Study on defect spot recognition method in metal soldering based on intelligent artificial vision

摘要

During the process of defect spot recognition among metal solder joint, if the defect spot is responsible characteristic within small region, the traditional identification method of metal solder joint defect spot is based on sparse representation, unable to express details of characteristics in small region accurately. An optimized metal solder defect spot identification model is proposed, based on the similar triangle principle to derive the relationship between the defect spot depth and weld area, using back projection map of background subtraction graph and color histogram to detect the defect spot region, and convert RBG color space to HSV color space, the color histogram in HSV space is extracted, and the brightness values of the region meets requirement need to be modified, so as to obtain the back projection image of color histogram after processing, and two value image of defect spots detection, with the algorithm based on 7Hu moment vector, on the basis of the solder joint defect spot detection two value image, through acquiring contour of defect spot to match the template in the library, so as to achieve recognition of defect spot in metal solder joint.