Study On The Visual Processing System Of LED Chip Sorter Based On Halcon

LED chip sort is an important part of the whole LED industry chain, and the visual processing module is an important part of separation equipment, which has a great influence on separation performance. The vision processing system of existing sort machine still has a lot of space to improve and promote. This paper's object is the vision processing module of sort machine, analyzing and improving the algorithm of the two core modules, template matching and defect detection. The main contents are as follows:The basic work flow of Led sort machine vision processing module is analyzed, and the desired speed and precision of the relevant performance indicators is proposed according to the characteristics of the fast speed, high precision, hardware limitations. Combined with matching and defect detection requirements, contrasting some current mainstream visual processing software package, selected the Halcon as a follow-up processing software package.Having analyzed grabbing error caused by center and angle deviation in the existing template selection algorithm, the automatic template region and angle correction algorithm is proposed, and the template is selected effectively. Analyzed the characteristics of LED chip matching, contrasted the gray and feature matching algorithm in the LED chip to identify the effect, compared the advantages and disadvantages, then proposed an double level matching scheme combined with the advantages of both, and according to the matching scheme, establishing the principle to select parameters.In the collection and reorganization of enormous images and data of defect chip, analysis the shortcomings and deficiencies of the existing classification methods of defects, putting forward three new system of defect classification. On this basis, designing and analyzing feature extraction and recognition algorithm of the most common and harmful shape defect in sort machine, and this algorithm is used to recognized defect chip, verifying the algorithm validly.For the double level matching and shape defect's testing algorithm, experiments of scanning positioning and defect detection were carried on respectively, verifying the algorithm in terms of speed, precision, recognition rate, false recognition rate and other performance indicators, and testing the validity of the algorithm. Through these two algorithms, improving the speed and accuracy of image recognition effectively in the LED chip sort process, and also improving the quality of identification chips, which lays a good foundation for subsequent chip sorting.