小波变换在三维医学图象分割中的应用  

Application of Wavelet Transform in 3D Medical Images Segmentation

作　　者：袁野[1] 仲崇权[2] 秦绪佳[3] 

机构地区：[1]上海交通大学图像与通信研究所,上海200030 [2]大连理工大学自动化系,辽宁大连116024 [3]浙江大学CAD&CG国家重点实验室,浙江杭州310027

出　　处：《小型微型计算机系统》2003年第6期1081-1083,共3页Journal of Chinese Computer Systems

摘　　要：本文将多尺度小波变换应用到三维医学图象分割中的阈值选取中 .由于小波变换在较大尺度 ,由噪音引起的细小突变较少 ,可描述信号的整体行为 ,取较大尺度的由负到正的零交叉点来确定图象的阈值 ,再逐步到相邻的小尺度由粗到精地确定精确的阈值 .用以上算法对三维医学图象进行二值化后 ,根据待提取组织或区域的特征 ,再选取合适的数学形态学操作 ,最后对区域进行种子填充 .从实验结果可以看出分割效果较好 。In this paper, a method in which multi scale wavelet transform was applied to finish threshold selecting in 3D medical images segmentation was developed. Based on the characteristic in big scale that there are little detail which was caused by noise , and the characteristic that the sketch will be keep in big scale, the zero crossings whose left value are negative and right value are positive in big scale were determined to be the initial thresholds, then the accurate thresholds in small scale will be found gradually. We binarize the image using this method, process the image using properly mathematical morphology operation according to the feature of tissues or regions extracted, then fill the region extracted using seed fill algorithm. The experimental results show that the algorithm is feasible.

关 键 词：医学图象 小波变换 图象分割 阈值 

分 类 号：TN911[电子电信—通信与信息系统]