Abstract: A quantizer employs a scaled integral inverse ratio division for quantization of an input T by a quantization step Q. The quantizer forms an integral approximation q of 2r/Q by either trunc(2r/Q) or round(2r/Q). A multiplier multiplies the absolute value of T by the q. An adjustment factor is added alternatively to the absolute value of T prior to multiplication or to the product after multiplication. This adjustment factor minimizes errors near transition points in the quantization. This invention is applicable to both trunc(T/Q) and round(T/Q).

Abstract: Motion compensation as for video compression with motion estimation with either frame mode or field mode by initial prediction of motion mode and when prediction error is small enough truncate motion estimation search.

Abstract: This invention is a method for speeding up block matching based motion estimation for video encoder. The invention 1) calculates statistics for a candidate motion vector for a predetermined fraction of the pixels of a macroblock, 2) makes an early decision based on this preliminary cost function, and 3) terminates the block matching process without calculating the cost function for other pixels if the preliminary cost function is not less than a predetermined threshold. This early decision for goodness estimation provides an economy of processing load when a large portion of data is left untouched (i.e. unprocessed). The present invention employs feedback control to reduce the predetermined threshold for quick convergence upon each detection of a better match.

Abstract: This invention corrects chrominance misalignment that occurs during chrominance down-sampling and up-sampling. The invention extracts a binary index from the corresponding luminance signal. The binary index enables generation of a filter window. On down-sampling the filter window is applied to a block of source chrominance pixels which are filtered or not based upon the binary index. On up-sampling the binary index of the filter window for the target chrominance pixels determines which are filtered or not.

Abstract: A quantizer employs a scaled integral inverse ratio division for quantization of an input T by a quantization step Q. The quantizer forms an integral approximation q of 2r/Q by either trunc(2r/Q) or round(2r/Q). A multiplier multiplies the absolute value of T by the q. An adjustment factor is added alternatively to the absolute value of T prior to multiplication or to the product after multiplication. This adjustment factor minimizes errors near transition points in the quantization. This invention is applicable to both trunc(T/Q) and round(T/Q).

Abstract: Motion compensation as for video compression with motion estimation with either frame mode or field mode by initial prediction of motion mode and when prediction error is small enough truncate motion estimation search.

Abstract: A method of variable length coding classifies each received symbol into one of a plurality of classifications having a corresponding variable length code table selected based upon a probability distribution of received symbols within the classification. The variable length codeword output corresponds to the received symbol according to the variable length code table corresponding to the classification of that received symbol. The plurality of classifications and the corresponding variable length code tables may be predetermined and fixed. Alternatively, the variable length code table may be dynamically determined with data transmitted from encoder to decoder specifying the variable length code tables and their configurations. Universal variable length code (UVLC) is used to code the symbols. Universal variable length code can instantiate to different variable length code tables with different parameters.

Abstract: This invention is a method to automatically configure the universal variable length coding (UVLC). The method is applicable to code the syntax elements with a lot of symbols like the transform coefficients and motion vectors. The configuration problem includes a configuration method based on a genetic algorithm (GA). The method can be applied to on-the-fly configuration of codewords during video encoding, or to off-line training of code tables. This invention also includes several techniques to reduce the required operations for applications which have only limited processing power.

Abstract: This invention is a method to automatically configure the universal variable length coding (UVLC). The method is applicable to code the syntax elements with a lot of symbols like the transform coefficients and motion vectors. The configuration problem includes a configuration method based on a genetic algorithm (GA). The method can be applied to on-the-fly configuration of codewords during video encoding, or to off-line training of code tables. This invention also includes several techniques to reduce the required operations for applications which have only limited processing power.