基于人类视觉特性的纹理分割方法

Texture Segmentation Method Based on the Property of HVS

纹理分割是将一幅图像依据纹理不同分成若干个不同的区域，目前广泛采用的是利用滤波器族（如Gabor）对图像进行分解。但由于图像纹理表现的各异性，通常要选择很多滤波器，导致提取的特征困难，分类效果不好，效率低，使用范围受限。文中提出了一种基于人类视觉系统（Human Visual System，HVS）二阶机理的纹理分割方法，即基于‘滤波器-〉整流-〉滤波器’（FRF）模型的纹理分割方法。该算法符合HVS区分纹理机理，计算过程简单、方便。针对各纹理选取的特征明显，分类效果好，效率较高。

The texture segmentation means dividing an image into several different areas based on the texture difference,now, it is widely that use filter group （such as Gabor） to decompose the image.However,because of the opposite sex of the texture,usually a lot of filters to choose,leading to difficulties in extracting the characteristics,classification results are not satisfactory, efficiency is low, the use of limited scope.In this paper,proposed a texture segmentation method based on second-order processing mechanism of human visual system （HVS）,which is a texture segmentation method based on ＇Filters-〉rectifier-〉Filters＇（FRF） model.The algorithm is consistent with the HVS distinguish the texture mechanism,the calculation process simple and convenient.Features for each texture were selected evidently, classification effect is good and efficiency is high.