Abstract
SMDs (Surface-Mounted Devices) are positioned or inspected using image processing methods, in chip mounters, used to place SMDs on PCBs (Printed-Circuit Boards), as well as in SMD inspection system. Currently, such methods require Part Shape Data made by time-consuming manual operation. Therefore, the demand for shortening the production lead-time grows by automation. In this paper, we propose a system for the hierarchical automatic classification of SMDs. We used Pixel Frequency, Edge Frequency and Circularity for feature extraction. We register the features of the image of the SMDs as reference data. We proposed the method for hierarchical classification. We classify the parts according to the relative histogram of circularity with the discriminant analysis as 1st classification. We classify the parts according to the rule base as 1st class segmentation. We classify the parts according to the DP (Dynamic Programming) distance, calculated using the reference data, with the discriminant analysis as 2nd classification. We evaluated this classification method with 715 parts, and obtained a classification rate of 95.9%.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have