基于视觉注意机制的医学图像压缩研究  

Research of Medical Image Compression Based on Visual Attention Mechanism

作　　者：徐瑞[1] 帅仁俊[1] 陈平 

机构地区：[1]南京工业大学计算机科学与技术系 [2]南京市信息卫生中心

出　　处：《中国数字医学》2013年第7期74-77,86,共5页China Digital Medicine

基　　金：十一五国家科技支撑计划项目(编号:2010BAI88B00)~~

摘　　要：为了解决医学图像压缩的图像高质量和高压缩比的矛盾问题,基于视觉注意机制的理论,提出了一种近无损压缩和有损压缩相结合的医学图像压缩方法:对具有医学诊断价值的病灶区域进行低压缩比的近无损压缩,而对没有诊断价值的非病灶区域进行高压缩比的有损压缩。实验结果表明,该方法有效地实现了对病灶区域和非病灶区域的分别压缩,相对于无损压缩不仅提高了图像整体的压缩比,同时又保证了病灶区域的诊断价值。To solve the problem of the contradiction between high quality and high compression ratio of medical image compression, a medical image compression method combining near-lossless compression and lossy compression is proposed based on the theory of visual attention mechanisms： compress the lesion areas that have medical diagnostic value with low compression ratio near-lossless compression, and compress the non-lesion areas that have no medical diagnostic value with high compression ratio lossy compression. The results show that this method can effectively achieve compressing lesion areas and non-lesion areas respectively, then it could improve the compression ratio of the image relative to lossless compression, as well as gnaranteeing the diagnostic value of the lesion areas.

关 键 词：视觉注意 小波变换 整数小波变换 近无损压缩 有损压缩 

分 类 号：R445[医药卫生—影像医学与核医学]