基于去降Mallat离散小波变换的彩色图像分割  

Color Image Segmentation Based on Mallat Discrete Wavelet Transform without Subsample

作　　者：刘昌进[1] 郭立[1] 朱俊株[1] 徐雯[1] 

机构地区：[1]中国科学技术大学电子科学与技术系,合肥230026

出　　处：《计算机工程与应用》2003年第11期93-95,102,共4页Computer Engineering and Applications

摘　　要：该文针对Mallat快速离散小波变换,提出了一种利用变换平移不变性的离散小波变换的彩色图像分割方法。首先对原始图像进行平移不变性的小波变换,然后提取颜色和纹理特征,并采用k均值算法进行分割。实验表明该方法对纹理图像和彩色自然图像都具有较好的分割效果。This paper introduces a method of color image segmentation based on a modified Mallat Discrete Wavelet transform.Translational invariance can be retrieved if subsample is ignored in Mallat algorithm.Authors apply this trans-form to color images,including texture images and natural images.Then color and texture features are obtained and k-means classification method is applied.Experiments demonstrate that this method can perform better than the traditional Mallat method.

关 键 词：MALLAT算法 小波变换 图像分割 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]