Research Of The Camera Module Defect Detection System Based On Machine Vision

The camera module is one of the most important parts of the smart phone. With the rapid development of the smart phone industry, demand for camera module increases. Higher pixel camera puts forward greater challenges on the detection technology. Now foreign manufacturers already have corresponding detection equipment. But the price is high and the technology is strictly confidential. Therefore, development of detection equipment with high accuracy is an inevitable trend. In this paper, around the specific needs of the camera module defect detection system, key technology of the visual system, image processing algorithm and system implementation are discussed.Firstly, this paper presents the design index for camera module defect detection device based on application requirements. In detail, hardware design of the system is discussed, including the visual system, motion system and mechanical system. Besides, analysis and selection of camera, lens and light source are carried out. Furthermore, achieved image of defect is compared with that by artificial detection, and the results meet the requirements of the system.Secondly, in order to reduce the complexity of the detection algorithm, the paper analyzes the characteristics of the captured image. Given the phenomenon of image offset during the detection process, template matching and rotation correction are conducted to locate and adjust the captured image. Furthermore, to solve the problem of image distortion and high time consumption, the thesis introduces the method of regional rotation angle. Concentrating on the grey varieties between particle in the wafer areas, several common threshold segmentation algorithms are analyzed. On this basis, the thesis puts forward the method of regional segmentation while gray scale conversion of the largest gold RGB componentis utilized on periphery gold line. Then, using Blob analysis to extract the object characteristic information of defects image after threshold segmentation. Experiments have been conducted to prove the feasibility of algorithms, and the result show that the Particle detection accuracy can reach to 3μ and all the defects of goldline can be detected, which meets the design index.Finally, based on the application requirements, the software framework of the system design and module partition is carried out, and UI of the system is built using Visual Studio. PLC is utilized to realize the operation logic of equipments. Stability and efficiency of the equipment are tested, and experimental results show that, related technology and implementation scheme of the system are feasible, which has great significance on accelerating the development of domestic camera module industry.