Image Processing And Algorithm Research On LED Chip Electrode Positioning

Image analysis and processing is a key point in LED chip detection andsorting equipment. The purpose of this project is extracting parameters from LEDchip image, such as shape, position and posture etc, and creating control commandbased on them. n order to meet the accuracy, robustness and real-time response of themachine vision system requirements, choosing right image processing algorithm, isparticularly important. The major research content is as follows:The image pretreatment algorithm has been studied, auto-adapted median filteralgorithm based on the concurrent operation was improved, the image was processedto be more smooth using an adaptive median filter; Simultaneously, the image wasprocessed with distortion correction and the influence of radial direction distortionwas eliminated; The image gradation and shape characteristic of LED chip wasanalyzed.The SSD, SAD and NCC template match algorithms have been studied and thesearch space was optimized using a pyramid algorithm. The image was processedwith threshold value division and region connecting based on its gradationinformation and. A new match algorithm based on the edge direction pattern platehas been researched, and the identification and location of different LED chip andcan be differed based on moment match algorithm. According to LED chip shape, Anew match algorithm based on the edge direction pattern plate has been studied, Theobjective of differentiation of LED chips has been achieved by using HDevelopshape match operator, and the operator parameter was optimized.After the recognization and location of LED chips, the sub-pixel picture elementedge was detected using the Canny edge examination operator. The picture elementwas localized using the revolution axis law to the abnormity electrode, the straightline edge ROI region of abnormity electrode was searched according to edge’srelative position. Finally, the accurate location of circular electrode was achieved byusing constraint least square circle fitting method. VC++6.0was used as the software development platform, the applicationsoftware used for the LED examination and sorting equipment was developed. Thefunction of this application includes image acquisition, image display, LED chipsdifferentiation, and electrode localization. We have approved that the algorithmstudied in this project is reasonable, both processing precision and speed can meetthe requirement.