Detection and Classification of Metal Workpiece Surface Defects Based on Machine Vision

In this era of mechanization and intelligence, all aspects of our daily life are inseparable from a variety of mechanical equipment, and metal workpieces are widely used in the manufacturing of mechanical equipment. With the development of society, the productivity of the manufacturing industry is increasing and the requirements for part quality are becoming stricter. However, defects will more or less exist in the production and transportation of metal artifacts. Therefore, the defect detection of metal workpieces shows its value. This paper designs a series of automatic defect detection and classification procedures of metal workpieces based on image processing and machine learning technology: image segmentation after pretreatment of the collected sample image, extract the values of grayscale, shape and HOG characteristics, input them into the SVM and KNN classifier for training, so that it can't only realize the classification of different types of defects, but also distinguish real defects and fake defects. Finally, through multiple flaw detection tests, the average recognition rate reached 92.6%, which has a certain practical value.