整数小波框架下基于阈值分割的静止图像编码  

Still Image Coding Based on Threshold Segmentation Using Integer Wavelet Transform

作　　者：张立保[1] 王丽荣[1] 

机构地区：[1]吉林大学通信工程学院,吉林长春130025

出　　处：《光电子．激光》2004年第2期212-215,220,共5页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金重点资助项目(59638220)

摘　　要：首先,选定分割阈值Th。然后,对大于Th的小波系数采用整数平方阈值进行量化、缩短阈值间的距离,并对阈值平面上的系数采用改进的二进制SPECK编码框架,对小于Th的系数采用2的整数幂作为量化阈值。最后,在每个阈值平面上均采用无乘法的二进制算术编码。通过与嵌入式零树小波(EZW)、SPIHT及SPECK算法的实验结果比较,ETSC算法不仅有较低的计算复杂度,而且提高了整数小波变换(IWT)在低比特率下的编码效率。此外,该算法支持有损和无损解码在单一码流下完成。The coefficients above segmentation threshold Th were quantized by integer square threshold for shortening the distance between threshold and threshold. An improved binary SPECK coding framework was adopted for these coefficients above threshold Th. The coefficients below Th were quantized by integer powers of two. Finally, binary arithmetic coding without multiplication was adopted for every threshold plane. Compared with embedded zerotree wavelet (EZW), SPIHT and SPECK, the experiment results show that ETSC algorithm has low computational complexity, and increases the encoding efficiency of IWT at low bit rates. Additionally, the new algorithm supports both lossless and lossy decoding using a single bitstream.

关 键 词：整数小波变换 阈值分割 静止图像编码 整数平方阈值 二进制算术编码 

分 类 号：TN919.81[电子电信—通信与信息系统]