Image Segmentation And Circle Detection In Visual Inspection Of High-density Flexible Integrated Circuit Substrate

Flexible Integrated Circuit Substrate(FICS)is a special component that provides a connection for integrated circuits.With the development of high integration of integrated circuits,FICS has gradually developed toward high density.At present,the line width,wiring distance and drilling size of FICS have reached the micron level.In FICS production and manufacturing process,quality inspection is a key step.At present,this step is mainly completed by manual visual inspection.However,with the increase of FICS production and the denser wiring density,inefficient human visual inspection methods have gradually failed to meet the detection requirements in actual production.The use of automatic optical inspection(AOI)is an effective means to improve the detection efficiency.In this paper,based on the detection problem of high-density flexible substrates,an automatic optical microscopy detection system is set up,and the image segmentation and circular drilling detection technologies in the detection algorithms are studied.The main work of the article is as follows:(1)For the detection of high-density flexible substrates,an optical microscopic inspection system that satisfies the requirements of high precision and high efficiency is built.The hardware of the system equipment is selected,and a reasonable software system framework is designed.(2)An SVM image segmentation method based on fuzzy membership information is proposed.The method uses the fuzzy membership information of image pixels and spatial structure information to construct the feature vector of the pixel,and uses SVM model to classify the pixels so as to complete the FICS image segmentation.This method has a good segmentation effect.(3)A fast threshold image segmentation method based on gray histogram is proposed.According to the gray histogram shape feature of FICS image,the method can find the appropriate threshold by searching the gray range of the image copper foil and the background,and can quickly and accurately segment the FICS image.(4)A fast circle detection algorithm based on contour extraction and validity determination is designed.First,the threshold image segmentation method proposed in this paper is used to segment the copper region of FICS,and then the contour of copper is extracted.Then the candidate set of circles is constructed by using the roundness information of the contours.Then the validity of the candidate circle is determined according to the Helmholtz principle.Finally,an iterative least squares fitting method is used.Calculate the effective circle parameters.The circle detection algorithm has the advantages of high speed,high precision,and low error detection rate,which is of great significance for real-time detection of FICS.This article provides a reference for the design,method and technology of the inspection system for the manufacturing process of high-density flexible substrates.