Study On Pantograph Porcelain Insulator Orientation And Defects Detection Algorithm

With great speed up in rail systems, people pay more and more attention on the safety of locomotives running. The pantograph porcelain insulators defects as an important part of locomotive inspection and received considerable attention. Nowadays it uses artificial method, the task is extremely arduous, low detection efficiency, the existence of security risks and other issues. Therefore, it is necessary to study a new detection method to enable accurate positioning porcelain vase, but also greatly improve working efficiency.The system is composed of image acquisition module, image detection module and retrieval control module. The image acquisition system uses two JAI industrial cameras symmetric photograph porcelain insulators'surface and back side images, by the photoelectric switch automatically trigger image acquisition program acquisition images. At the same time through image quality control, image format conversion and image pre-processing, change the acquisition images to the JPG format suitable for network transmission. And then sent the images to the image detection module, the detection program adopts digital image processing technology for image processing, recognition and send the process results to retrieve control module. The retrieval control module can realize retrieve relative locomotive's insulators'images and alarm notification according to the detection system records.In the part of software algorithms, this paper according to the pantograph porcelain insulators characteristics to determine a visual analysis combined with the gradient feature orientation algorithm. In the part of visual analysis, the visual characteristics of insulators are overall color changes uniformly. Through analysis a large number of insulator pictures, we can confirm the judgment basis which is composed of single factors, at last determine the location of insulators that may exist. In the part of gradient characteristics, the shape of the insulator is uniqueness, through image processing to calculate the gradient map:the insulator gradient is positive and negative changes alternately. Through the limitation of gradient, we can ascertain the porcelain insulators accurate central position. After that, through statistical analysis a large number of locate images data, using a rectangular box positioning the location of porcelain insulators, extracted pantograph porcelain insulators'images, and then determine the pollution and defection status of porcelain.Experiments and field application results show that:The system uses industrial camera system, relying on digital image processing technology, based on computer vision and gradient characteristics of the pantograph porcelain insulators detection algorithm, achieves non-contact automatic defect detection. In a word, this paper not only lays the foundation for pantograph porcelain insulator defects characteristics and defects extraction, but also has certain reference value in the design and implementation of pantograph porcelain insulators detection methods and orientation algorithm.