Abstract: Disclosed is a method for constructing Slepian-Wolf codes, wherein the designed Slepian-Wolf codes are robust to mismatched decoding. The disclosed method for constructing Slepian-Wolf codes includes the steps of: choosing representative probability distributions from a set of possible probability distributions; choosing a probability distribution as a decoding metric; converting the chosen decoding metric to a cyclic-symmetric channel; computing the initial message value given the cyclic-symmetric channel; computing a set of probability distributions of the initial message given the initial message values and the representative probability distributions; optimizing the degree distribution given the set of probability distributions of the initial message; optimizing the decoding metric.

Abstract: A method of and system for rateless source coding are disclosed. The method comprises the steps of providing a set of low-density parity check (LDPC) codes, each of which accepts a range of data input lengths and a range of target compression rates; identifying a data input having a data input length; and identifying a desired compression rate. The method comprises the further steps of selecting one of said LDPC codes based on said data input length and desired compression rate; encoding the data input, using the selected LDPC code, to generate a sequence of data values; and puncturing some of said encoded data values to achieve the desired compression rate. Preferably, the encoding step includes the steps of generating a syndrome and a parity sequence from the data input, puncturing the generated parity sequence, and mixing a remaining portion of the data input with the punctuated parity sequence.

Abstract: Disclosed are a method and system for optimized, dynamic data-dependent program execution. The disclosed system comprises a statistics computer which computes statistics of the incoming data at the current time instant, where the said statistics include the probability distribution of the incoming data, the probability distribution over program modules induced by the incoming data, the probability distribution induced over program outputs by the incoming data, and the time-complexity of each program module for the incoming data, wherein the said statistics are computed on as a function of current and past data, and previously computed statistics; a plurality of alternative execution path orders designed prior to run-time by the use of an appropriate source code; a source code selector which selects one of the execution path orders as a function of the statistics computed by the statistics computer; a complexity measurement which measures the time-complexity of the currently selected execution path-order.

Abstract: A system for the introduction of controlled correlation among multiple redundant representations of predictively encoded signals while avoiding predictive mismatch at a receiver when any given sub-set of the multiple representations is received. The system embodies a signal encoder and decoder. The decoder can comprise at least two signal adders for respectively receiving coefficient values and adding at least one predictive value transform to the coefficient value in order to generate and transmit a second set of coefficient values. The second set of coefficient values is subsequently received by a decoder means, wherein the decoder means transforms the received coefficient values and transmits the resultant coefficient values to a signal adder. Upon reception of the transformed coefficient values, the signal adder generates a third set of coefficient values; the third set of coefficients being used to reconstruct an approximate version of an encoded signal.

Abstract: A computer-implemented method for decompression in data compression systems with decpder side-information including a plurality of signals each of which is correlated to a source, includes determining a conditional probability function of the source conditioned upon a subset of decoder side-information signals, wherein the decoder side-information signals include pre-stored and received statistical information, estimating an a-posteriori probability function based on the conditional probability function and extrinsic information, evaluating a stopping criterion for decompresiion, generating the extrinsic information based on the a-posteriori probability function, and determining a likelihood threshold for determining a most probable value of a quantized source signal based on the a-posteriori probability function and outputting the quantized source upon determining to stop decompression.

Abstract: A method and apparatus for decompressing data in a data-compression system with decoder-only side information is provided. In one aspect, the method comprises generating side information using a source reconstruction and decoding using the generated side information to generate a new source reconstruction. The method further includes iterating the steps of generating and decoding, the generating step using at least the new source reconstruction output by the previous decoding step, and the decoding step using the side information output by the previous generating step. The method may stop the iteration when one or more predetermined criteria are met.

Abstract: A system for the introduction of controlled correlation among multiple redundant representations of predictively encoded signals while avoiding predictive mismatch at a receiver when any given sub-set of the multiple representations is received. The system embodies a signal encoder and decoder. The decoder can comprise at least two signal adders for respectively receiving coefficient values and adding at least one predictive value transform to the coefficient value in order to generate and transmit a second set of coefficient values. The second set of coefficient values is subsequently received by a decoder means, wherein the decoder means transforms the received coefficient values and transmits the resultant coefficient values to a signal adder. Upon reception of the transformed coefficient values, the signal adder generates a third set of coefficient values; the third set of coefficients being used to reconstruct an approximate version of an encoded signal.

Abstract: The present invention relates to a method, system and computer program product for the predictive encoding of digital video sequences. The objectives of the invention are accomplished by dynamically determining the resolution of a current frame being encoded and outputting the determination. The determination process is based on statistical and coding information of a plurality of frames, including at least one previous frame and the current frame. Further, general encoding parameters and the encoding parameters of a current frame at a chosen resolution are determined, wherein the encoding parameter selection step takes into account the determination of the dynamic resolution determination step in determining the encoding parameters.

Abstract: The present invention relates to a method, system and computer program product for the predictive encoding of digital video sequences. The objectives of the invention are accomplished by dynamically determining the resolution of a current frame being encoded and outputting the determination. The determination process is based on statistical and coding information of a plurality of frames, including at least one previous frame and the current frame. Further, general encoding parameters and the encoding parameters of a current frame at a chosen resolution are determined, wherein the encoding parameter selection step takes into account the determination of the dynamic resolution determination step in determining the encoding parameters.

Abstract: A system for the introduction of controlled correlation among multiple redundant representations of predictively encoded signals while avoiding predictive mismatch at a receiver when any given sub-set of the multiple representations is received. The system embodies a signal encoder and decoder. The decoder can comprise at least two signal adders for respectively receiving coefficient values and adding at least one predictive value transform to the coefficient value in order to generate and transmit a second set of coefficient values. The second set of coefficient values is subsequently received by a decoder means, wherein the decoder means transforms the received coefficient values and transmits the resultant coefficient values to a signal adder. Upon reception of the transformed coefficient values, the signal adder generates a third set of coefficient values; the third set of coefficients being used to reconstruct an approximate version of an encoded signal.

Abstract: A method and apparatus are described for constructing efficient Slepian-Wolf codes for inputs that do not have uniform probability distributions, and for channels which are not symmetric. Whereas conventional Wyner-Ziv video compression systems assume symmetrical probability distributions and ignore the marginal distribution component, the invention incorporates the marginal distribution component by combining the marginal distribution and conditional distributions as input to a transform that generates a symmetrical output, which is then fed into the conventional density evolution method.

Abstract: An improvement to a predictive video encoding method or apparatus that includes selecting a mode class within which to choose an encoding algorithm to encode individual blocks of video information. Rather than implementing each algorithm in each mode class to ascertain an acceptable compression, the improvement eliminates searching through a class of encoding modes based simply on heuristics. The method comprises obtaining statistical information related to previous blocks of encoded video information, determining a mode class within which to chose a particular encoding algorithm based on the statistical information (e.g., heuristics) gathered, choosing an algorithm within the selected mode class using conventional techniques, and encoding the video information according to the chosen algorithm. Statistical information may include quantization parameters, prior encoding decisions, intensity or frequency values, or Hadamard transform coefficients of previously encoded macroblocks.

Abstract: A system for the introduction of controlled correlation among multiple redundant representations of predictively encoded signals while avoiding predictive mismatch at a receiver when any given sub-set of the multiple representations is received. The system embodies a signal encoder and decoder. The decoder can comprise at least two signal adders for respectively receiving coefficient values and adding at least one predictive value transform to the coefficient value in order to generate and transmit a second set of coefficient values. The second set of coefficient values is subsequently received by a decoder means, wherein the decoder means transforms the received coefficient values and transmits the resultant coefficient values to a signal adder. Upon reception of the transformed coefficient values, the signal adder generates a third set of coefficient values; the third set of coefficients being used to reconstruct an approximate version of an encoded signal.

Abstract: The present invention relates to a method, system and computer program product for the predictive encoding of digital video sequences. The objectives of the invention are accomplished by dynamically determining the resolution of a current frame being encoded and outputting the determination. The determination process is based on statistical and coding information of a plurality of frames, including at least one previous frame and the current frame. Further, general encoding parameters and the encoding parameters of a current frame at a chosen resolution are determined, wherein the encoding parameter selection step takes into account the determination of the dynamic resolution determination step in determining the encoding parameters.