Research On FPC Surface Defect Detection Based On Machine Vision

FPC has been used for precision control in various fields widely,making the quality inspection of FPC become an essential process.In order to improve efficiency,it is important theoretical significance and application value that carry out FPC surface detection based machine vision.In this thesis,the surface quality of the irregular precision FPC board with 25 × 20 mm surface is analyzed,and the detection accuracy is 0.1mm.The FPC surface automatic defect detection system is intended.The research contents are as follows:(1)The general design of irregular FPC surface defect detection system: according to the overall needs and functions of the system,and the needs of users,design process detection process and software frame structure.And based on the existing mechanical structure and mechanical motion control system,the hardware platform of visual inspection is designed to obtain high quality images.(2)Image preprocessing and image segmentation algorithm design: based on the collected image and its characteristics in the color channel,design the algorithm of transforming the color space for image feature extraction,so that the image features can be extracted completely.The ROI extraction algorithm based on edge detection and region growing is designed to solve the problem that the collected FPC and the background are difficult to divide.The ROI of the standard chip works as a mask template to extract the follow-up to be measured graph to simplify the algorithm.(3)Feature extraction and defect detection algorithm design: based on the characteristics of FPC with dense solder joints,the method of Blob region analysis is designed to extract the data features of Blob region and set the rules of defect region.Study defect recognition algorithm,and get the defective area features.The global defect detection algorithm is designed.BP neural network is used to classify the defects.Aiming at the recognition of the underlying defects,an improved image subtraction method is proposed to extract the defect features of the Blob region and simplify the algorithm.For the recognition of potential defects,the defective feature data are extracted from B channel of RGB color space and H component of HIS color space respectively for defect detection.(4)A software system for automatic recognition and classification of FPC surface defects is developed: based on the above research,the detection software architecture,flow and human-computer interaction interface are designed and the corresponding functional modules are designed.The software realizes the automatic recognition and classification of FPC surface defects and stores the results of defect detection and recognition.The software systems for FPC surface defects on the overall defect detection accuracy reaches to 90% or more.Detection accuracy meets the design requirements.The system has high accuracy in the detection of different defects on the surface of FPC,and has a great improvement in reliability and efficiency compared with the traditional detection methods.