Research On Defect Detection Technology Of LED Welding Wire Based On Machine Vision

LED chip is an indispensable part of LED lamp because its welding quality determines the quality of LED lamp.In the process of production,welding may appear various welding defects,which affect the performance of LED lamp.It is of great significance to realize the defect detection of LED chip welding line for the control of production quality and improvement of process,and can improve the finished product rate and reduce the labor intensity of workers.The main research contents and innovations of this paper are as follows:(1)Based on the general process of machine vision measurement,combined with the LED chip detection types and image characteristics,the technical difficulties are analyzed,and the LED chip imaging scheme is designed,which provides a good data source for the subsequent defect detection.(2)This paper proposes an algorithm of LED chip location based on shape matching,which is realized by coarse to fine matching method.Firstly,the edge contour of LED chip image is extracted by Sobel detection operator,and the edge contour points are decomposed quantitatively,and then the image pyramid is used for hierarchical search to complete the coarse matching.Based on the data structure of KD tree,fine matching makes the benchmark image and target image point set correspond one by one by iterating the closest point.Compared with the existing algorithms,it can achieve LED chip positioning more accurately and faster.(3)This paper introduces a solder joint region segmentation algorithm based on adaptive local threshold segmentation algorithm and morphological operation.This method uses adaptive local threshold segmentation algorithm to segment the target and background,divide the ROI,and finally perform multiple morphological operations to segment the solder joint region,which can accurately and effectively extract the target region.In addition,a weld line recognition and detection algorithm combining Canny edge detection operator,image contrast and least square method is proposed.This method obtains the image edge line through Canny edge detection,sets the contrast of extraction line and edge line,extracts and filters the second derivative threshold,then uses the least square method to fit the line,and finally obtains the weld line through the expansion algorithm,It improves the recognition accuracy and has strong adaptability.Finally,based on the summary of the full text,the research focus and direction of LED chip defect detection algorithm are prospected.