Research On The Algorithm Of Precision Detection Of Surface Defects Of High Density Flexible Integrated Circuit Substrate

Flexible Integrated Circuit Substrate(FICS),as one of the printed wire subgrade boards,is widely used in aerospace,electronic and information industries.Due to the complexity of the processing process,the deviation of any process will lead to different types of defects.With the high precision and miniaturization of industrial production,the process of quality inspection needs to be updated,from manual visual inspection to image vision processing,improve detection efficiency,and constantly adapt to new products.In this paper,aiming at FICS surface discoloration and etching,a visual detection algorithm is designed based on the images collected by the self-developed FICS defect detection system for the characteristics of geometric and structural distribution of defects.The main research contents of this paper are as follows:(1)For the problem of serious noise interference in the collected images of high-precision flexible substrate,an adaptive noise reduction filtering method based on improved non-local mean value is proposed.Experimental results show that,compared with the conventional spatial domain based filtering methods,the proposed method can preserve the edge details while removing more noise from the image of flexible substrate.(2)Due to defects such as oxidation and etching,some defects are hidden in the dark area of the background,which cannot be observed and found by human eyes.A local low-illumination image enhancem-ent algorithm is proposed,which can improve the contrast of the local dark area of the image without causing correction problems such as blurring,chromaticity and overexposure.(3)On the surface of the FICS uneven discoloration defect problems,in order to avoid by color and texture of the impact of substrate itself,this paper proposes a fast segmentation based on improved ultra pixels change color defect detection algorithm,the algorithm based on segmentation after super pixel multi-scale feature extracting and analysis,accurate positioning and segmentation target defects.(4)In view of the complex etching defect types of flexible substrate,the research target was subprobl-ematic.According to the structural characteristics of etching on the edge of the line,a contour curvature defect detection algorithm based on differential geometry theory was proposed to identify the etching defect of the line by looking for the extreme points of the discrete curvature of the contour and screening them.The defect area is located by the prior knowledge backprojection for the underetched defects.Experimental results show that the algorithm has high detection accuracy and can effectively improve the detection rate.This paper studies the algorithm of discoloration and etching defects for FICS quality inspection,which has certain technical reference value for realizing automatic and efficient detection of FICS appearance defects.