Research Of Video Image Detection And Recognition In Contenary Critical Components

Catenary is an important part in traction power supply system of electrified railway. When the catenary malfunction occurs, the normal operation of traction power supply sys-tem will be affected, or even the traffic function of electrified railway will be interrupted. With the development of computer vision technology, those to be utilized in the high-speed rail inspection device. The requirements for image processing technology have to be im-proved continuously due to the complexity of the scene along the railway line, and the com-plicated of capture picture background. For an amount of video date acquired by the catenary device, image recognition is necessary to detect malfunction. At present, there isn’t research on the parts detection of catenary video image with the vast amounts of complex background and the intelligent recognition of abnormal circumstances. Therefore, the research of how to intelligent recognize the catenary video image efficiently is significant important to high-speed rail inspection technology.In this paper, for the railway video image data by catenary inspection device, an explo-ratory research of abnormal circumstances in catenary base on the image processing and pattern recognition methods. The main research work of this paper are as follows:Firstly, establish the image library of catenary key parts required for detection and iden-tification. Based on the inspection video data along railway line, the features of the catenary key parts, insulators and support devices are counted, summarized and analyzed. In order to provide experimental data for future research, an image library of catenary normal and abnormal image experiment image, and experimental sample image library of insulators are established.Secondly, complete the location of the support device. Based on the theory of target de-tection and identification, the positions of insulators are located by the classification and identification of the video image features with AdaBoost classifier. Then, according to the geometric relationship of the insulators, the position of support device is located and the ca-tenary key part area of interest is obtained automatically.Thirdly, implement the malfunction intelligent recognition of the key parts. Based on the digital image processing technology, the interference of illumination and thin wire of in-terest area are eliminated. The intelligent recognition method of catenary support device in-tegrity is proposed, and the selection of abnormal images of catenary key parts is imple-mented from an amount of image data.