基于熵和变异度的织物疵点图像分割方法

针对现有织物疵点图像分割方法对光照不均匀敏感的问题，提出了一种基于局部熵和变异度的织物疵点图像分割方法。首先对织物图像进行局部熵和变异度计算，提取疵点的类边缘和区域信息；然后基于人工神经网络脉冲耦合（PCNN ）的区域生长法分割织物疵点图像。通过对T ILDA数据库中的疵点图像和基于线阵CCD在线检测的织物疵点图像进行测试，并与已有的相关方法进行对比实验和评价。结果表明，该方法不仅能有效地抑制光照不均匀和复杂背景干扰的影响，而且分割质量有了明显改进。

A fabric defect image segmentation method based on entropy and variation degree

A fabric defect image segmentation method based on local entropy and variation degree is pro-posed ,w hich aims to overcome the over or under segmentation of fabric defect image for uneven illumina-tion in some existing segmentation algorithms .First ,the edge and regional information is extracted from local entropy and variation degree of the fabric defect image ,then fabric defects are segmented by a means of region growing based on PCNN .Comparing with some existing segmentation algorithms by the test of fabric defect images in TILDA database and that captured by CCD ,the experimental results show that this algorithm can effectively overcome the shortcomings of uneven illumination and complex back-ground .