基于链码改进算法的钢轨表面缺陷识别

Rail surface defect detection based on improved chain-code algorithm

从基本链码算法的数据结构、程序流程和结束条件的设计分析入手,针对易出现的孤立点、单列链、局部链、开链等错误跟踪问题,提出最大区域数限定下的节点标记法与自适应后退两步搜索法相结合的Freeman链码改进算法。实验研究中,采用迭代阈值法对钢轨表面图像进行二值化分析,确定最佳阈值,采用改进算法跟踪钢轨表面缺陷边界,获得较好的缺陷区域提取效果。该算法结合了最佳阈值分析方法,避免了传统链码算法的错误跟踪问题,能够有效识别图像的区域性缺陷。

Starting with the data structure,program flow and termination condition of basic chain code algorithm,an improved Freeman chain code tracing（FCCT）algorithm combining adaptive two-steps back searching（ATSBS）and node labeling under limitation of max zone number was presented,so that specified problems of isolated point,single column chain,local chain and open loop chain were solved.In experimental study,an iteration thresholding method was used for threshold analysis of rail surface image which gave instruction for best threshold selection.The improved FCCT algorithm was applied in boundary tracing of rail surface defects and desired defects extraction effect was achieved.Integrated with best thresholding analysis,this algorithm avoids incorrect trace of the traditional tracing algorithm and effectively recognizes regional defects.