JPEG2000压缩算法研究

由于JPEG2000采用零树编码方式实现压缩,编码过程中零树的形成个数与压缩效率CE密切相关;对小波滤波器及JPEG2000零树熵编码进行研究,发现高的小波变换层数和5／3滤波器的使用有利于增加零树个数,从而提高压缩率;实现基于拼接块小波变换的JPEG2000压缩算法,结果显示图像分成拼接块会影响编码效率和解码保真度;对JPEG和JPEG2000进行比较,得出在超低码率下,JPEG2000解码图像虽然没有采用JPEG时出现的“方块”效应,但有“飞萤”效应。

Study of Wavelet -based JPEG2000 Image Compression Algorithm

Using the Embedded Zemtree Wavelet (EZW), wavelet-baseded JPEG2000 standard realizes high image compression rate. In course of enceding, the selection of the wavelet filter, WT levels and tiles size are extremely critical. They have impact on the numbers of zerotrees. As a result, they have impact on the compression efficiency (CE). In this work, after studying EZW algorithm, we reach five important conclusions. One is that the numbers of zerotrees has a plus impact on CE. Secondly, zerotree count increased as the number of WT levels were increased. Thirdly, the 5/3 filter, on the average, performed slightly best than others. Fourthly, tiles size has a minus impact on CE. Fifthly, DCT-based JPEG algorithm produces blocking effect at a very low rate, but JPEG2000 produces "fly effect".