Font Size: a A A

Automatic Detection And Classification Of Surface Defects

Posted on:2016-03-19Degree:MasterType:Thesis
Country:ChinaCandidate:Q LiFull Text:PDF
GTID:2272330470967739Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
In the automobile manufacturing industry, the appearance of the body outer surface is a measure of body mass and an important aesthetic criteria. The requirement of the surface quality detection has become increasingly high. However, there is no good way to its automatic detection and classification. And most applications require complex user interaction. Automatic detection and classification not only help to improve the accuracy of detection, but also reduce production costs. Therefore, automatic surface defect detection and classification methods seem indispensable.This paper include automatic surface defect detection module and surface defects classification module. In automatic surface defect inspection module, firstly, I implement Reflection Map algorithm. The result of this algorithm depend on the position of viewpoint and need lots of user interaction. To solve this problem, this paper presents the Integral Reflection Map algorithm which integrate the defect value in valid viewpoint area. Finally, this paper propose Local Alignment Integral Reflection Map algorithm to automatically deal with some special cases which two models can’t be aligned globally. This algorithm can also detect different scale defects.The algorithms in surface defects detection module are based on reference model and manufacture model. For the absence of the reference model, surface defects detection algorithms will not be able to meet customer requirements. Therefore, I propose and implement the surface defects classification algorithm to classify defects on manufacture model. Firstly, I apply local height map Zernike coefficient, curvature Zernike coefficient, smoothing curvature Zernike coefficient and curvature histogram as surface descriptor. Secondly, this paper use Support Vector Machine to obtain a classifier. I label defects classification on manufacture model by using this classifier. Finally, this paper analyze and discuss the advantages and disadvantages of these different kinds of surfaces descriptor.
Keywords/Search Tags:Surface Defect, Defect Classification, Surface Descriptor, SVM
PDF Full Text Request
Related items