基于人类视觉感知特性的一种纹理图象分割方法

APPROACH OF TEXTURE IMAGE SEGMENTATION BASED ON HUMAN VISION PERCEPTION

在纹理图象分割的研究领域中 ,基于人类视觉感知特性的图象分割方法是一个重要的新方向 .本文给出了一种纹理图象分割的方法 ,设计了一组参数不同的 2 D最佳正交极可分方向滤波器来提取图象特征 ,然后经过四叉树平滑 ,用模糊 C-均值聚类对图象进行分类 ,最后进行边缘确定 .实验结果表明 。

In the field of texture image segmentation, the research based on human vision perception is a crucial question. This paper gives an approach for texture image segmentation. A group of 2D optimal orientational filters with different parameters are designed to extract features of the image. Then the extracted feature image is smoothed and clustered through fuzzy c-means network. Finally we determine the edge of the image, thus getting the result of the image segmentation. Experiments on artificial and natural texture images show that this approach is effective.