一种新的纹理图像特征提取的方法

灰度图像的纹理反映了一个区域中像素灰度级及其局部变化的空间分布属性，用图像的局部统计特征能较好地刻画不同纹理的差异。基于这种思想，本文中首次提出了一种基于PCNN点火序列图的纹理图像特征提取的新方法。通过对PCNN的运行行为和基本特性的分析。指出PCNN的点火时刻序列图不仅包含了局部图像的灰度分布信息，更重要的是还包含了图像中相邻像素之间的几何信息，这恰是纹理图像的个性特征所在，最后给出了部分仿真实验的结果，以验证该方法的有效性。

Research of fabric defect detection based on improved PCNN

Texture analysis is the emphasis and difficulty of researches in the fields such as computer vision, image understanding and pattern recognition , while the feature extraction of texture image is a basic research topic . The texture of gray scale image is a property that reflects the spatial distribution of the gray scale levels of image pixels in a region and their local variations. So the local statistical features of an image can characterize the differences of different textures. In view of this thought, a new method of feature extraction of texture images based on PCNN firing series image frame is proposed in this paper for the first time. After analysis of the running action and basic properties of PCNN, it is pointed out that PCNN firing series image frame contains not only information about the distribution of gray scale of local images, but also geometry information between adjacent pixels, which is more important and just the individual characteristics of texture images. The validity of this method is well verified by our experiments.