Precision Micro Imaging Detection System For Ultrathin Flexible Substrate Of High Density Integrated Circuit

Flexible Integrated Circuit Substrate(FICS)is flexible printed circuit boards.Because of its thin thickness,good bending property and high density wiring,it can be used as the carrying material for connection between electronic components.With the development of electronic technology and the trend of light and thin electronic products,FICS also has been developing on the direction of miniaturization.As a key link in the manufacturing process of the manufacturer,the quality detection of FICS is still by eyes of people,which has the disadvantages of strong subjectivity and superposition of fatigue with growth working time.Therefore,using machine vision technology and algorithm with high speed and high precision to detect the quality of FICS automatically has become the certain trend of its development.Based on the detection problem of FICS,an image acquisition platform is built.For five kinds of defect characteristics of the two kinds of FICS,we do a lot of study and write the algorithm of defect recognition.The main work of this article is as follows:(1)In view of the overall requirements and functions of the system,an automatic detection system of flexible substrate based on micro-vision is designed.Based on the original mechanical structure and mechanical motion control system,aiming at the new system requirements,the automatic detection system is improved and designed from hardware and software system.The hardware includes the selection of the camera,the lens,the light source,and so on.The software includes the update of system parameter list,the addition of the functional modules and so on.(2)The algorithm of image preprocessing and image segmentation is designed.Based on the study on the captured image of FICS and its gray scale,the defect region is analyzed by our custom image preprocessing algorithm.If the defect region exists,it'll be segmented to be used for the feature calculation;if not,the following steps will be skipped and the warnings window will be popped up.(3)The feature extraction and feature classification recognition algorithm is designed.Based on the study of different types of defects of the two type of FICS,the available defect feature sets are determined.To do the segmentation on FICS input patterns with defect areas,and the feature values will be calculated after the segmentation.According to the classification and recognition rules set by the algorithm,the defect types will be identified.(4)In order to improve the recognition accuracy as much as possible,the idea based on matching a variety of features can be used to design the algorithm.For example,this paper uses the BP neural network classifier as the final optimization of the defect recognition classifier.This paper provides the design,algorithm implementation and technical reference of a quality detection system for the manufacturing process of FICS.