Title :
Source coding by classification-91-715
Author_Institution :
Philips Kommunikations Ind. AG, Nurnberg, Germany
fDate :
11/1/1995 12:00:00 AM
Abstract :
Adaptive source coding is investigated which is performed by a blockwise classification of the source symbols and selection of a coder depending on the class. As justification for adaptive coding, an upper bound is used for the rate distortion function of a stationary source depending on blockwise classification. The main result of the paper states that classification is appropriate to exploit statistical dependencies other than correlation. Such statistical dependencies are described for the prediction error of a coding scheme for moving pictures at 64 kbit/s and are exploited for upper bounding the rate distortion function of the prediction error and improving the coding scheme
Keywords :
adaptive codes; error analysis; image classification; prediction theory; rate distortion theory; source coding; statistical analysis; video coding; 64 kbit/s; 91-715; adaptive source coding; blockwise classification; classification; coder; coding scheme; moving pictures; prediction error; rate distortion function; source symbols; stationary source; statistical dependencies; upper bound; Adaptive coding; Image communication; Image reconstruction; Quantization; Rate-distortion; Resource description framework; Source coding; Speech coding; Switches; Upper bound;
Journal_Title :
Communications, IEEE Transactions on