Research On Surface Defect Detection Method Based On Photometric Stereo Vision

In recent years,with the development of science and technology and the acceleration of global industrialization,the public demand for various industrial products is also increasing.Therefore,the quality detection of industrial products is extremely important.Manual detection is a traditional method to detect product defects,which is timeconsuming and labor-intensive,high cost and low efficiency,so the use of machine vision to detect industrial product defects has become a research topic of great practical significance.Using cameras instead of human eyes and using computers instead of human brains to identify defects can greatly reduce labor costs,but also greatly improve the detection accuracy and reduce the false detection rate.At present,the defect detection based on traditional two-dimensional vision has the problem of information limitation,which can not well display the information of all sides of the object.However,stereo vision technology can effectively capture the 3D information of the measured object and obtain the 3D coordinates of any point on the 3D outline of the measured object.Therefore,stereo vision technology has a wide range of application prospects in the field of machine vision.Therefore,this paper aims to study the surface defect detection method based on photometric stereoscopic vision,take the high temperature resistant container(alumina)produced by the cooperative enterprise as the research object,and design the container surface defect detection method for the surface defects of the container.The specific research content of this paper is as follows:A hardware prototype system for image acquisition of high temperature resistant container is designed and built.In order to build a hardware prototype system suitable for the research object in this paper,this paper designed a lighting scheme based on the four-light source photometric stereoscopic method suitable for the surface defect detection of high-temperature container by comparing the corresponding angles of different light sources and the number of different light sources The container photometric stereo images of high quality can be obtained effectively.An image enhancement algorithm combining curvature filtering and bilateral filtering is proposed.The quality of the image obtained by the four-light source photometric stereoscopic method is much higher than that of the two-dimensional image collected under the single light source,but the features of some defects are not well displayed,which will affect the accuracy of the subsequent defect detection.Therefore,the image enhancement algorithm combined with curvature filtering and bilateral filtering can enhance the image effect of defect images,which can effectively highlight the difference between the defect part and the background part and facilitate the subsequent defect feature extraction.A surface defect detection method based on photometric stereoscopic vision is designed.Firstly,four-light source photometric stereoscopic method was used to reconstruct the surface image of the high temperature resistant container.Then,the image enhancement algorithm proposed in this paper was used for filtering processing to obtain the image with enhanced defect features.Then,the region of interest was separated from the background image by extracting ROI for the convenience of subsequent defect detection.The automatic global threshold segmentation method based on gray value was used to segment the ROI region,and then the defect region was extracted according to the shape features and displayed on the albedo image.Based on the above key algorithms,the surface defect detection system of high temperature resistant container is designed and developed.According to field research,the container defect detection system is designed,the hardware system and the software system are built according to the requirements of actual industrial applications and enterprises,and a container defect detection system with high efficiency and accuracy has been developed.This paper has done some technical exploration on the surface defect detection of industrial products,and has made some technical accumulation for subsequent enterprises to develop more optimized performance of high-end surface defect detection system of industrial products.Moreover,the accuracy of the experimental results in this paper is reliable.The method has been applied to defect detection of high temperature resistant container in cooperative enterprises,and has been tested online and achieved good detection results.It provides certain experience accumulation for stereo vision technology in the field of product defect detection.