Design of bolt defect recognition system based on machine vision

With the rapid development of artificial intelligence, machine vision, one of its application fields, has received extensive attention. Applying it to the manufacturing process has become an important transformation direction of the manufacturing industry. The quality of threaded fasteners is directly related to the reliability and safety of mechanical products. However, bolt workpieces are prone to cracks, scratches, wrinkles and geometric defects in the production process. Traditional artificial defect recognition has the problems of slow speed, low accuracy and high labor cost. Therefore, a workpiece recognition system based on machine vision is proposed to realize the automatic recognition of defective bolts. Firstly, the principle and use process of machine vision are introduced, and the system is developed in OpenCV open source vision library. Then the basic process and key technologies of defect recognition are studied. The key technologies include image graying, mean filtering, edge detection, threshold segmentation, contour extraction and other technologies. These technologies can greatly reduce the image size, remove most of the interference factors of the image, strengthen the target area in the image, improve the speed and accuracy of defect detection, and lay a theoretical foundation for subsequent system development. Then the workpiece recognition system is designed, including two modules: data set and defect recognition. The BP neural network is used to identify the defect, and the principle of this method and how to realize the defect detection function are studied. Finally, the defect workpiece recognition experiment was carried out. The system displayed the experimental results by labeling the defect type and displaying the code. The experimental results verified the accuracy and feasibility of the system. The recognition system has high reliability for bolt defect recognition, and has certain reference value for bolt defect recognition research based on machine vision.