Defect Detection Based On X-Ray Image Ofconveyor Belt

The defects detection is the foundation of mending conveyor belt.lt is also a guarantee for the safe using of conveyor belt.The traditional artificial detectionhas gradually beenreplaced byautomaticdetection with the development of computertechnology and machine vision. Defect detection is one of the mainstream application in digitalimageprocessing.lt consists of image denoising and enhancement, segmentation, and pattern recognition algorithm. But the current defect detection technology still not meet the requirements of practical application for high efficiency and accuracy, therefore,researching efficient and accurate detection algorithm for a specific application is necessary.The X-ray image of conveyor belt not only can present the surface defects, it also can show the internal defects in the conveyor belt. Therefore, defect detection based onX-ray imagenot only can save cost and improve the quality and efficiency of inspection,but also can detect the internal defect on the conveyor belt.In this thesis,the defects detection algorithmof the rapid and accurate is studied according to the actual needs of industrial production. In view of the defects of different types of diversity and distribution area,different algorithm is choosed to detect defect in this thesis. Before detection defect of belt,the local threshold segmentation method called Sauvola is used to binary the image.Then,the image of conveyor belt was divided into different areas according to the texture characteristics.Each image segment is very similar. In view of the main defects in the images of the conveyor belt area, this thesis uses the support vector machine (SVM) method to detect the defects of the rubber.In order to improve the detection speed, the method that the iterative learning is used to simplify the decision function of SVM, and the effective feature vector is selected to reduce dimension of feature vector.After optimizing these two aspects,the efficiency of the SVM to detect defects is improved greatly. For the wire defects.the defect area is got by examining two wire breaking point.Other defects in the image, threshold method is mainly used to detect them.It can be seen from the experimental results, this thesis accurately detect the various defects of conveyor belt, and through optimized the algorithm, to achieve the purpose of real-time detection.