Research On Detection Method Of Circumferential Surface Defects Of Cylindrical Coated Lithium Battery Under Color Image

Cylindrical coated lithium battery is a kind of lithium battery,which has many advantages such as high safety,high energy density,fast charging,long endurance,green and pollution-free,long service life and so on.Therefore,its application range is becoming more and more extensive and it gradually occupies a larger market.In the process of battery production,various defects may appear on the surface of battery coating,thus affecting the stability and assemblability of products.At present,the surface defects of coated batteries mainly rely on manual inspection,which has slow inspection rate and high missed inspection rate.With the increasing labor cost,the inspection method using machine vision has attracted more attention.In this paper,the detection method of circumferential defects of cylindrical lithium battery under color images is studied,and the detection of film-coating broken defects,film-coating wrinkle defects,and the extraction of spray code on the film-coating surface are realized.The research contents of this paper mainly include the following aspects:(1)Introduced the related defects and interference on the surface of the battery coating,analyzed the causes of the defects,discussed the key and difficult points of this topic,and introduced the overall design scheme of the system algorithm.(2)In the aspect of cell region segmentation,different methods are used to transform the original color image into single-channel gray image,and the characteristics of these singlechannel gray images are analyzed.A method of using hue histogram search to realize cell region segmentation is adopted.The method first obtains the standard hue range of the battery coating with different colors through statistics,and then judges the degree of coincidence between the dense interval in the hue histogram and the standard hue range,and then realizes the segmentation of the coated battery area.This method improves the shortcomings of edge detection and region growth in region segmentation of coated battery.(3)In the aspect of bar code extraction,the difference of gray value between spraying code and coating film on lightness image is analyzed.Aiming at the deficiency of extracting fine bar codes and characters by dynamic threshold and extracting characters by gray distribution curve,a bar code and character extraction method based on local mean and standard deviation is adopted.This method first sets the size of the local window according to the overall size of the bar code and the character,then sets the threshold parameters by calculating the local mean and standard deviation in the window,and finally uses the minimum gray difference to reduce the interference of background fluctuations to realize the extraction of bar codes and characters.(4)In the aspect of defect detection,the characteristics of broken film,wrinkle and dirt are analyzed,and the detection method of broken film and wrinkle defects on the circumferential surface based on gray distribution curve is adopted.Firstly,the suspected defect area is selected by dynamic threshold,and the gray distribution curves in four directions of the suspected defect area are calculated,and then the defects in this area are judged according to the gray distribution curve types.To a certain extent,this method improves the problem that the detection method of broken film and wrinkle defects based on HSV color space has a high false detection rate of broken film and dirt.