Method and apparatus for accomplishing multiple description coding for video
View Patent ↗A method and apparatus for utilizing temporal prediction and motion compensated prediction to accomplish multiple description video coding is disclosed. An encoder receives a sequence of video frames and divides each frame into non-overlapping macromacroblocks. Each macromacroblock is then encoded using either an intraframe mode (I-mode) or a prediction mode (P-mode) technique. Both the I-mode and the P-mode encoding techniques produce an output for each of n channels used to transmit the encoded video data to a decoder.
1. An encoder comprising:
a first module responsive to an applied video frame that divides the video frame into non-overlapping macroblocks,
an intra-frame coding (I-mode) module that develops multiple descriptions of an applied macroblock;
a predictive coding (P-mode) module that develops multiple predictive descriptions of said applied macroblock;
a mode selection element that applies each of the macroblocks created by said first module to either said I-mode module or to said P-mode module;
a buffer, connected to said mode selector element, for storing macroblocks that were reconstructed by decoding macroblocks of previously applied video frames, which had been encoded, and
a rate control unit that is connected to said mode selection element, said I-mode module and said P-mode module, for influencing whether said mode selection element applies the macroblocks created by said first module to said I-mode module or to said P-mode module.
2. The encoder of claim 1 where said buffer stored reconstructed macroblocks from a plurality of previously applied frames.
3. The encoder of claim 1 where said P-mode module encodes error between the applied macroblock and a macroblock from a previous frame that is used to predict said applied macroblock (best matching macroblock).
4. The encoder of claim 3 where said P-mode module also encodes a motion vector that describes displacement between said applied macroblock and said best matching macroblock.
5. The encoder of claim 4 where both the encoded error and the encoded motion vector are coded into said multiple descriptions.
6. The encoder of claim 1 where said mode selection element applies a macroblock of said created macroblocks to said I-mode module periodically.
7. The encoder of claim 6 where said mode selection element applies a macroblock of said created macroblocks to said I-mode module every x macroblocks that are routed to said P-mode module, where x is a number between 10 and 15.
8. The encoder of claim 1 where said rate control unit influences choice by said mode selection element as to whether to apply a macroblock to said I-mode module or to said P-mode module.
9. The encoder of claim 1 further comprising a redundancy allocation module, responsive to said I-mode module and to said P-mode module, for affecting the applying performed of said mode selection element based on desired redundancy.
10. The encoder of claim 1 where said buffer stores macroblocks of video frames that were reconstructed by using a first of said multiple descriptions, ψ 1 , macroblocks of said video frames that were reconstructed by using a second of said multiple descriptions, ψ 2 , and macroblocks of said video frames that were reconstructed by using all of said multiple descriptions, ψ 0 .
11. The encoder of claim 10 where said P-mode module comprises
a motion estimation coding element that, based on said ψ 1 , ψ 2 , and ψ 0 macroblocks and on said applied macroblock, develops multiple description coded motion vectors for said applied macroblock;
a first motion compensated predictor responsive to a first of said multiple description coded motion vectors and to said ψ 1 macroblocks for developing code P 1 ,
a second motion compensated predictor responsive to a second of said multiple description coded motion vectors and to said ψ 2 macroblocks for developing code P 2 ,
a third motion compensated predictor responsive to all of said multiple description coded motion vectors and to said ψ 1 , ψ 2 , and ψ 0 macroblocks for developing code P 0 ; and
a multiple description encoder responsive to said P 1 , P 2 and P 0 for developing multiple prediction error descriptions.