基于小波变换的纹理图像分割  

A Study Of Segmentation Of Textured Images Based on Wavelet Transform

作　　者：黄兴滨[1] 谷光琳[2] 刘伟东 

机构地区：[1]黑龙江大学 [2]哈尔滨工业大学 [3]省科技开发中心

出　　处：《自动化技术与应用》2003年第2期13-14,24,共3页Techniques of Automation and Applications

摘　　要：基于小波的纹理图像分割方法是把小波变换应用于纹理的特征提取。通过对原始纹理图像进行高阶小波分解,构成每个象素对应一个特征矢量,对所提取的特征利用均值举类算法进行分类,最终获得图像的分割结果。为了提高纹理分割图像的边缘准确性和区域性以及降低分割错误率,提出了利用特征加权来进行后分割的方法。The method of texture image segmentation based on wavelet-transforms to feature extraction . Through processing pyramid hierarchical wavelet for original images. each picture element correspond to a feature vector. The result of segment image can be get finally with the mean of duster-ing algorithm to classify the feature extracted. To improve the accuracy of boundary locations and region homogeneity as well as to reduce error rate in texture image segmentation, a novel approach based on using feature weighting is proposed in this paper.

关 键 词：纹理图像分割 小波变换 图像处理 特征提取 聚类 特征矢量 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TN911.73[自动化与计算机技术—计算机科学与技术]