中文摘要：

针对热轧板带生产线相对恶劣的生产环境及表面缺陷检测精度低的问题,设计了一套热轧板带表面缺陷在线检测系统,深入研究了系统构成、图像去噪处理、缺陷的检测和分类、特征提取和缺陷识别。本文首先对图像进行形态滤波去噪处理,提出了一种基于融入熵的BING检测算法,采用P-relief特征检测和多分离器集成识别算法对热轧板带多种缺陷图像进行处理。实验证明本文算法能够有效地实现缺陷图像的检测和精确的分类。

英文摘要：

In the light of the problem of poor production environment and low precision of surface defect detection for hot strip production line,a set of on-line inspection system for the surface defects of hot rolled strip is designed.The system structure,image denoising,defect detection and classification,feature extraction and defect identification are studied.Firstly,morphology image filtering denoising is employed,then a BING detection algorithm integrating based on entropy is proposed,and a P-relief feature detection and recognition algorithm are utilized to process hot rolled plate w ith a variety of defect images.Experimental results show that the proposed algorithm can effectively detect the image defect,and accurate classification is also realized.