Fang Chen
Machine vision, spare parts, surface defects, distinguish
The traditional defect detection of artificial mechanical parts has the problems of high detection cost and low efficiency. To solve this problem, the algorithm design for the surface defect recognition of mechanical parts is studied. The algorithm is divided into two parts: recognition algorithm and classification algorithm. Among them, the recognition algorithm is mainly correlation wrapper (CW) algorithm designed by research, while the classification algorithm is mainly support vector machine (SVM) algorithm adapted to design, they classify different types of mechanical surface defects through the identification classification step. The results show that in terms of detection accuracy, the detection classification accuracy of the research design model is basically higher than 90%, and the detection accuracy is high. In terms of detection time, the detection time of the research design model is basically between 0.5 s and 0.6 s. It can be seen that the model studied and designed can achieve high accuracy and high efficiency detection behaviour at the same time, make up for the low efficiency caused by the insufficient detection ability of the finishing link in the fully automated production line, and is conducive to the full automation, efficiency, and large-scale development of the industrial production line.
Important Links:
Go Back
