基于自适应整数小波变换的侦察图像编码  

Reconnaissance image encoding based on adaptive integer-to-integer wavelet transform

作　　者：胡高军[1] 吴芝路[1] 高真真[1] 

机构地区：[1]哈尔滨工业大学电子与信息技术研究院,黑龙江哈尔滨150001

出　　处：《光电工程》2006年第10期132-136,共5页Opto-Electronic Engineering

摘　　要：由于侦察图像相关性低,基于传统小波变换的图像编码在渐进恢复时低频近似图像边缘模糊。针对这个问题,提出了基于自适应小波变换结合EBCOT的编码方法。该方法通过区别对待平滑区域和边缘部分,降低边缘被平滑的程度,保持边缘清晰。为了达到无损编码,还推导了整数到整数的自适应小波变换。实验表明,该方法达到了图像的无损编码,并在压缩比基本不变的情况下,图像的低频近似效果得以改善。In view of the blurriness of the edge in the approximation image based on traditional wavelet transform and the lower correlation of reconnaissance image than general image, an encoding method based on adaptive wavelet transform combined with Embedded Block Coding with Optimized Truncation （EBCOT） is proposed. The method deals respectively with the smooth area and keen-edged area by different means, so that the edge is not smoothened and kept clear. And an integer-to-integer adaptive wavelet transform is derived to implement the lossless compression. Experiments show that the perfect reconstruction is completed, and the edge of the approximation image is clearer than non-adaptive one while compression rate is almost same.

关 键 词：图像相关性 自适应小波变换 图像编码 EBCOT 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]