On Optimality of Context Modeling for Bit-Plane Entropy Coding in the JPEG2000 Standard

摘要: The JPEG2000 image compression standard exploits two empirically designed context models for conditional entropy coding of bit-planes wavelet transform coefficients. use the so called significance (binary) values eight adjacent coefficients as a template, which are mapped into 9 contexts. This paper addresses problem optimality this approach. In other words, we answer question: given is it possible to design that would result in better performance? our work, exploited optimization techniques model design. We show chosen results only marginal improvement performance (about 0.3%) compared class natural images. Our conclusion efficiency can be improved by choosing larger template.