Research On Welding Defects Detection Of Camera Compact Module Based On Machine Vision

Camera Compact Module(CCM)is an important component of smart devices.With the rapid development of intelligent equipment industry,the demand for CCM increases drastically,and higher requirements are put forward for the welding quality inspection of CCM products.At present,the existing quality inspection solutions of domestic manufacturers have the shortcomings of high cost,high false positive rate and low efficiency.Therefore,the study of an effective CCM welding defect detection method to achieve the localization of equipment is an inevitable trend in the development of the industry.This paper focuses on the existence of bridge joint,leak welding and less welding defects in CCM welding,and focuses on key technologies such as machine vision optical imaging solutions and image processing algorithms.The main contents include:(1)Designed a welding defect detection system based on the criteria for the detection of welding defects in CCM.By analyzing the optical imaging characteristics of the welding area to be detected,the CCM welding spot detection imaging model was constructed.The low-cost and high-efficiency optical imaging equipment was selected,and the lighting method to highlight the welding characteristics was designed to complete the construction of the optical imaging experimental platform.(2)The feature matching method of multi-scale pyramid hierarchical search strategy is used to locate the welding area quickly and accurately,so as to solve the problem that the welding area is difficult to locate accurately due to the small offset in CCM detection.A multi-exposure image fusion algorithm based on exposure function and homomorphic filter function is proposed,which effectively reduce the high reflection of welding area and enhance the details,and finally the high quality welding area image is obtained.(3)To meet the high precision and efficient inspection requirements of CCM solder joints,a preliminary quality inspection by module and a secondary defect detection by weld joint are designed.Firstly,the preliminary quality inspection uses the image enhanced of the weld area to extract the contour geometric features of the solder joint,and quickly determines whether the module passes or fails by a minimum risk Bayesian decision method with adjustable risk function.Secondly,a Cbayes-Le Net neural network based on a Bayesian approach is designed for secondary defect detection to perform defect recognition on fused high-quality weld joint images.Comparing the Cbayes-Le Net method with a variety of traditional neural network models on the data set in this paper,the experimental results show that the Cbayes-Le Net method is more accurate and faster.(4)The experimental platform is built and the process design of the software module is presented,and experiments are carried out to verify the validity of the research methodology.The results show that the method of this paper can achieve better results in the detection of CCM welding defects,with an accuracy rate of 98.96%.The method in this paper effectively realizes the detection of bridge joint,leak welding and less welding defects in CCM welding,which provides an important reference value for accelerating the rapid development of CCM welding quality automation detection industry in China.