Motion compensation for fine-grain scalable video

Abstract

A fine-grain scalable video data apparatus, system, method and data structure is disclosed. An encoder (110) for encoding input video data as minimum bitrate macroblock data to produce DCT data having DCT coefficients representing a minimum bitrate version of the macroblock data. The encoder (110) also encodes the input video data as intermediate bitrate macroblock data to produce DCT data having DCT coefficients representing an intermediate bitrate version of the macroblock data. An adaptive motion compensator (132) (whether incorporated within the encoder or externally) communicates with the encoder for predicting whether a decoded version of the intermediate bitrate macroblock data has an accumulated predicted error frame energy exceeding a maximum threshold (228).

Description

FIELD OF THE INVENTION

[0002] The present disclosure is directed towards fine-grain scalable (“FGS”) video encoders, and in particular, to an adaptive motion compensation technique for coding of video data using fine-grain scalability.

BACKGROUND OF THE INVENTION

[0003] Digital video data is often processed and transferred in the form of bit streams. A bit stream is fine-grain scalable (“FGS”) if the bit stream can be decoded at any one of a finely spaced set of bitrates between pre-determined minimum and maximum rates. Unfortunately, this type of scalability typically results in a coding efficiency that is significantly less than that of a non-scalable video coder-decoder (“CODEC”).

[0004] The Moving Picture Experts Group (“MPEG”) has adopted standards for streaming video. The MPEG-4 standard includes a mode for FGS video. In MPEG-4 FGS video, the current frame is predicted using the minimum-bitrate reconstructed version of the previous frame. With MPEG-4, if a higher-bitrate version of the previous frame were used for prediction, this would lead to prediction drift any time the bit stream was decoded at a rate lower than the rate used for prediction in the encoder. The prediction drift is caused by the difference between the encoder's reference frame and the decoder's reference frame. Accordingly, it is desirable to improve the CODEC efficiency over that of typical FGS schemes such as, for example, the FGS video scheme adopted in the MPEG-4 standard.

SUMMARY OF THE INVENTION

[0005] These and other drawbacks and disadvantages of the prior art are addressed by an apparatus and method for motion compensation of fine-grain scalable video data. Fine-grain scalable video data is generated by an encoder for encoding input video data as minimum bitrate macroblock data to produce Discrete Cosine Transform (“DCT”) data having DCT coefficients representing a minimum bitrate version of the macroblock data, and for encoding the input video data as intermediate bitrate macroblock data to produce DCT data having DCT coefficients representing an intermediate bitrate version of the macroblock data, and an adaptive motion compensator in signal communication with the encoder for predicting whether a decoded version of the intermediate bitrate macroblock data will have an accumulated predicted error frame energy exceeding a maximum threshold.

BRIEF DESCRIPTION OF THE DRAWINGS

[0006] The present disclosure teaches an efficient approach to motion compensation for fine-grain scalable video in accordance with the following exemplary figures, in which:

[0007] FIG. 1 is a block diagram of a fine-grain scalable (“FGS”) encoder with only base layer motion compensation;

[0008] FIG. 2 is a block diagram of a fine-grain scalable (“FGS”) encoder with adaptive motion compensation according to a preferred embodiment of the present invention; and

[0009] FIG. 3 is a flow diagram for the adaptive motion compensation of FIG. 2, in accordance with the principles of the present invention.

DETAILED DESCRIPTION

[0010] A video data coder-decoder (“CODEC”), in accordance with the embodiments of the present invention described herein, employs discrete cosine transform (“DCT”) based manipulation of video data. The video data is preferably organized as macroblocks.

[0011] MPEG-4 fine-grain scalability (“FGS”) uses a minimum-bitrate previous frame for motion compensation. In accordance with the principles of the invention, the encoder chooses between the minimum-bitrate previous frame and a higher-bitrate previous frame, on a macroblock basis. The encoder tracks the accumulated prediction drift at each frame.

[0012] For a given macroblock, if using the higher-bitrate previous frame for motion compensation would result in a prediction drift energy above a maximum limit, the encoder chooses the minimum-bitrate previous frame to predict that macroblock. Otherwise, the encoder chooses the higher-bitrate previous frame to predict the macroblock. The encoder sets a bit (flag) in the coded macroblock to convey to the decoder which version of the previous frame was used for the prediction.

[0013] As shown in FIG. 1, an FGS encoder 10 can be functionally broken up into a Base Layer portion 11 and an Enhancement Layer portion 33. The Base Layer portion 11 includes an input terminal 12 that is coupled in signal communication to a positive input of a summing block 14. The summing block 14 is coupled, in turn, to a function block 16 for implementing a discrete cosine transform (“DCT”). The block 16 is coupled to a function block 18 for implementing the quantization transform Q. The function block 18 is coupled to a function block 20 for implementing variable length coding (“VLC”). The block 18 is further coupled to a function block 22 for implementing the inverse quantization transform Q−1.

[0014] The block 22, in turn, is coupled to a function block 24 for implementing an inverse discrete cosine transform (“IDCT”). The block 24 is coupled to a positive input of a summing block 26, which is coupled to a block 28 for implementing a frame buffer. The block 28 is coupled to a function block 30 for performing motion estimation. The input terminal 12 is also coupled to the block 30 for providing an input video signal. The frame buffer 28 and the motion estimation block 30 are each coupled to a block 32 for performing motion compensation. The function block 32 is coupled to a negative input of the summing block 14 and also passed to a positive input of the summing block 26.

[0015] The enhancement layer portion 33 includes a summing block 34 having its positive input coupled to the output of the DCT 16, and its negative input coupled to the output of the inverse quantization block 22. The output of the block 34 is coupled to a function block 36 for implementing bit-plane coding. The output of the bit-plane coder 36 is coupled, in turn, to a function block 38 for implementing variable length coding (“VLC”).

[0016] In operation, the FGS encoder of FIG. 1 uses only the base layer for prediction, as is done in MPEG-4 FGS. The base layer encoder 11 is simply a single layer DCT-based motion-compensated encoder. Initially, input video is motion-compensated, using motion vectors obtained from the motion estimation process. Then the prediction error is transformed using the DCT, and the resulting DCT coefficients are quantized and entropy coded using a variable-length code. To reconstruct the base layer frame, initially, inverse quantization is performed, then an IDCT is performed. The prediction that was subtracted in the motion compensation process is then added back in, and the reconstructed frame is stored in the frame buffer in order to be used as a reference for future pictures.

[0017] An initial step in encoding the enhancement layer is to subtract the inverse quantized DCT coefficients in the base layer from the unquantized coefficients. The bit planes are then scanned one at a time and variable-length coded. The decoder will decode some subset of these bitplanes according to the bitrate available at the time of decoding.

[0018] Turning to FIG. 2, an exemplary FGS encoder 110, in accordance with the principles of the present invention, can be conceptually broken up into a Base Layer portion 111 and an Enhancement Layer portion 133. The Base Layer portion 111 includes an input terminal 112 that is coupled in signal communication to a positive input of a summing block 114. The summing block 114 is coupled, in turn, to a function block 116 for implementing a discrete cosine transform (“DCT”). The block 116 is coupled to a function block 118 for implementing the quantization transform Q. The function block 118 is coupled to a function block 120 for implementing variable length coding (“VLC”). The block 118 is further coupled to a function block 122 for implementing the inverse quantization transform Q−1.

[0019] The block 122, in turn, is coupled to a function block 124 for implementing an inverse discrete cosine transform (“IDCT”). The block 124 is coupled to a positive input of a summing block 126, which is coupled to a block 128 for implementing a frame buffer. The block 128 is coupled to a function block 130 for performing motion estimation. The input terminal 112 is also coupled to the block 130 for providing an input video signal. The frame buffer 128 and the motion estimator 130 are each coupled to a function block 132 for performing adaptive motion compensation. The function block 132 is coupled to a negative input of the summing block 114 and also passed to a positive input of the summing block 126.

[0020] The enhancement layer portion 133 includes a summing block 134 having its positive input coupled to the output of the DCT 116, and its negative input coupled to the output of the inverse quantization block 122. The output of the block 134 is coupled to a function block 136 for implementing bit-plane coding. The output of the bit-plane coder 136 is coupled, in turn, to a function block 138 for implementing variable length coding (“VLC”). The output of the bit-plane coder 136 is also coupled to a positive input of a summing block 139 comprised by the base layer portion 111.

[0021] Returning to the base layer portion 111, the summing block 139 has another positive input coupled from the output of the inverse quantization block 122. The output of the summing block 139 is coupled to a function block 140 for implementing another IDCT. The IDCT 140 is coupled to a positive input of a summing block 142, which has another positive input coupled from the output of the adaptive motion compensator 132. The output of the summing block 142 is coupled to an enhancement layer frame buffer 144. The enhancement layer frame buffer 144 is coupled, in turn, to the adaptive motion compensator 132. A drift frame buffer 146 is coupled in bi-directional signal communication with the adaptive motion compensator 132.

[0022] In operation, the FGS encoder of FIG. 2 implements a preferred FGS method, in accordance with the principles of the present invention. A significant difference between the FGS encoder of FIG. 2 and that of FIG. 1 is that in the encoder of FIG. 2, the output of the bit plane coding for a subset of the bit planes in the enhancement layer is added to the inverse-quantized DCT coefficients in the base layer, as an initial step in obtaining the reconstructed enhancement layer frame fmid. An IDCT is then performed and the prediction from the motion compensation step is added back in. The result, fmid, is stored in the enhancement layer frame buffer. The reconstructed base layer frame, fmin, is stored in the base layer frame buffer. In the adaptive motion compensation method, apparatus, and system which incorporates the principles of the present invention, the base layer and enhancement layer predictions are read, the accumulated prediction drift is computed assuming the enhancement layer prediction is used, and the appropriate prediction is selected. If the enhancement layer prediction is selected, the accumulated prediction drift is updated and written to the drift frame buffer.

[0023] Referring to FIG. 3, the process of adaptive motion compensation referred to in FIG. 2 and accompanying description, and in accordance with the principles of the present invention, is illustrated as method 200. Method 200 begins at start block 210, and proceeds to decision block 212. At decision block 212, it is determined whether the current image begins a new group of pictures (“GOP”). If the current image does begin a new GOP, control passes to function block 214 to reset the accumulated predicted error frame, Fd, to zero. After block 214, or if a new GOP was not detected at block 212, control passes to function block 216, which chooses an intermediate bitrate Rmid, where Rmid is any value between a minimum bitrate Rmin and a maximum bitrate Rmax. For exemplary purposes, Rmid may be considered to be halfway between Rmin and Rmax. Block 216 then passes to function block 218, which fetches a macroblock from the frame, Fmin, corresponding to the previous frame coded at the minimum bitrate Rmin. Block 218 then passes to function block 220, which fetches a macroblock from the frame, Fmid, corresponding to the previous frame coded at the intermediate bitrate Rmid. Block 220 then passes to function block 222, which fetches a macroblock from the frame, Fd, corresponding to the previous frame accumulated prediction error.

[0024] Function block 222 passes control to function block 226. Block 226 computes the energy E of the intermediate bitrate prediction Pmid relative to the accumulated prediction error Fd, and passes to decision block 228. Decision block 228 determines whether the computed energy E is greater than a threshold Emax, and if it is not greater, passes control to function block 230. Function block 230 chooses the intermediate bitrate prediction Pmid, and passes to function block 232. Function block 232 updates the accumulated prediction error frame Fd, and passes to a return block 236. At decision block 228, if the energy E is greater than the threshold Emax, then control is passed to function block 234. Block 234 chooses the minimum bitrate prediction Pmin, and passes to return block 236.

[0025] In operation of the present motion compensation method, the minimum and maximum bitrates for the encoded data stream are Rmin and Rmax respectively. Rmid is any intermediate bitrate between Rmin and Rmax. Thus, to encode a macroblock, the encoder fetches a motion-compensated block from the previous frame at Rmin and a motion-compensated block from the previous frame at Rmid.

[0026] The encoder also fetches another block from a frame representing the accumulated prediction drift error. The accumulated prediction drift error frame is reset to zero at the beginning of every group of pictures (“GOP”). The blocks representing the minimum-rate prediction, intermediate rate prediction, and accumulated prediction drift error are referred to as Pmin, Pmid, and Pd, respectively. In order to determine which prediction to use, the encoder computes the energy of the prediction drift error for the Pmid prediction. If the energy “E” is defined as a function measuring the energy of a block and if Emax is the maximum permitted drift energy threshold, then the appropriate prediction is selected as follows:

If E(Pd+Pmin−Pmid)>Emax  (1)

Prediction=Pmin

[0027] Else

Prediction=Pmid

P d =P d +P min −P mid

[0028] End If

[0029] In this exemplary embodiment, a bit is included in the macroblock header to convey to the receiving decoder which prediction block was selected. In the decoder, two decoded versions of each frame, Fmin and Fmid, respectively, are written into memory to be used as reference frames. The frame Fmin represents the frame at the minimum bitrate, while the frame Fmid represents the frame at the intermediate bitrate. If the frame is decoded at a bitrate lower than Rmid, then Fmid is approximated using the decoded frame at that lower bitrate.

[0030] These and other features and advantages of the present disclosure may be readily ascertained by one of ordinary skill in the pertinent art based on the teachings herein. It is to be understood that the teachings of the present disclosure may be implemented in various forms of hardware, software, firmware, special purpose processors, or combinations thereof.

[0031] Most preferably, the teachings of the present disclosure are implemented as a combination of hardware and software. Moreover, the software is preferably implemented as an application program tangibly embodied on a program storage unit. The application program may be uploaded to, and executed by, a machine comprising any suitable architecture. Preferably, the machine is implemented on a computer platform having hardware such as one or more central processing units (“CPU”), a random access memory (“RAM”), and input/output (“I/O”) interfaces. The computer platform may also include an operating system and microinstruction code. The various processes and functions described herein may be either part of the microinstruction code or part of the application program, or any combination thereof, which may be executed by a CPU. In addition, various other peripheral units may be connected to the computer platform such as an additional data storage unit and a printing unit.

[0032] It is to be further understood that, because some of the constituent system components and methods depicted in the accompanying drawings are preferably implemented in software, the actual connections between the system components or the process function blocks may differ depending upon the manner in which the present disclosure is programmed. Given the teachings herein, one of ordinary skill in the pertinent art will be able to contemplate these and similar implementations or configurations of the present disclosure.

[0033] Although the illustrative embodiments have been described herein With reference to the accompanying drawings, it is to be understood that the present disclosure is not limited to those precise embodiments, and that various changes and modifications may be effected therein by one of ordinary skill in the pertinent art without departing from the scope or spirit of the present disclosure. All such changes and modifications are intended to be included within the scope of the present disclosure as set forth in the appended claims.

Claims

1. A fine-grain scalable video data apparatus comprising: an encoder (110) for encoding input video data as minimum bitrate macroblock data to produce DCT data having DCT coefficients representing a minimum bitrate version of the macroblock data, and for encoding the input video data as intermediate bitrate macroblock data to produce DCT data having DCT coefficients representing an intermediate bitrate version of the macroblock data; and an adaptive motion compensator (132) in signal communication with the encoder for predicting whether a decoded version of the intermediate bitrate macroblock data has an accumulated predicted error frame energy exceeding a maximum threshold (228).

2. The apparatus as defined in claim 1 wherein the input video data is fine-grain scalable between the minimum bitrate and a maximum bitrate, the intermediate bitrate falling anywhere therebetween.

3. The apparatus as defined in claim 1 wherein the encoder comprises the adaptive motion compensator.

4. The apparatus as defined in claim 1, the encoder comprising: an enhancement layer frame buffer (144); the adaptive motion compensator (132) in signal communication with the enhancement layer frame buffer; and a drift frame buffer (146) in signal communication with the enhancement layer frame buffer.

5. The apparatus as defined in claim 1, the adaptive motion compensator comprising: a group-of-pictures detector for resetting a drift frame buffer for each new group-of-pictures (212); an energy unit (226) for computing the energy of an intermediate-rate prediction relative to the drift frame buffer (226); and a prediction unit for selecting one of the intermediate-rate prediction and a minimum-rate prediction for each block of pixels to be predicted from the data of the previous picture using the motion vectors for the macroblock data (228).

6. A fine-grain scalable video data apparatus for receiving encoded video macroblock data wherein each macroblock is represented by one of DCT coefficients representing a minimum bitrate version of the macroblock data and DCT coefficients representing an intermediate bitrate version of the macroblock data, the apparatus comprising a decoder for decoding one of the intermediate and minimum bitrate encoded DCT data for each macroblock received from the encoder to produce reconstructed macroblock data responsive to a predicted energy of an accumulated predicted error frame.

7. A method for performing fine-grain scalable video data operations, the method comprising: encoding input video data as minimum bitrate macroblock data to produce DCT data having DCT coefficients representing a minimum bitrate version of the macroblock data; encoding the input video data as intermediate bitrate macroblock data to produce DCT data having DCT coefficients representing an intermediate bitrate version of the macroblock data; and compensating the encoded data to predict whether a decoded version of the intermediate bitrate macroblock data has an accumulated predicted error frame energy exceeding a maximum threshold.

8. A method as defined in claim 7, further comprising decoding one of the intermediate and minimum bitrate encoded DCT data from the encoded input video data to produce reconstructed macroblock data responsive to the predicted energy of the accumulated predicted error frame.

9. A method as defined in claim 7 wherein the input video data is fine-grain scalable between the minimum bitrate and a maximum bitrate, the intermediate bitrate falling anywhere therebetween.

10. A method as defined in claim 7, further comprising compensating enhancement layer block data with enhancement layer data of a previous picture and motion vectors for the macroblock data to produce compensated enhancement layer block data such that DCT is performed with respect to the compensated enhancement layer block data to produce the enhancement layer DCT data.

11. A method for receiving encoded video macroblock data wherein each macroblock is represented by one of DCT coefficients representing a minimum bitrate version of the macroblock data and DCT coefficients representing an intermediate bitrate version of the macroblock data, the method comprising decoding one of the intermediate and minimum bitrate encoded DCT data for each macroblock received from the encoder to produce reconstructed macroblock data responsive to a predicted energy of an accumulated predicted error frame.

12. A program storage device readable by machine, tangibly embodying a program of instructions executable by the machine to perform steps for performing fine-grain scalable video data operations, the steps comprising: encoding input video data as minimum bitrate macroblock data to produce DCT data having DCT coefficients representing a minimum bitrate version of the macroblock data; encoding the input video data as intermediate bitrate macroblock data to produce DCT data having DCT coefficients representing an intermediate bitrate version of the macroblock data; and compensating the encoded data to predict whether a decoded version of the intermediate bitrate macroblock data will have an accumulated predicted error frame energy exceeding a maximum threshold.

13. A program storage device as defined in claim 12, the steps further comprising decoding one of the intermediate and minimum bitrate encoded DCT data from the encoded input video data to produce reconstructed macroblock data responsive to the predicted energy of the accumulated predicted error frame.

14. A program storage device as defined in claim 12 wherein the input video data is fine-grain scalable between the minimum bitrate and a maximum bitrate, the intermediate bitrate falling anywhere therebetween.

15. A program storage device as defined in claim 12, the steps further comprising compensating enhancement layer block data with enhancement layer data of a previous picture and motion vectors for the macroblock data to produce compensated enhancement layer block data such that DCT is performed with respect to the compensated enhancement layer block data to produce the enhancement layer DCT data.

16. A program storage device readable by machine, tangibly embodying a program of instructions executable by the machine to perform steps for receiving encoded video macroblock data wherein each macroblock is represented by one of DCT coefficients representing a minimum bitrate version of the macroblock data and DCT coefficients representing an intermediate bitrate version of the macroblock data, the steps comprising decoding one of the intermediate and minimum bitrate encoded DCT data for each macroblock received from the encoder to produce reconstructed macroblock data responsive to a predicted energy of an accumulated predicted error frame.

17. A fine-grain scalable video data system for performing fine-grain scalable video data operations, the system comprising: minimum encoding means for encoding input video data as minimum bitrate macroblock data to produce DCT data having DCT coefficients representing a minimum bitrate version of the macroblock data; intermediate encoding means for encoding the input video data as intermediate bitrate macroblock data to produce DCT data having DCT coefficients representing an intermediate bitrate version of the macroblock data; and compensating means for compensating the encoded data to predict whether a decoded version of the intermediate bitrate macroblock data will have an accumulated predicted error frame energy exceeding a maximum threshold.

18. A system as defined in claim 17, further comprising decoding means for decoding one of the intermediate and minimum bitrate encoded DCT data from the encoded input video data to produce reconstructed macroblock data responsive to the predicted energy of the accumulated predicted error frame.

19. A system as defined in claim 17 wherein the input video data is fine-grain scalable between the minimum bitrate and a maximum bitrate, the intermediate bitrate falling anywhere therebetween.

20. A system as defined in claim 17, further comprising compensating means for compensating enhancement layer block data with enhancement layer data of a previous picture and motion vectors for the macroblock data to produce compensated enhancement layer block data such that DCT is performed with respect to the compensated enhancement layer block data to produce the enhancement layer DCT data.