摘要
在研究小波变换及布朗分形模型的基础上，定义了水平分维数，垂直分维数，对角分维数，并将它们作为纹理特征量来实现纹理的分割．实验结果表明，这些特征量能够很好地反映纹理特征，从而分割时不需有关图象中纹理类别的先验信息，较以往方法在准确度和计算复杂程度上都有很大提高．
Based on research of wavelet transform and Brownial fractal model, the paper defines horizon fractal dimension, vertical fractal dimension and diagonal fractal dimension as texture features to perform texture segmentation. Experimental study has shown the features are effective and the approach can well perform texture segmentation without any prior information about texture kinds in a image. It yields high accuracy and.less calculation.
出处
《武汉大学学报（自然科学版）》
CSCD
1998年第1期118-120,共3页
Journal of Wuhan University(Natural Science Edition)
关键词
小波变换
布朗分布模型
纹理分割
图象处理
wavelet transform,brownial fractal model,fractal dimension,texture segmentation
