| The traditional cylindrical film-covered lithium battery circumferential film breakage detection is performed by the human eye.The human eye has the disadvantages of slow speed and fatigue,which is difficult to adapt to the detection requirements of the high-speed battery production line.Machine vision is a new type of industrial automation detection technology that can be combined with a computer to greatly improve the detection efficiency of the filmbreaking defects on the circumferential surface of cylindrical coated lithium batteries,while saving costs.In response to this requirement,this paper has designed the relevant algorithms of machine vision to detect membrane breakage defects,which reduces the rate of missed detection and false detection of membrane breakage defects on the battery circumferential surface.The main research contents are as follows:The characteristics of the cylindrical membrane-coated lithium battery's circumferential surface breaking film were analyzed,and it was divided into two types: exposed shell and fold according to whether it exposed the metal shell.When the mechanical transmission device drives the battery to rotate quickly,the mechanical black oil may contaminate the battery surface.The size and gray value of the oil stain in the battery image are similar to the exposed shell,which may be mistakenly detected as the exposed shell,and the oil stain needs to be analyzed.Distinguish from the difference between the exposed shell.Analyzed the reasons for the defect of film breaking,and analyzed the various factors that affect the gray scale of the image,and the influence of the light source on the defect detection in the imaging system,formulate the detection plan,and plan the algorithm flow of the entire system.Preprocessed the lithium battery image,studied how the lithium battery is positioned and corrected in the image,in order to adapt to different colors of batteries,a variety of automatic threshold extraction methods are analyzed,and the maximum between-class variance method is the best.In order to correct the battery offset,a method for correcting the rotation of the entire lithium battery image after positioning is proposed in the image rotation angle acquisition method.The angle obtained by this method is accurate and the correction effect is good.Analyzed the lithium battery image filtering,image segmentation,morphological processing and other algorithms,and preprocessed the original images collected by the machine vision imaging system to facilitate defect identification and extraction.Through a large number of tests,the problems that are easy to miss and misdetect are summarized.In order to improve the detection efficiency,this paper proposes four detection schemes.Option 1 selects the template image and the original image as the difference image,highlights the defect in this way,and then sets the gray threshold to extract the defect through the 3 criterion;Option 2 uses a dynamic threshold of different template sizes to extract the defect according to the position and size of the broken film;3 Divide the battery into multiple areas,so that the defects cause the standard deviation in the area to change significantly,and use the 3 criterion to set the gray threshold to extract the defects;Scheme 4 uses the median filtered image and the original image to make the difference image.Highlight the defects in a way,and then set the gray threshold to extract the defects through the 3 criterion.In addition,oil pollution detection is performed to distinguish it from the film-breaking defects and reduce the false detection rate.Through the demonstration of the scheme,scheme 4 can achieve the lowest false detection rate without missing the inspection,make up for the interference caused by the light and the color of the film,the algorithm has strong adaptability,and has high efficiency,high precision and good practical value. |
PDF Full Text Request