Prediction Truncation Algorithm Based on Regression Analysis for Embedded Block Coding

Abstract

In order to solve the problems of execution complexity and time waste in compression process induced by Post Compression Rate Distortion(PCRD) Optimization in embedded block coding in JPEG2000, this paper proposes an effective method to promote coding efficiency-embedded block coding algorithm with prediction truncation based on regression analysis. This algorithm predicts the truncation points based on regression formula derived from linear regression analysis. It avoids scanning and encoding the bit planes that can be truncated, discards the time-consuming rate-distortion truncation processing, so that the algorithm is able to speed up the compression process. The experiment results show the algorithm obtains obvious promotion efficiency of coding speed. high compression ratio such as 0.125 or 0.0625, the majority of the coefficients which have been encoded in tier-2 stage are discarded through truncation. Obviously, a lot of running time is wasted in the coding of coefficient bits that will be truncated. This paper proposes the predict truncation algorithm based on regression analysis, while embedded coding it can predict plane scan code stream position of the truncated points, give up coding to the coefficients bits that will be discarded, and give up the rate distortion calculation and optimization truncation based on cycle search in Tier-2 coding, thus greatly improves the encoding speed.