Research On Visual Inspection Method For Defects Of Round End Surface Of Steel Shell

The steel shell is a necessary part for the production of cylindrical lithium batteries.In order to ensure the quality of lithium batteries,the surface of the steel shell needs to be tested for defects.Defects such as pits,bumps and scratches are common defects on the surface of the steel shell.When collecting the round end image of the steel shell,the edge defects are difficult to detect due to the reflection of the steel shell surface and the uneven illumination on the edge area.In response to these problems,this paper designs a visual inspection system for the defects of the round end surface of the steel shell,and improves the accuracy of the defects detection of the round end surface of the steel shell from both optical imaging and software algorithms.For the problem of serious reflection on the round end surface of the steel shell,the optical imaging part selects the high-resolution industrial camera,coaxial light source and custom industrial lens to suppress the impact of the reflected light of the end surface on the imaging quality and effectively improve the image quality of the round end surface of steel shell.Since the steel shell is manufactured by stamping process,the edge of the round end surface of the steel shell is arc-shaped,resulting in the middle part of the image taken by the camera is bright and the edge part is dark.This paper proposes a method of regionbased defect detection based on gradient operation to detect defects on the round end surface of the steel shell.The specific implementation method is as follows:Firstly,a comparison experiment of image denoising based on mean filtering,median filtering,and gaussian filtering of steel shell round end surface is performed.By comparing the mean square error(MSE)and peak signal-to-noise ratio(PSNR),the gaussian filtering algorithm is used for image denoising on the image of the shell's round end surface.By comparing and studying the histogram equalization,CLAHE algorithm,horizontal and vertical difference method on the steel shell round end image enhancement effect,the horizontal and vertical difference method is finally selected to enhance the image of the round end image of the steel shell.Secondly,aiming at the difference between the non-edge area defect and the edge area defect of the round end surface of the steel shell,a region-based defect detection method based on gradient operation is proposed.For the non-edge area defects,the method of binarizing the gradient image is used for defect detection.For the edge area defects,first extract the inner ring curve of the approximate ring in the binarized image,calculate the distance between the inner ring curve and the inner ring curve fitting ellipse,calculate the distance threshold according to the 3? criteria to locate the edge defect,and then calculating the number of connected regions of the binarized image of the edge defect area further eliminates false defects.Finally,this paper selects 300 non-defective samples and 300 defective samples to conduct a feasibility test on the visual inspection method of the steel shell round end surface defects proposed in the paper.The test results show that the average missed detection rate and the average false detection rate of defects on the round end surface of the steel shell are 3.33% and 7% respectively,the average detection time of a single steel shell is 128.43 ms,and the detection system can continue to run steadily,which can meet the actual production requirements.