基于神经网络的零件缺陷机器视觉识别系统

零件缺陷检测是保证零件使用安全的重要手段。传统的零件缺陷检测法需要有操作人员参与其中，易受主观因素影响，检测的效率及精度得不到良好的保证。而采用机器视觉技术的检测法可实现实时在线的自动检测，无需人工参与，这就极大的提高了生产效率。本文以小轴承表面为研究对象，针对微小轴承的表面结构、尺寸、检测精度和缺陷特征，设计了基于BP神经网络的零件缺陷机器视觉在线自动检测系统，其采用机器视觉技术，构建了BP神经网络检测识别模型，采用进行图像特征提取的间接识别方法，对微小轴承缺陷进行实时检测。实验结果证明了人工神经网络模型的检测能力的可靠性。

Computer Vision Recognition System for Component Defect Based on Neural Network

Small bearings, axles, and etc are important components for machines,vehicles, engines and etc. In order to improve the detection efficiency and detection accuracy of its surface defects, Taking small bearing surface as the object of study, and putting forward the method to realize real-time online automatic detection of bearing surface defects based on machine vision technology and designing the online automatic detection system of defective parts machine vision based on BP neural network. According to the micro bearings surface structure, size, accuracy and defect characteristics, using machine vision technology, preprocessing for the collected image, and constructing BP neural network detective model, extracting target area in the image by Hough transform and Roberts operator. Based on the combined-moment invariants, the defects of the bearings are judged, and thus the defects of the small bearings are detected in real time.The simulation results of MATLAB verify the reliability and effectiveness of ANN detection model.