OPTIMAL HEEGARD-BERGER CODING SCHEMES
Optimal Heegard-Berger coding methods, devices, and systems are provided based on the disclosed coding schemes. The disclosed schemes facilitate decoding even in the absence of side information, with lower coding complexity than conventional Wyner-Ziv based distributed coding techniques. The disclosed details enable various refinements and modifications according to system design considerations.
1 . A decoding method comprising:
receiving in a first decoder a succession of encoded source information including last and current information from an encoder;
reconstructing a sequence of coarse version information including last and current coarse version information in the first decoder;
determining whether side information is absent or substantially unreliable;
transmitting the current coarse version information if side information is absent or substantially unreliable; otherwise
performing a motion search based at least in part on the last coarse version information;
reconstructing in a second decoder for the current version information from the encoder, a fine version information based at least in part on the motion search, the second decoder having a spatial domain; and
transmitting the fine version information.
2 . The method of claim 1 , wherein the fine version information reconstruction is performed in the spatial domain.
3 . The method of claim 1 , further comprising:
receiving in the second decoder a coded quantization noise variance from the encoder to facilitate the fine version information reconstruction.
4 . The method of claim 3 , wherein the quantization noise variance from the encoder is coded using fixed length coding.
5 . The method of claim 3 , wherein the encoder includes a lossy coder communicatively coupled to a quantization noise variance estimator and a transform module having a transform domain.
6 . The method of claim 5 , further comprising performing linear minimum mean square error computation in the estimator to facilitate quantization noise variance estimation.
7 . The method of claim 5 , further comprising:
performing quantization and entropy coding in the lossy coder to facilitate producing the succession of encoded source information.
8 . The method of claim 5 , further comprising:
storing in the transform domain, data associated with the last encoded source information to facilitate determining a prediction of the current encoded source information in the encoder.
9 . The method of claim 7 , wherein the entropy coding is one of Huffman coding, arithmetic coding, and static coding.
10 . The method of claim 2 , wherein the encoder includes successive refinement encoding based at least in part on second decoder side information communicatively coupled to the encoder.
11 . A decoding system comprising:
a first decoder component configured to receive coded source data from an encoder component including current and prior coded data segments and produce coarse version decoded data including a current and prior coarse version data segments;
a motion prediction component configured to provide a motion prediction based at least in part on a prior coarse version data segment;
a second decoder component configured to produce fine version decoded data based at least in part on the motion prediction in the spatial domain; and
a transmission component configured to transmit either the fine or the coarse version decoded data, respectively depending on whether side information is available and reliable or otherwise.
12 . The system of claim 11 , wherein the second decoder component is further configured to receive a fixed length coded quantization noise variance from the encoder component to facilitate producing the fine version decoded data.
13 . The system of claim 12 , wherein the encoder component comprises a lossy coding component communicatively coupled to a quantization noise variance estimation component and a transform component having a transform domain to facilitate producing the coded source data.
14 . The system of claim 13 , wherein the quantization noise variance estimation component is configured to perform linear minimum mean square error computation to facilitate quantization noise variance estimation.
15 . The system of claim 13 , wherein the lossy coding component further comprises a quantization component and an entropy coding component, the lossy coding component is further configured to store information associated with the prior coded data segments in the transform domain to facilitate determining a prediction of the information associated with the current coded data segment.
16 . In a video coding system, a coding method comprising:
encoding source data to form encoded source data to facilitate first-pass decoding by a first decoder, wherein the encoded source data includes a prior-pass and a current-pass encoded source data, and wherein the first-pass decoding produces first-pass decoded data for the prior-pass and the current-pass encoded source data; and
encoding quantization noise variance in an estimator module to form a coded quantization noise variance to facilitate second-pass decoding by a second decoder based on second decoder side information, wherein the second decoder side information is based in part on the first-pass decoded data for prior-pass encoded source data.
17 . The method of claim 16 , wherein the encoding in the estimator module includes performing linear minimum mean square error determination to facilitate encoding the quantization noise variance.
18 . The method of claim 16 , wherein the encoding source data includes encoding in a lossy coder communicatively coupled to the estimator module and storing in a transform domain of the lossy coder data associated with the prior-pass encoded source data to facilitate determining a prediction of data associated with the current-pass encoded source data.
19 . The method of claim 18 , further comprising:
performing entropy coding after quantization in the lossy coder to facilitate first-pass decoding by the first decoder.
20 . The method of claim 16 , wherein the encoding source data includes successive refinement encoding based at least in part on the second decoder side information received from the second pass decoder.