热轧重轨表面缺陷在线检测识别的关键技术研究

Study on the Key Technology of Hot Rolling Heavy Rail Surface Faults of Online Detecting and Recognition

目前热轧重轨表面缺陷检测速度慢、精度低。为此，提出了一种基于机器视觉的热轧重轨表面缺陷在线检测系统。分析了过暗过曝区域交叠融合法与图像像素线互相关校验法两种方法提取特征缺陷等关键技术，并对模糊脉冲神经网络的表面缺陷分类效果进行了研究。实际应用证明，采用上述机器视觉的检测关键技术对热轧重轨表面进行缺陷检测识别，较大提高了检测速度和精度，且检测正确率在90％以上。

In currently hot roiling heavy rail surface faults detecting, speed is slow and its precision is low. So a suit of surface defect detection system for hot rolling heavy rail based on the machine vision is produced. Too dark and sun regional overlapping fusion method and image correlation between pixel hnes algorithm is analysised, and a fuzzy spiking neural network used to make a classification for the characteristics of low SVM training algorithm is researched. Using above key machine vision technology for detection of hot heavy rail surface defects identification, the speed and accuracy of online testing can be greatly improved, and the detection correction rate is over than 90%.