Research On Image Compression Algorithm Based On JPEG2000

JPEG2000 is the new image compression standard. It supports good lossless compression, good compression of low bit-depth imagery, scalability, and ROI coding and so on. It has a great variety of applications such as Internet, digital cameras, wireless communication, medical imaging. But the compression system of JPEG2000 are more complex and needs long coding time which is not fit for the high real-time system. Therefore, we aim to find a solution to speed up coding.In this thesis several problem are discussed: Firstly, the deep analysis of two main techniques DWT and compression algorithm EBCOT (Embedded Block Coding With Optimal Truncation) in JPEG2000 is presented. EBCOT is composed of block coding and bit-stream organization. Block coding determines the compression performance, but bit-stream organization adds extra information bits. Accordingly, we focus on block coding. We discover that the three coding passes of block coding have different need-to-be-coded pixels and there are many redundancies in the three passes scanning. Secondly, based on this feature and analyzing the image transform coefficients, we propose an improved algorithm coding the highest and the sencond highest bit plane together and two coding passes on the 0 bit plane. The coding pass union algorithm not only can speed up coding through reducing scanning times but also gives a way to improve lossess compression. At last, we introduce the core coding system of JPEG2000 and design the experiment system in c/c++ language. Experiment shows that the improved algorithm can reduce the coding time with not making compression performance worse and gaining slightly better lossless performance for a great number of images. The improved algorithm is especially suitable for the real-time transmission image compression system.