Research On Surface Defect Detection Of Lithium Battery

Mobile phones have become an indispensable part of people's daily life.According to the survey conducted by relevant departments,the global handset sales exceeded 1.3 billion units in 2016 alone.According to statistics from related departments,the domestic handset production reached 120 million in the first half of 2017.The phone's battery life and the size of the appearance decide the popularity of mobile phones,after years of technical competition,the phone lithium battery with its high energy,long life advantages dominate.Taking lithium-ion cell phone as an example,the surface measurement,scratch detection and foreign object detection are mostly based on manual inspection.However,due to the small size of components,many details cannot be accurately identified by the human eye.This paper combines the machine vision and industrial testing to develop a mobile phone lithium battery surface detection system to achieve the appearance of mobile phone lithium battery size defects,the top seal defects and poor coding auto-detection.Firstly,the visual inspection hardware system is constructed according to the demand of industrial measurement.The camera,lens and strip light source model are selected,the relationship between the coordinates is established,the camera imaging model is built,and the camera parameters are calibrated by Zhang Zhengyou calibration method for subsequent acquisition image.Then,the measurement method based on the least square method is used to measure the appearance size.Filtration,binarization,morphology and edge detection of the image are used to get the edge points.Then the algorithm is modified by the Hough transform and the improved algorithm based on least square Compared with the measurement,it is found that the accuracy of the Hough transform measurement is three pixels.The improved algorithm in this paper can achieve one pixel with higher accuracy and higher robustness,which can effectively eliminate the interference of some singular points and avoid some deformed battery edge point is ignored.Secondly,the detection of peak seal foreign objects was realized by the feature extraction + support vector machine(SVM)algorithm.After binarization,morphology and edge detection,four features were extracted,such as area,contrast,position and orientation,Support vector machine algorithm extract of horizontal projection,vertical projection and variance,the detection results improved significantly.Finally,based on the method of affine transformation image registration,the difference method is used to detect the poor coding(deviation,leakage,etc.)in the area of the back of the battery.Experiments show that compared with the traditional method,the lithium-ion cell phone detection method designed in this paper has been improved both in detection efficiency and accuracy,and has good robustness.