基于小波分析的图像自适应阈值选择  

IMAGE AUTOMATIC THRESHOLD SELECTION BASED ON WAVELET ANALYSIS

作　　者：赵立初[1] 王积分[2] 

机构地区：[1]上海交通大学 [2]天津大学计算机科学与工程系,300072

出　　处：《模式识别与人工智能》1999年第1期79-84,共6页Pattern Recognition and Artificial Intelligence

基　　金：国家自然科学基金

摘　　要：本文充分利用小波作为"数学显微镜"检出奇异信号的特殊功能,对医学显微图像—白内障进行了自适应的阈值选择。图像直方图的特征点(波峰和波谷)可以通过小波变换的特征点(过零点或极值点)由粗到精地表示。通过理论分析各小波的特性以及类比几种小波的图像处理后的效果,确定出相对最佳小波和分解级数。文章指出,小波自适应阈值选择为白内障边界抽取和定量诊断提供可靠保障。As ' mathematical microscope ', wavelet can detect the singularity of the signal precisely. This paper uses this properties to determine the thresholds of the medical microscope images automatically. The characteristic points of histogram (peak and valley) can be represented by the characteristic points of wavelet transform (zero-crossing or extreme points) from coarse to fine. Analyzing the property of different, wavelets and comparing the processing results using different wavelets, the better wavelet, and the better decomposition stage are discussed. Finally, wavelet automatic threshold selection provides the reliable guarantee for cataract edge detection and quantitative diagnosis.

关 键 词：小波分析 阈值选择 边界强度 图像处理 白内障 

分 类 号：R776.1[医药卫生—眼科] TP391.41[医药卫生—临床医学]