Recognition and classification of surface defects of aluminum castings based on machine vision

In order to improve the efficiency of surface defect detection of casting parts in industrial production, machine vision method was used to automatically identify and classify the surface defects of castings. Image processing technology is used to preprocess the collected sample image, enhance the image information and segment the defect area. The geometric features, shape features and gray features of various defect images are analyzed, and the supervised learning classifier is designed to classify and recognize the parameters of casting feature set. By comparing radial basis function kernel function SVM, BP neural network and polynomial kernel function SVM. Experimental verification: The average recognition rate of RBF kernel SVM is 96%, which is higher than the other two methods. The results show that the classifier can meet the needs of industrial detection.