基于四分树的多特征纹理图像分割

MULTI-FEATURE TEXTURE IMAGE SEGMENTATION BASED ON QUADTREE

本文提出了一种基于四分树的多特征纹理图像分割方法。该方法针对四分树和纹理特征本身的特点,灵活地选择了分形模型,灰度共现矩阵和纹理能量方法提取纹理特征,用一种简洁的方法基于多特征判断图像块属性的一致性,用分开-合并算法分割图像。该方法能够准确和快速地分割纹理图像。实验结果是令人满意的。

A texture image segmentation approach using multi-features based on quadtree is proposed in the paper. Methods of the fractal model, the gray co-occurrence matrix and the texture energy are selected flexibly to extract texture features according to the characteristics of the quadtree and the texture features. The consistence of image blocks based on multi-features is determined by a simple method, an image is segmentated by the split-and-merge algorithm. The approach can segment texture images precisely and fast. The experiment, result is satisfied.