channel coder 116 assigns a higher priority ~o the low frequency 35 quantized prediction error signal and the motion vectors than is assigned to the high f~quency coefflcients.
~ it '~
S
Modon compensation prediction unit 124 also generates MCPEL, which is a block of pels that is shifted by an amount corresponding to each of the aforementioned motion vectors. MCPEL represents the prediction to be used in coding the corresponding block of original pels now being supplied as an output by 5 delay 102. Thus, an original block of pels on signal line 104 passes into DCT 106 at the sarne time as the corresponding predicted MCPEL block of pels passes into DCr 126. The low frequency coefficients supplied as an output from DCI' 126 appear as signal PRED, which is the aforementioned prediction for the low frequency coefficients LFREQ. DCT 126 does not generate any high frequency coefficients.
A corresponding example decoder is shown in FIG. 2. Channel decoder 202 decodes a signal supplied from the channel or from storage and supplies as outputs signals QHFREQ2, QLPERR2 and MOVECT2. IJnder errorless conditions, a signal with a suffix of 2 is a reconstructed version of the signal having the same base name as shown in Fig. 1. High frequency coefficients QHFREQ2 are inverse 15 transformed by INVERSE D(~ 204 and the resulting signal, HPEL, is supplied asan output. Signal HPEL represents a high frequency pel domain version of the original irnage. Quantized low frequency coefficient prediction error signal QLPERR2 is added to prediction signal P~D2 by adder 206 to generate reconstructed low frequency coefficients RLFREQ2, which are supplied as an 20 output. Reconstructed low frequency coefficients RLFREQ2 are inverse transformed by INVERSE DCI' 208 to form a signal LPEL2 which represents a low frequency motion compensated pel domain version of the original image. Adder 210 sums signals LPEL2 ancl HPEL and supplies the result as output signal VIDOUT. Signal VIDOUT is a full frequency reconstructed version of the original video image signal 25 in the pel domain.
Addi~onally, low frequency pel signal LPEL2 is supplied to a frame memory 211 of motion compensation prediction unit 212. Motion compensation prediction unit 212 is also supplied with motion vectors MOVECT2. A shifted block of pels, signal MCPEL2, which represents an uncorrected motion 30 compensation p~dicted version of the low ~equency pel domain irnage is generated by motion compensation prediction unit 212. Signal MCPEL2 is supplied to DCI
214 wherein it is transfo~med into the frequency domain. The resulting output ofD(~ 214 is a set of low frequency prediction coefficients, PRED2, which are supplied to abovementioned adder 206. D~ 214 also does not generate any high 35 frequency coefficients.
;~ 9, ~ 3 At start up, to initialize the motion compensation unit and establish a base frame, motion compensation of the low frequency image is suspended for one frame. Thereafter, at predetermined intervals, motion compensation of the low frequency image may be suspended for additional frame periods. This is useful in5 applications requiring random access from storage such as compact disks or where the image signal is to be transported over a paclcet network that cannot guaranty 100% delivery for packets that contain the motion vectors and low fiequency error signal. Networks that guaranty 100% delivery would not require suspension of motion compensation.
In the encoder (FIG. 1), the time of suspension is determined by a control mechanism (not shown) and effected by changing the logical condition of signal CI~RL supplied to motion compensation prediction unit 124. Suspension of motion compensation causes signals MOVECT, MCPEL and PRED to have values of æro (0). Nothing is therefore subtracted from the low frequency coefficients at 15 subtracter 112 and LPERR and low frequency coefficients LFREQ are therefore equal. This causes the low frequency coefficients LFREQ to be directly encoded rather than prediction error signal LPERR and motion vectors MOVECI thereby resulting in the transmission of a new base frame.
Similarly in the receiver (FIG. 2), a corresponding time of suspension of 20 motion compensation is dete~nined by a control mechanism (also not shown) andeffected by changing the logical condition of signal CI'RL supplied to motion compensation prediction unit 212. When motion compensation is suspended in the encoder (E;IG. 1) signal QLPERR2 in the receiver (E~IG. 2) is equal to the low frequency coefficients LFREQ (FIG. 1). Suspension of motion compensation causes 25 signals MCPEL2 and PRED2 to have values of zero (0). In turn, signal RLFREQ2 is equal to signal QLPERR2 and correspondingly LPEL2 represents a low frequency base image.
Separating the DCI coefficients has several advantages. First, the low frequency image and the high frequency image can be computed in parallel, since 30 errors in~roduced by compressing the low frequency information do not translate into additional errors in the high frequency information. Also, since not all D~
coefficients ~e necessa~y to generate either the low frequency image or the highfrequency irnage, the entire DCI need not be computed for each, and an intelligent implementation of the DCT can improve the efficiency of the computation.
35 Additionally, a better prediction of the coefficients in the motion compensation loop is obtained, since fewer coefficients are predicted. Furthermore, by motion compensating only the low frequencies, the overall image quality is not sacrificed in areas with large motion. In this case, motion compensation cannot adequately predict the high frequencies anyway, so little is lost by not motion compensating them. This coding method has all the advantages for packet transn~ission inherent in S any variable bit-rate method. A balance is struck between data compression androbustness to packet loss, by motion compensating low frequencies, yet intraframe coding the high frequencies.
An embedded code, as is well known, is a digital code that allows bi~s or symbols to be dropped and inserted without causing mismatch between the 10 memories of the encoder and the decoder. The encoding and decoding system descnbed above is such a system. Thus, another advantage of this system is that if some of the high frequency coefficients transmitted by the encoder are lost they can be zero-filled at the decoder and an image with less error than prior systems will be displayed. Purther, only the low frequency portions of the intraframe coded base15 frame and subsequent interfrarne coded frames until and including a desired random frarne need be read. These are then combined with the high frequency portion of the desired frame and displayed. Therefore, a high quality image can be randomly accessed more rapidly than with prior coded video systems.
Back~round of the Invention Prior video signal interframe coding, transmission and reproduction systems initially encode a base frame image. This base frarne image incorporatesboth the low and the high spatial frequency components of the image. The base 10 frame image is transmitted and thereafter only information representative of the difference between the base frame and each subsequent image is transmitted. Bothlow and high frequency spatial components are contained within the difference information. Using the base frame image and the subsequent difference information a high quality version, i.e., an accurate or error free representation, of any desired 15 subsequent image can be reconstructed. If difference information is lost during transmission, only a low quality version of the desired image, i.e., an inaccurate or error strewn representation, can be reconstructed. Such prior a~angements are inefficient because they lack the ability to rapidly reproduce high quality images in random access applications. Sin~ilarly, this prior arrangement requires a long time to 20 recover from loss of the difference inforrnation caused by transmission errors. The inefficiency and delays are caused by the leveraged encoding of each image beyond the base frarne since all of the information prior to the selected frame is required, i.e., all the information contained in the base frame and all the difference information relating to each sequentially subsequent image. However, it is known that 25 interframe coding tends to reduce the number of bits that are required to be transmitted.
In a prior attempt to reduce the reconstruction difficulty a new base frame is periodically incorporated into the bit stream. However, this technique drarnatically increases the average number of bits required to represent all the30 information in an average frarne because each base f:rarne is encoded using intrafirame coding and all of the information representing a base frame is used by each subsequent image. It is kno~wn that intraframe coding tends to require a larger number of bits than interframe coding. In intrafrarnç coding, characteristics ofneighboring picture elements (pels or pixels) are predicted based on the values of those characteristics of neighboring pels in the same frame, and the error or difference between the actual value and the predicted value is encoded. This type of encoding is illustrated in an article entitled "Adaptive Coding of Monochrome and Color Images", EEE Trans. Communications, ~rol. COM-25, pp. 1285-1292, Nov.
5 1977.
Summary of the Invention The problems with prior image coding and reproduction systems are overcome, in accordance with an aspect of the invention, by encoding and transmitdng a low spatial frequency representadon of an image using interframe 10 prediction techniques. High spatial frequency coefficients of the image are encoded directly for transmission.
In a specific embodiment of an encoder, a video signal is divided into blocks comprising an array of pels. Each block is transformed into the frequencydomain. The high and the low frequency coefficients of each block are separately15 extracted frorn the ~esulting transformed matli?c. A version of the original image represented by the low frequency coefficients of each block is modon compensatedto yield a prediction error signal and motion vectors. The prediction error signal, the motion vectors and the high frequency coefficients are encoded using a method suitable to the medium of transrmission or storage. The prediction error signal and 20 motion vectors are assigned a high priority. The high frequency coefficients are assigned a low priority and may be dropped by a network during congestion or notretrieved from storage.
A receiver decodes and separates the prediction error signal, the motion vectors and the high frequency coefficients. The prediction error signal and motion 25 vectors are used to recons~uct a motion compensated version of the low frequency image. The high frequency coefficients are inverse transforrned and the resulting reconstructed high frequency version of the image is combined with the reconstructed, motion compensated low frequency version of the image to produce a reconstructed version of the original image.
3û Brief Description of the Drawin~
FIG. 1 shows, in simplified block diagram form, an encoder embodying aspects of the invention; and FIG. 2 shows, in sitnplified block diagram form, a decoder embodying aspects of the invention.
1l r ~ 3 ~
_etailed Description There is some experimental evidence that for regions of high motion in a sequence of moving image firames the interframe co~relation of high frequency coefScients of a discrete cosine transform (DCT) of those regions is approximately zero (0). Therefore, the bit-rate to code these coefficients in an intraframe format is not much different than coding them in an inter~rame frame or Modon Compensated format. FIG. 1 shows an encoder that encodes the high frequency coefficients of a DCT of a sequence of image frames in an intraframe format. An additional DCI
beyond those used in prior motion compensated systems is used as a low-pass loop-lO filter to generate a low frequency prediction.
In an example encoder, an original video signal VIDlN is first delayedby delay 102 for reasons to be given later. The delayed signal 104 is then converted from the pel domain to the Discrete Cosine Transform domain by DCT 106. An example DCI` 106 groups input pels, i.e. picture elements, into 2- dimensional 15 blocks, e.g., 8x8 pels each. DCI' 106 produces a set of output frequency coefficients that are grouped into 2-dimensional blocks of the same size as the input bloclcs. The transform coefficients supplied as an output by DC~ 106 are then partidoned by separator 108 into two groups, namely, high spatial frequency coefficients and low spadal frequency coefficients. The high spatial frequency coefficients, HFREQ, are 20 selected based upon a predetermined threshold and are supplied to quandzer 111).
The remaining low spatdal frequency coefficients, LFREQ, are supplied to subtracter 112.
High spatial frequency coefficients HFREQ are quandzed by quantizer 110. A quandzer reduces the number of levels available for coefficients to assume.
2S The quantizing c,an optionally be made responsive to the quantity of inforrnation stored in elasdc store 114 so as to prevent elastic store 114 from overflowing. The output from quantizer llû, QHFREQ, is supplied to elastic store 114. Elastic store 114 subsequently supplies QHFREQ to channel coder 116 for encoding appropriate for transmission or storage, depending on the application. If elastic store 114 is not 30 present, QHFREQ is supplied direcdy to channel coder 116 for encoding. Low spatial frequency coefficients LFREQ are coded by interframe prediction techniques that are well known in the art and exploit the correlation normally present between image frames. For example, here we show simple Motion Compensation prediction.
(See Dig;ital Pictures Representa~on and Compression by Arun N. Netravali and 35 Barry G. Haskell, Plenum Press 1988 pp. 334-340.) Other methods might includeConditional Motion Compensated Interpolation (see co-pending application Serial ~ ' t ~ 7 Number 413,520 Filed on Septernber 27, 1989 and allowed on April 5, 1990).
Low frequency coefficient prediction signal PRED, supplied to subtracter 112, is subtracted from the low frequency coefficients LFREQ to yield an initial low frequency prediction error signal LPERR. Low frequency prediction error S signal LPERR is quantized by quandzer 118 and then supplied as low frequency prediction error signal QLPERR tO elastic store 114. In turn, QLPERR is subsequently encoded and supplied for transmission by channel coder 116. Data supplied to channel coder 116 from elastic store 114 is encoded using well knownreversible data compression methods. A serial bit-stream output that can be 10 transmitted or stored is supplied as an output from channel coder 116.
It should be noted that in some appl*ations quantizers 110 and 118 and elastic store 114 may be eliminated. If quantizer 118 is eliminated, low frequency prediction error signal LPERR is supplied directly to elastic store 114 and adder 120.
Additionally, a significance detector (not shown~ may be incorporated into quantiær 15 118. This detector would determine whether the values of the low frequency coefficients in a block of QLPERR were too small to warrant transn~ission.
Insignificant blocks would be rçplaced by zero (0) before being supplied as an outpu~.
After any significance thresholding is performed, the quantiæd 20 prediction error signal QLPERR is added to low frequency coefflcient prediction signal PRED by adder 120 to produce "reconstructed" low frequency coefficients RLFREQ. These reconstructed coefficients are inverse transformed by INVERSE
DCT 122 to form a low frequency pel domain signal LPEL. LPEL is passed into a frame memory 123 of motion compensation prediction unit 124. Motion 25 compensation plediction units are well known in the art. Original signal VIDIN is also supplied to motion compensation prediction unit 124. Delay 102 is required to insure that coefficients LFREQ correspond in time to predicted coefficients PREDdue to the processing delay in motion compensation plediction lmit 124. Motion compensadon prediction unit 124 compares each block of the original signal VIDIN30 with signal LPEL stored in its frarne memory 123 and calculates a motion vector for each block. Motion vectors are necessary to reconstruct a shift-matrix for the motion compensated low frequency version of the image. The motion vectors are supplied as output signal MOVE~T to elastic store 114 for subsequent transmission by channel coder 116. Channel coder 116 assigns a higher priority ~o the low frequency 35 quantized prediction error signal and the motion vectors than is assigned to the high f~quency coefflcients.
~ it '~
S
Modon compensation prediction unit 124 also generates MCPEL, which is a block of pels that is shifted by an amount corresponding to each of the aforementioned motion vectors. MCPEL represents the prediction to be used in coding the corresponding block of original pels now being supplied as an output by 5 delay 102. Thus, an original block of pels on signal line 104 passes into DCT 106 at the sarne time as the corresponding predicted MCPEL block of pels passes into DCr 126. The low frequency coefficients supplied as an output from DCI' 126 appear as signal PRED, which is the aforementioned prediction for the low frequency coefficients LFREQ. DCT 126 does not generate any high frequency coefficients.
A corresponding example decoder is shown in FIG. 2. Channel decoder 202 decodes a signal supplied from the channel or from storage and supplies as outputs signals QHFREQ2, QLPERR2 and MOVECT2. IJnder errorless conditions, a signal with a suffix of 2 is a reconstructed version of the signal having the same base name as shown in Fig. 1. High frequency coefficients QHFREQ2 are inverse 15 transformed by INVERSE D(~ 204 and the resulting signal, HPEL, is supplied asan output. Signal HPEL represents a high frequency pel domain version of the original irnage. Quantized low frequency coefficient prediction error signal QLPERR2 is added to prediction signal P~D2 by adder 206 to generate reconstructed low frequency coefficients RLFREQ2, which are supplied as an 20 output. Reconstructed low frequency coefficients RLFREQ2 are inverse transformed by INVERSE DCI' 208 to form a signal LPEL2 which represents a low frequency motion compensated pel domain version of the original image. Adder 210 sums signals LPEL2 ancl HPEL and supplies the result as output signal VIDOUT. Signal VIDOUT is a full frequency reconstructed version of the original video image signal 25 in the pel domain.
Addi~onally, low frequency pel signal LPEL2 is supplied to a frame memory 211 of motion compensation prediction unit 212. Motion compensation prediction unit 212 is also supplied with motion vectors MOVECT2. A shifted block of pels, signal MCPEL2, which represents an uncorrected motion 30 compensation p~dicted version of the low ~equency pel domain irnage is generated by motion compensation prediction unit 212. Signal MCPEL2 is supplied to DCI
214 wherein it is transfo~med into the frequency domain. The resulting output ofD(~ 214 is a set of low frequency prediction coefficients, PRED2, which are supplied to abovementioned adder 206. D~ 214 also does not generate any high 35 frequency coefficients.
;~ 9, ~ 3 At start up, to initialize the motion compensation unit and establish a base frame, motion compensation of the low frequency image is suspended for one frame. Thereafter, at predetermined intervals, motion compensation of the low frequency image may be suspended for additional frame periods. This is useful in5 applications requiring random access from storage such as compact disks or where the image signal is to be transported over a paclcet network that cannot guaranty 100% delivery for packets that contain the motion vectors and low fiequency error signal. Networks that guaranty 100% delivery would not require suspension of motion compensation.
In the encoder (FIG. 1), the time of suspension is determined by a control mechanism (not shown) and effected by changing the logical condition of signal CI~RL supplied to motion compensation prediction unit 124. Suspension of motion compensation causes signals MOVECT, MCPEL and PRED to have values of æro (0). Nothing is therefore subtracted from the low frequency coefficients at 15 subtracter 112 and LPERR and low frequency coefficients LFREQ are therefore equal. This causes the low frequency coefficients LFREQ to be directly encoded rather than prediction error signal LPERR and motion vectors MOVECI thereby resulting in the transmission of a new base frame.
Similarly in the receiver (FIG. 2), a corresponding time of suspension of 20 motion compensation is dete~nined by a control mechanism (also not shown) andeffected by changing the logical condition of signal CI'RL supplied to motion compensation prediction unit 212. When motion compensation is suspended in the encoder (E;IG. 1) signal QLPERR2 in the receiver (E~IG. 2) is equal to the low frequency coefficients LFREQ (FIG. 1). Suspension of motion compensation causes 25 signals MCPEL2 and PRED2 to have values of zero (0). In turn, signal RLFREQ2 is equal to signal QLPERR2 and correspondingly LPEL2 represents a low frequency base image.
Separating the DCI coefficients has several advantages. First, the low frequency image and the high frequency image can be computed in parallel, since 30 errors in~roduced by compressing the low frequency information do not translate into additional errors in the high frequency information. Also, since not all D~
coefficients ~e necessa~y to generate either the low frequency image or the highfrequency irnage, the entire DCI need not be computed for each, and an intelligent implementation of the DCT can improve the efficiency of the computation.
35 Additionally, a better prediction of the coefficients in the motion compensation loop is obtained, since fewer coefficients are predicted. Furthermore, by motion compensating only the low frequencies, the overall image quality is not sacrificed in areas with large motion. In this case, motion compensation cannot adequately predict the high frequencies anyway, so little is lost by not motion compensating them. This coding method has all the advantages for packet transn~ission inherent in S any variable bit-rate method. A balance is struck between data compression androbustness to packet loss, by motion compensating low frequencies, yet intraframe coding the high frequencies.
An embedded code, as is well known, is a digital code that allows bi~s or symbols to be dropped and inserted without causing mismatch between the 10 memories of the encoder and the decoder. The encoding and decoding system descnbed above is such a system. Thus, another advantage of this system is that if some of the high frequency coefficients transmitted by the encoder are lost they can be zero-filled at the decoder and an image with less error than prior systems will be displayed. Purther, only the low frequency portions of the intraframe coded base15 frame and subsequent interfrarne coded frames until and including a desired random frarne need be read. These are then combined with the high frequency portion of the desired frame and displayed. Therefore, a high quality image can be randomly accessed more rapidly than with prior coded video systems.
Claims (10)
1. Apparatus for use in a video coding system being CHARACTERIZED
BY:
means for receiving as an input an original digitized video image signal;
means for delaying said video image signal;
means for deriving frequency transform coefficients from said delayed video image signal;
means for separating said frequency transform coefficients into a set of low frequency coefficients and a set of high frequency coefficients, said separation being based upon a predetermined threshold;
means responsive to said set of low frequency coefficients and said input video image signal for generating and supplying as an output a representation of a motion compensated low frequency version of said original video image signal that includes a set of motion vectors and a prediction error signal;
means for encoding said set of high frequency coefficients, said prediction error signal and said set of motion vectors into a predetermined format specific to a predetermined output medium; and means for interfacing said predetermined format to said predetermined output medium.
BY:
means for receiving as an input an original digitized video image signal;
means for delaying said video image signal;
means for deriving frequency transform coefficients from said delayed video image signal;
means for separating said frequency transform coefficients into a set of low frequency coefficients and a set of high frequency coefficients, said separation being based upon a predetermined threshold;
means responsive to said set of low frequency coefficients and said input video image signal for generating and supplying as an output a representation of a motion compensated low frequency version of said original video image signal that includes a set of motion vectors and a prediction error signal;
means for encoding said set of high frequency coefficients, said prediction error signal and said set of motion vectors into a predetermined format specific to a predetermined output medium; and means for interfacing said predetermined format to said predetermined output medium.
2. The apparatus as defined in claim 1 further CHARACTERIZED BY
elastic storage means for storing said set of high frequency coefficients, said prediction error signal and said set of motion vectors and for supplying said set of high frequency coefficients, said prediction error signal and said set of motionvectors as an output to said means for encoding.
elastic storage means for storing said set of high frequency coefficients, said prediction error signal and said set of motion vectors and for supplying said set of high frequency coefficients, said prediction error signal and said set of motionvectors as an output to said means for encoding.
3. The apparatus as defined in claim 1 further CHARACTERIZED BY
means for quantizing said set of high frequency coefficients.
means for quantizing said set of high frequency coefficients.
4. The apparatus as defined in claim 1 further CHARACTERIZED BY
means for generating an initial error signal and means for quantizing responsive to said initial error signal and supplying as an output said prediction error signal.
means for generating an initial error signal and means for quantizing responsive to said initial error signal and supplying as an output said prediction error signal.
5. The apparatus as defined in claim 1 CHARACTERIZED IN THAT
said means for generating and supplying includes means for generating a reconstructed low frequency pel domain version of said digitized video image signal, motion compensation prediction means responsive to a current frame of said original digitized video image signal and an at least one stored frame of said reconstructed low frequency pel domain version of said digitized video image signal for generating a motion compensated pel domain prediction signal and motion vectors, means for transforming said motion compensated pel domain prediction signal into said set of low frequency prediction coefficients and means responsive to said set of low frequency coefficients and said set of low frequency prediction coefficients forgenerating and supplying said prediction error signal.
said means for generating and supplying includes means for generating a reconstructed low frequency pel domain version of said digitized video image signal, motion compensation prediction means responsive to a current frame of said original digitized video image signal and an at least one stored frame of said reconstructed low frequency pel domain version of said digitized video image signal for generating a motion compensated pel domain prediction signal and motion vectors, means for transforming said motion compensated pel domain prediction signal into said set of low frequency prediction coefficients and means responsive to said set of low frequency coefficients and said set of low frequency prediction coefficients forgenerating and supplying said prediction error signal.
6. The apparatus as defined in claim 5 CHARACTERIZED IN THAT
said means for generating and supplying further includes means for supplying as an output a signal representing a combination of said low frequency prediction coefficients and said prediction error signal to said means for generating said reconstructed low frequency pel domain version of said digitized video image signal.
said means for generating and supplying further includes means for supplying as an output a signal representing a combination of said low frequency prediction coefficients and said prediction error signal to said means for generating said reconstructed low frequency pel domain version of said digitized video image signal.
7. The apparatus as defined in claim 5 CHARACTERIZED IN THAT
said means for generating and supplying further includes means for supplying a quantized version of said frequency domain prediction error signal and means forsupplying as an output a signal representing the combination of said low frequency prediction coefficients and quantized version of said frequency domain prediction error signal to said means for generating said reconstructed low frequency pel domain version of said digitized video image signal.
said means for generating and supplying further includes means for supplying a quantized version of said frequency domain prediction error signal and means forsupplying as an output a signal representing the combination of said low frequency prediction coefficients and quantized version of said frequency domain prediction error signal to said means for generating said reconstructed low frequency pel domain version of said digitized video image signal.
8. Apparatus for use in a video decoding system comprising:
means for interfacing to a predetermined input medium for receiving, separating, decoding and supplying as an output a set of high frequency coefficients, a prediction error signal and a set of motion vectors from data supplied from said medium in a predetermined format;
means for transforming said set of high frequency coefficients into a high frequency pel domain version of an original video image;
means responsive to said prediction error signal and said set of motion vectors for deriving a motion compensated low frequency pel domain version of said original video image signal;
means for combining said high frequency pel domain version of said original video image and said motion compensated low frequency pel domain version of said original video image signal into a reconstructed full frequency version of said original video image signal in the pel domain; and means for supplying said reconstructed full frequency version of said original video image signal in the pel domain as an output.
means for interfacing to a predetermined input medium for receiving, separating, decoding and supplying as an output a set of high frequency coefficients, a prediction error signal and a set of motion vectors from data supplied from said medium in a predetermined format;
means for transforming said set of high frequency coefficients into a high frequency pel domain version of an original video image;
means responsive to said prediction error signal and said set of motion vectors for deriving a motion compensated low frequency pel domain version of said original video image signal;
means for combining said high frequency pel domain version of said original video image and said motion compensated low frequency pel domain version of said original video image signal into a reconstructed full frequency version of said original video image signal in the pel domain; and means for supplying said reconstructed full frequency version of said original video image signal in the pel domain as an output.
9. The apparatus as defined in claim 10 wherein said means for deriving includes motion compensation prediction means responsive to a said motion vectors and an at least one prior stored frame of a reconstructed low frequency pel domain version of said digitized video image signal for generating a current motion compensated prediction pel domain signal, means for transforming said current motion compensated pel domain prediction signal into a set of low frequency prediction coefficients, means responsive to said set of low frequency prediction coefficients and said frequency domain error signal for generating a set of reconstructed low frequency prediction coefficients, means supplied with said reconstructed low frequency prediction coefficients for generating a reconstructed low frequency pel domain version of said digitized video image signal and means for storing said low frequency pel domain version of said digitized video image signal.
10. The apparatus as defined in claim 10 wherein said means for receiving, separating, decoding and supplying further includes means for supplying null value high frequency coefficients if said high frequency coefficients are not supplied from said data within a predetermined length of time.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US07/517,991 US5001561A (en) | 1990-05-02 | 1990-05-02 | Embedded coding system for video signals |
US517,991 | 1990-05-02 |
Publications (1)
Publication Number | Publication Date |
---|---|
CA2034418A1 true CA2034418A1 (en) | 1991-11-03 |
Family
ID=24062078
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CA002034418A Abandoned CA2034418A1 (en) | 1990-05-02 | 1991-01-17 | Embedded coding system for video signals |
Country Status (5)
Country | Link |
---|---|
US (1) | US5001561A (en) |
EP (1) | EP0454927A3 (en) |
JP (1) | JPH04229791A (en) |
CA (1) | CA2034418A1 (en) |
FI (1) | FI912110A (en) |
Families Citing this family (48)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5140417A (en) * | 1989-06-20 | 1992-08-18 | Matsushita Electric Co., Ltd. | Fast packet transmission system of video data |
US5422675A (en) * | 1990-01-29 | 1995-06-06 | Massachusetts Institute Of Technology | Adaptive modulation/demodulation signal processing |
US5128757A (en) * | 1990-06-18 | 1992-07-07 | Zenith Electronics Corporation | Video transmission system using adaptive sub-band coding |
US5218435A (en) * | 1991-02-20 | 1993-06-08 | Massachusetts Institute Of Technology | Digital advanced television systems |
US5111292A (en) * | 1991-02-27 | 1992-05-05 | General Electric Company | Priority selection apparatus as for a video signal processor |
US5285276A (en) * | 1991-03-12 | 1994-02-08 | Zenith Electronics Corp. | Bi-rate high definition television signal transmission system |
JPH04323963A (en) * | 1991-04-23 | 1992-11-13 | Canon Inc | Picture processing method and device |
JPH04358486A (en) * | 1991-06-04 | 1992-12-11 | Toshiba Corp | High efficiency code signal processing unit |
US6091884A (en) * | 1991-08-19 | 2000-07-18 | Index Systems, Inc. | Enhancing operations of video tape cassette players |
US6487362B1 (en) | 1991-08-19 | 2002-11-26 | Index Systems, Inc. | Enhancing operations of video tape cassette players |
JP3109185B2 (en) * | 1991-11-12 | 2000-11-13 | ソニー株式会社 | Digital VTR |
US5227878A (en) * | 1991-11-15 | 1993-07-13 | At&T Bell Laboratories | Adaptive coding and decoding of frames and fields of video |
US5339108A (en) * | 1992-04-09 | 1994-08-16 | Ampex Corporation | Ordering and formatting coded image data and reconstructing partial images from the data |
US5253055A (en) * | 1992-07-02 | 1993-10-12 | At&T Bell Laboratories | Efficient frequency scalable video encoding with coefficient selection |
US5270813A (en) * | 1992-07-02 | 1993-12-14 | At&T Bell Laboratories | Spatially scalable video coding facilitating the derivation of variable-resolution images |
US5253056A (en) * | 1992-07-02 | 1993-10-12 | At&T Bell Laboratories | Spatial/frequency hybrid video coding facilitating the derivatives of variable-resolution images |
JPH06197334A (en) * | 1992-07-03 | 1994-07-15 | Sony Corp | Picture signal coding method, picture signal decoding method, picture signal coder, picture signal decoder and picture signal recording medium |
JPH0698288A (en) * | 1992-09-10 | 1994-04-08 | Nikon Corp | Still picture storage device |
GB9312022D0 (en) * | 1993-06-10 | 1993-07-28 | Snell & Wilcox Ltd | Video signal processing |
JP2902284B2 (en) * | 1993-11-12 | 1999-06-07 | ケイディディ株式会社 | Video encoding device |
US5566284A (en) * | 1993-12-22 | 1996-10-15 | Matsushita Electric Industrial Co., Ltd. | Apparatus and method for mip-map generation using low-pass filtering based on resolution ratio |
US5452009A (en) * | 1993-12-29 | 1995-09-19 | Zenith Electronics Corp. | Digital transmission system with data rate optimized for noise in transmission medium |
JP2956464B2 (en) * | 1993-12-29 | 1999-10-04 | 日本ビクター株式会社 | Image information compression / decompression device |
US5592226A (en) * | 1994-01-26 | 1997-01-07 | Btg Usa Inc. | Method and apparatus for video data compression using temporally adaptive motion interpolation |
GB2302488B (en) * | 1994-09-20 | 1997-04-09 | Ricoh Kk | Method for compression |
JPH08275160A (en) * | 1995-03-27 | 1996-10-18 | Internatl Business Mach Corp <Ibm> | Discrete cosine conversion method |
JP3788823B2 (en) | 1995-10-27 | 2006-06-21 | 株式会社東芝 | Moving picture encoding apparatus and moving picture decoding apparatus |
SE512291C2 (en) * | 1997-09-23 | 2000-02-28 | Ericsson Telefon Ab L M | Embedded DCT-based still image coding algorithm |
SE9800088D0 (en) | 1997-10-10 | 1998-01-16 | Ericsson Telefon Ab L M | Long filter lossless region of interest coding |
SE9803454L (en) | 1998-10-09 | 2000-04-10 | Ericsson Telefon Ab L M | Procedure and system for coding ROI |
US6289297B1 (en) | 1998-10-09 | 2001-09-11 | Microsoft Corporation | Method for reconstructing a video frame received from a video source over a communication channel |
US6438136B1 (en) | 1998-10-09 | 2002-08-20 | Microsoft Corporation | Method for scheduling time slots in a communications network channel to support on-going video transmissions |
US6618363B1 (en) | 1998-10-09 | 2003-09-09 | Microsoft Corporation | Method for adapting video packet generation and transmission rates to available resources in a communications network |
US6507587B1 (en) | 1998-10-09 | 2003-01-14 | Microsoft Corporation | Method of specifying the amount of bandwidth to reserve for use in network communications |
US6754266B2 (en) * | 1998-10-09 | 2004-06-22 | Microsoft Corporation | Method and apparatus for use in transmitting video information over a communication network |
US6445701B1 (en) * | 1998-10-09 | 2002-09-03 | Microsoft Corporation | Channel access scheme for use in network communications |
US6519004B1 (en) | 1998-10-09 | 2003-02-11 | Microsoft Corporation | Method for transmitting video information over a communication channel |
US6490705B1 (en) | 1998-10-22 | 2002-12-03 | Lucent Technologies Inc. | Method and apparatus for receiving MPEG video over the internet |
US6317462B1 (en) | 1998-10-22 | 2001-11-13 | Lucent Technologies Inc. | Method and apparatus for transmitting MPEG video over the internet |
EP1081958B1 (en) | 1999-08-31 | 2006-02-01 | Lucent Technologies Inc. | Method and apparatus for macroblock DC and AC coefficient prediction for video coding |
US6639943B1 (en) * | 1999-11-23 | 2003-10-28 | Koninklijke Philips Electronics N.V. | Hybrid temporal-SNR fine granular scalability video coding |
US6826232B2 (en) | 1999-12-20 | 2004-11-30 | Koninklijke Philips Electronics N.V. | Fine granular scalable video with embedded DCT coding of the enhancement layer |
KR100359821B1 (en) * | 2000-01-20 | 2002-11-07 | 엘지전자 주식회사 | Method, Apparatus And Decoder For Motion Compensation Adaptive Image Re-compression |
US6647149B2 (en) | 2001-01-03 | 2003-11-11 | Electronics For Imaging, Inc. | Methods and apparatus for securely transmitting and processing digital image data |
WO2006054507A1 (en) | 2004-11-19 | 2006-05-26 | Matsushita Electric Industrial Co., Ltd. | Moving-image encoding method, and moving-image decoding method |
FR2906433B1 (en) * | 2006-09-22 | 2009-02-27 | Canon Kk | METHODS AND DEVICES FOR ENCODING AND DECODING IMAGES, COMPUTER PROGRAM USING THEM AND INFORMATION SUPPORT FOR IMPLEMENTING THEM |
US7939167B2 (en) * | 2008-12-30 | 2011-05-10 | Cheil Industries Inc. | Resin composition |
FR2944936A1 (en) * | 2009-04-23 | 2010-10-29 | Thomson Licensing | METHODS OF ENCODING AND DECODING AN IMAGE DATA BLOCK, CODING AND DECODING DEVICES IMPLEMENTING SAID METHODS |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4245248A (en) * | 1979-04-04 | 1981-01-13 | Bell Telephone Laboratories, Incorporated | Motion estimation and encoding of video signals in the transform domain |
FR2575351B1 (en) * | 1984-12-21 | 1988-05-13 | Thomson Csf | ADAPTIVE METHOD OF ENCODING AND DECODING A SUITE OF IMAGES BY TRANSFORMATION, AND DEVICES FOR CARRYING OUT SAID METHOD |
US4805030A (en) * | 1986-01-27 | 1989-02-14 | Fuji Photo Film Co., Ltd. | Method of image signal encoding by orthogonal transformation |
EP0339589A3 (en) * | 1988-04-28 | 1992-01-02 | Sharp Kabushiki Kaisha | Orthogonal transform coding system for image data |
US4941043A (en) * | 1988-06-14 | 1990-07-10 | Siemens Aktiengesellschaft | Method for reducing blocking artifacts in video scene coding with discrete cosine transformation (DCT) at a low data rate |
JPH0262176A (en) * | 1988-08-26 | 1990-03-02 | Fujitsu Ltd | Picture encoding/transmission equipment |
US4958226A (en) * | 1989-09-27 | 1990-09-18 | At&T Bell Laboratories | Conditional motion compensated interpolation of digital motion video |
-
1990
- 1990-05-02 US US07/517,991 patent/US5001561A/en not_active Expired - Lifetime
- 1990-12-18 EP EP19900313839 patent/EP0454927A3/en not_active Withdrawn
-
1991
- 1991-01-17 CA CA002034418A patent/CA2034418A1/en not_active Abandoned
- 1991-03-18 JP JP3077226A patent/JPH04229791A/en active Pending
- 1991-04-30 FI FI912110A patent/FI912110A/en unknown
Also Published As
Publication number | Publication date |
---|---|
JPH04229791A (en) | 1992-08-19 |
EP0454927A3 (en) | 1993-01-13 |
EP0454927A2 (en) | 1991-11-06 |
FI912110A0 (en) | 1991-04-30 |
US5001561A (en) | 1991-03-19 |
FI912110A (en) | 1991-11-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US5001561A (en) | Embedded coding system for video signals | |
CA2084178C (en) | Statistical multiplexer for a multichannel image compression system | |
EP0691054B1 (en) | Efficient transcoding device and method | |
US4821119A (en) | Method and apparatus for low bit-rate interframe video coding | |
US4723161A (en) | Method and arrangement of coding digital image signals utilizing interframe correlation | |
US5438374A (en) | System and method for filtering video signals | |
EP0892560B1 (en) | Picture encoding method and picture encoding apparatus | |
US5500678A (en) | Optimized scanning of transform coefficients in video coding | |
US5241383A (en) | Pseudo-constant bit rate video coding with quantization parameter adjustment | |
KR100238509B1 (en) | Method and apparatus for decoding video signals | |
JP3425377B2 (en) | Visual object encoding method | |
KR100253931B1 (en) | Approximate mpeg decoder with compressed reference frames | |
KR100317412B1 (en) | Video Signal Compressor Using Noise Reduction | |
CA1278381C (en) | Data compression using orthogonal transform and vector quantization | |
CA1216058A (en) | Predictive coding/decoding apparatus for video signal | |
EP1296524A1 (en) | Process and apparatus for the compression of digital video signals, a system and a computer program product therefor | |
JPH0746594A (en) | Encoding method and its device | |
KR20000053028A (en) | Prediction method and device with motion compensation | |
US7236521B2 (en) | Digital stream transcoder | |
Reibman | DCT-based embedded coding for packet video | |
US7436889B2 (en) | Methods and systems for reducing requantization-originated generational error in predictive video streams using motion compensation | |
US5614953A (en) | Image signal decoding apparatus having an encoding error compensation | |
JPH1013826A (en) | Picture encoder, its method, picture decoder, its method, picture transmitter, picture receiver and recording medium | |
Gharavi | Low bit-rate video transmission for ISDN application | |
Ng et al. | Buffer control algorithm for low bit-rate video compression |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
EEER | Examination request | ||
FZDE | Discontinued |