Surface defect detection method for round rob products based on similarity of feature matrix
To solve the problems of low efficiency,low precision,and inconsistent standards in the surface de-fect detection of round rod products in the manual sampling inspection process of quality inspection laboratory,a product surface defect detection method based on feature matrix similarity is proposed.Firstly,through analy-sis of product surface defects,the multi-threshold segmentation method based on the maximum inter-class vari-ance is used to segment the product surface defect area,and the defect features are extracted from many aspects such as shape contour,color,location and regional features,and the data feature distribution of various defect images is established.Then,the defect features with different characteristics and quantities are uniformly con-verted into the form of feature detection matrix,and the cosine similarity is used to measure the similarity be-tween the defect feature matrix and the product surface defect feature matrix,so as to judge whether the product has defects,and classify the product surface defects.Taking packaged products as an example,4 820 product sample images are collected for defect detection experiment.The results show that the average accuracy,false detection rate,missed detection rate and Kappa coefficient of the proposed method for the detection of product surface defects are 97.48％,3.30％,0.77％and 0.955 7,respectively,which verifies the effectiveness and con-sistency of the method.
NETL
NSTL
万方数据
