Research On The Inspection Method Of The Surface Defects Of The Small Cylindrical Components Based On Machine Vision

The detection of surface defects of components has always been an indispen-sable part in industrial production.If there are defects on the surface of components,it may cause certain safety hazards.In current industrial manufacturing,the detection of surface defects in many products still relies on manual inspection.The manual inspection process is a very time-consuming and labor-intensive process and it is difficult to meet the large-scale production of components.Under long-term high-intensity pressure,manual detection may cause missed judgment and misjudgment,which greatly affects the accuracy of product detection.With the continuous advancement and development of science and technology,defect detection of products has been continuously developing in the direction of automation.In order to improve the deficiency of manual detection,this paper proposes an automatic defect detection technology based on machine vision on the basis of in-depth study of surface defect detection.According to the characteristics of the sample to be measured,an automatic end-face image acquisition device was designed.The contrast experiment was performed on the lighting conditions of the end face of the sample.Finally,the end-surface lighting was selected using the structured light-driving method that best showed the end face defects.In order to reduce the missed detection rate of defects and improve the detection efficiency of defects,two pictures with different defect information are obtained by using left and right structured light lighting methods.After the pretreatment of the face image,the sample target area is extracted,and the left lighting image is obtained according to the template matching and statistical analysis method to obtain the rough area of the defect,and the scatter chart of the two target areas is image fusion to find the cross scatter chart.Finally,the Delaunay triangulation technique was used to analyze the data of the cross scatter plots,and finally an accurate face defect region was obtained.This article also uses the photometric stereo algorithm and DBSCAN clustering algorithm to analyze the surface defects,and uses three methods for defect detection: repeatability error(RSD),runtime(T),standard deviation(MSE),and correct recognition rate.Effectiveness analysis.The experimental results show that the Delaunay triangulation technique has faster running time,less repetitive errors,and higher recognition rate of central defects and cross-sectional defects than the photometric stereo algorithm and the DBSCAN clustering algorithm.