The Research On JPEG XR Rate Contorl Algorithm

JPEG XR is a new still-image compression algorithm and file format forcontinuous tone photographic image, which is developed by Microsoft. As the nextgeneration image coding standard, JPEG XR can outperform the original JPEG standardin terms of compression efficiency in many situations, and obtain comparablecompression efficiency to JPEG2000, but with much lower computational complexity.The overall goal of JPEG XR is to provide a compressed format specificationappropriate for a wide range of applications while keeping the implementationrequirements for encoders and decoders as simple as possible. Rate control is a keytechnology in JPEG XR encoding process like JPEG for CBR application. Thequantization scheme of JPEG XR is so flexible that it can be tuned to optimize codingperformance. But the organization of JPEG hasn’t released any algorithms of ratecontrol with optimization processing. It’s indeed extremely meaningful to researchJPEG XR rate control algorithm.This paper first introduces JPEG XR from its file format and its encoder. For JPEGXR rate control research, we study distortion rate theory and some approaches of ratecontrol, and use these to direct our research for designing appropriate rate controlalgorithms.Then, this paper introduces some rate control algorithms about JPEG. The iterativealgorithm is a novel rate control technique. The main feature of this algorithm is that itcan get a good rate-distortion performance, but having a very high arithmeticcomplexity. By researching JPEG XR from its image structure and encoder, we candesign a modified iterative algorithm for JPEG XR. This algorithm can compress fileaccurately, but its rate-distortion performance is unsatisfactory.Finally, for improving JPEG XR rate control algorithm’s performance, this paperresearches image complexity description by analyzing image internal structure, anddistributes bit rates by image complexity. The rate-distortion performance of modifiediterative algorithm can be improved as considering image complexity. In most cases,image has own R-D relationship. We research a new R-D model of JPEG XR image byits coefficient distribution characteristics, and use this exponential R-D model andimage complexity to propose a new rate control algorithm based exponential R-D model.The algorithm can compress image accurately and have a good rate-distortion performance with low arithmetic complexity.The experimental result shows the iterative algorithm considered image complexityoffers rate-distortion gains of around1dB in PSNR over iterative algorithm, and therate control algorithm based exponential R-D model performs better than iterativealgorithm, with a gain in PSNR around of3dB.