Study On The Techniques Of Surface Defect Detection For Board Strips

Quality control is a key step in the process of board strips production, and thesurface defects are the important factors affecting the quality of board strips. In order tocontrol and improve the quality of board strips, it is necessary to conduct surface defectdetection. In recent years, with the rapid development of computer and artificialintelligence techniques, the surface defect detection based on machine vision technologyhas become the focus of the current research. This paper mainly studies the techniques ofsurface defect detection for board strips based on image processing and patternrecognition techniques. The main contents are as follows:1. According to the basic requirements of surface defect detection system, theoverall design scheme of the system is proposed. And the hardware, software structuresare designed. The image acquisition program and image processing process are designed.2. According to the gray-scale characteristics of strip surface defect images, theimage preprocessing methods are studied. By studying and comparing the traditionalsegmentation methods, a new segmentation method is proposed. It is based on gray-scalemorphology and adaptive threshold. By enhancing the contrast of image and usingvariable threshold, the segmentation result of the proposed method is more accuratecompared with the traditional ones. Finally, the smallest rectangular area that contains thedefective part of board strip is located by using projection method.3. The common feature types of defect images are studied. And the featureextraction methods are introduced. Finally, a set of feature vector is determined andregarded as the input of defects identification classifier. The classification results showthe representativeness of the feature vector.4. By studying the pattern recognition theory and common identification methods, adefects identification classifier based on improved BP neural network is designed. The classifier is trained and tested with defects samples and the experiment results are ideal.Therefore, the detection techniques studied in the paper are feasible, effective and havesome practical value.