Detection Of Catenary Equipment And Abnormal State Based On Monitoring Video

In recent years, with the gradual popularization of China’s high-speed railway, more and more attention is paid to the safety of the railway. In order to ensure the safety of railway traffic, the potential safety hazard must be eliminated in a timely manner. The security risks of catenary equipment, can only be ruled out by means of inspections. Traditional manual inspection methods have been unable to meet the requirements, because of its low efficiency, long cycle, high cost, the intelligent inspection methods of using image processing technology began to receive more attention. The intelligent inspection device of the catenary is installed in the cab of the train, complete image acquisition in the train travel, and realize real-time intelligent detection of the catenary equipment, as well as abnormal state detection. This way not only can liberate the human, but also to find security risks more timely, this has great significance for the safe operation of the high-speed rail.The research work of this paper was based on the technical specifications of the contact network security inspection device in 6C system. In this paper, the algorithm took the catenary inspection image as the experimental data, used image processing method to realize the intelligent recognition of equipment and abnormal state. Finally, the effectiveness of the proposed algorithm was demonstrated by experiments.The main work and innovation of this paper include the following aspects:In the image preprocessing stage, firstly, this paper used the MSRCR algorithm to carry out the fog to the image which was affected by fog. Then, this paper used Radon algorithm for angle detection of the inspection image, and used affine transformation to realize the image level correction. Finally, the test showed that the algorithm has good applicability.In the detection of catenary pillar, in order to reduce the interference and improve the speed of detection, the method to partition the image by the way of detecting vanishing points was researched. And then, the area of the catenary pillar in the motion foreground was extracted by using the method of analyzing the optical flow value of lattice. Finally, in order to realize the identification of the catenary device, the method for the integration of the structural characteristics of the catenary device with the idea of generalized Hough transform was proposed in this paper.In the detection of the nest on the catenary, the method based on HOG feature to detect nest of railway catenary was proposed in this paper. First, this paper extracted the samples of suspected nest and calculated HOG characteristics of these regions. Then, these characteristics were classified by SVM. Finally, the regions of nest were obtained automatically by analyzing the distribution of samples.In the end, the algorithm of this paper was tested by using the image of different terrain. The algorithm has good applicability, and it has a high recognition rate. It is proved that the algorithm has a certain engineering application value.