摘要
噪声去除是带钢缺陷自动检测系统的一个重要环节,传统的滤波方法在去噪的同时会造成边缘的模糊,影响检测及质量评价结果。为解决上述问题,文章提出一种基于邻域信息的非迭代、非线性倒数加权滤波方法-双倒数波方法,该方法同时结合邻域内像素的几何位置信息与灰度信息进行滤波,采用邻域内各像素与中心点之间的几何距离差倒数与灰度相似性倒数的乘积做为最终的权值。文中分别采用高斯滤波、各向异性扩散滤波及双倒数滤波三种方法对带钢表面缺陷进行处理,实验结果表明:双倒数滤波去噪和边缘保持效果明显优于其它两种滤波方法,可满足带钢缺陷检测系统在图像噪声去除同时保留重要的边缘细节信息的要求。
Noise reduction is an important part o the surface defect inspection system. The traditional filter method would blur the edge, which has a severe effect on the accuracy of detection and the results of quality evaluation. In order to solve this incompatible problem, a locally non-iterative and nonlinear filter based on reciprocal weight, we call it double-reciprocal filter is presented. It combines gray levels based on both their geometric closeness and their photometric similarity, and prefers near values to distant values in both domain and range, it adapts the produce of the geometric closeness reciprocal and photometric similarity reciprocal as the weight. The Gaussian filter, anisotropic diffusion and double-field filter have been adopted to process several kinds of typical defect. The results have been compared. It is testified that： this method can remove the noise affectedly and preserve the edge distinctly, it can satisfy the demand of steel strip defect detection.
出处
《组合机床与自动化加工技术》
北大核心
2009年第8期45-48,53,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
国家自然科学基金资助项目(50574019)
国家高技术研究发展计划(863计划)资助项目(2008AA04Z135)
关键词
带钢缺陷
图像去噪
倒数加权
双倒数滤波
边缘保持
steel strip defect
image denoising
reciprocal weight
double-reciprocal filter
edge preserving
