Research On The Visual Inspection Method For The Defects Of The Cylindrical Lithium Battery Shell

As a new energy battery,cylindrical lithium batteries have been used as important energy supply components in various fields.With the rapid development of science and technology,the demand for batteries is increasing,and the quality of batteries has also attracted people's attention.Due to the production process and other reasons,there are pits,bumps and scratches on the surface of the battery casing.These defects will not only affect the quality of the finished battery and cause losses to the battery manufacturer,but also due to the electrochemical discharge nature of the lithium battery.It may even cause hidden safety hazards such as fire and cause personal and property safety.Therefore,research on battery shell defect detection is very necessary.The traditional defect detection methods can no longer meet the production requirements of industrial automation,and the detection accuracy cannot reach modern standards.In this thesis,a complete visual inspection system for the defect of the circumferential surface of the battery shell is introduced,including hardware and software.Based on the classic threshold segmentation method,a method for quickly segmenting the circumferential image of the battery shell is improved.Uneven detection and misdetection caused by the grayscale difference between the edge background and the inner surface.This paper innovatively proposes a one-dimensional partitioned median-Gaussian filter image preprocessing algorithm,which can eliminate image noise while simultaneously Retain defect information,especially the defect information at the edge position;flexibly apply the moving average filtering algorithm to the grayscale of the preprocessed image according to the defect illumination imaging model to highlight the defect;set the dynamic threshold to segment the defect according to the characteristics of the residual curve Value processing;mark the defect location on the original image through morphological closure processing and connected domain analysis.The detection system designed in this thesis conducted a group comparison experiment on 1200 battery shells to be tested.The results prove that the battery shell defect detection algorithm designed in this paper has faster detection speed and higher detection accuracy,and can detect concaves quickly and accurately Pits,bumps,scratches,brushing,etc.mainly affect defects in battery quality.Compared with the results of traditional un-partitioned filter processing,the detection accuracy of partitioned processing is about 20% higher than that of un-partitioned processing,which significantly reduces the missed detection rate;compared with other defect detection methods,the method proposed in this paper can be very good Taking into account the speed and accuracy,it can process the defect situation of the image of the circumferential surface of the battery shell of 4 pixels per second and the detection accuracy is not less than 95%,which fully meets the real-time detection requirements of industrial automation,which shows the effectiveness and practicability of the algorithm.