| With the increasing demand for social intelligence,the chip manufacturing industry has played an increasingly important role.In order to ensure the quality of chip production,more and more integrated chip manufacturing companies have adopted a variety of automatic detection technologies.Among them,the detection method based on machine vision technology has the advantages of strong real-time and high precision,and is gradually selected by more manufacturers.As a key part of machine vision technology,image processing technology plays a key role in detecting defects.The main purpose of this paper is to detect and classify the macro defects on IC chip packaging surface.This paper takes surface mounted chip as an example to simulate the possible defects in the chip packaging process and take the defect image.The research focuses on the extraction of defect location and numerical calculation,and the generation of defect database and defect classification rules.The main contents of this paper are as follows:(1)In the image preprocessing section,the enhanced contrast algorithm is added.By comparing the advantages and disadvantages of different algorithms,the median filtering algorithm is selected to filter the image,the multi-scale Retinex algorithm is used for illumination correction,and the Hough transform method is used for tilt correction.(2)The Otsu algorithm is optimized to improve defect extraction effect,filtering and selection of mathematical morphology algorithm for defect contour noise.Finally,five kinds of edge detection operators are selected for contour extraction,and the results are compared.Finally,the best edge detection effect of Canny detection operator is obtained.(3)The defect data is measured to obtain information such as the location,area,and perimeter of the defect.According to the characteristics of the defect data,it is divided into four types,and data samples of four types of defects are established to classify the unknown defect data. |
PDF Full Text Request