| Machine Vision System is a key part in all-in-one LED equipment, which plays an important role in its efficiency and accuracy. The main function of machine vision system is LED chip localization and chip duplicate removal. According to the deficiency of previous studies, below shows the main contents of this thesis.Analyze the statistical distribution of the pixel gray value of various regions in LED chips, and then present an intelligent template making method. Meanwhile, present an intelligent template selecting method based on the pixel number on the edges of the chip and the average and standard deviation of the pixel gray value of various regions. Finally, the experiments have verified the effectiveness of these methods.Analyze the feature of NCC, Shape and Shape +NCC template matching method and how to set the key parameters, and then design the LED chip localization scheme based on the requirements of all-in-one LED equipment.Analyze the existing duplicate removal method to find the reason of its low efficiency, and then improve this method. Firstly, save the information of LED chips based on their location which can remove the duplicate chips directly. In addition, give the same priority to the four orientation of searching lines which improve the efficiency very much. Finally, compare with the efficiency of the existing method to verify the method.The thesis presents the method to align the wafer on the worktable automatically and the experiments have verified the effectiveness of the method. Analyze the feature of polycrystalline LED chips and presented the method to detect the polycrystalline chips among the matched LED chips. The experiments have verified the effectiveness of the method. In order to improve the accuracy when picking the LED chips, present the scheme to compute the error between the position of chips and the center of the image and the experiments have verified the effectiveness of the scheme. |
PDF Full Text Request