Research On Fast Programming Solution For Automatic Optical Inspecion System Of Printed Circuit Board

Automatic Optical Inspection system (AOI) has been widely used in Surface MountTechnology (SMT) nowdays. However, the ability and efficiency of automatical programmingof AOI system is paid more attention, and become the problems urgently need to be solved inorder to get further development.In order to promote the efficiency of programming, program database image retrievalalgorithm is investigated and fast programming solution is promoted, based on AOI systemfor design and process rule check.(1) First of all, data structure of the program database is analyzed in order to discoursethe feasibility of realizing fast programming based on program database image retrieval.Model of image feature database is designed according to a red-green-blue-color illuminator.Model of program database image retrieval is established according to content based imageretrieval theory.(2) Method to position and gain component image is put forward, after the componentposition deviation is analyzed. Region divided method is promoted in order to outstanding theimportance of the ROI of the image, after the regionally distributed characteristics ofcomponent color is found out. The following experiments certify the effectiveness of themethod for positioning and gaining component image and gain reasonable weightscombination.(3) As the basis for the above work, image feature extraction and macthing algorithmsare investigated and region-color-feature based image retrieval algorithm is realized. In orderto improve color feature extraction algorithm, the color space is quantified and an improvedalgorithm based on threshold segmentation and weighted average is proposed replace of thetraditional clustering algorithm. The experiments followed prove the superiority of thealgorithm.(4) Finally, optimized solution of algorithm executing and data inquiries is put forwardbased on parallel computing and hash principle in order to improve the efficiency of imageretrieval. Then the component database data structure is improved to optimize the retrieval process. And both of these methods proved to be efficiency during the following experiment.(5) The fast programming solution based on program database image retrieval is tested inthe final experiment. Both program editing and debugging are included. And the experimentalresults show that the solution is able to improve the efficiency and automatical degree of AOIprogramming process.This issue comes from the investigation towards theory of automatic optical inpectionand human-computer interaction and technology of content based image retrieval. And it iscontribute to improve the performance of AOI system, by realizing fast programming. All thework has improved the ease of use and automation of the system and provided a reference tothe research in the future.