Research On Obstacle Detection For Railway Auto-detection System

With the increase of speed of train and quantity of passengers and freights, security and reliability are both more and more demanded. The traditional detection method with human-vision and routine check is less efficient at railway's detection and maintenance. The railway auto-detection system is an application urgently demanded by the department of railway now. The application of the system can overcome the disadvantage of the traditional detections, which are considered as production process and periodical overhaul, and change the human-control model to the computer combining with human-control model until the final implementation of the complete computer-control model.The main work of this paper can be summarized as follows: based on a presently hot research point, the Computer Vision, the obstacle likely to cause the damage of trains or rails can be automatically detected and the decision to examine and repair can be made, accordingly the clue about the abnormity of trains or rails can be present. The methods of digit image process based on the feature analysis and selection have been designed to realize the automatic detection system, in which the eyes' identification can be substituted by the computers'. It can consummate the system' automatic detection function as well as probe into the perfect application about the completely automatic detection.The framework of the system is a Server\Client distributed model. The communication between the server and clients is based on the network protocol TCP/IP. In the design of the machine vision, image filter arithmetic, image enhancement, edge detection of the Canny operator and image binarization are analyzed and improved to pretreat railway images; The pick-up method for the frame of railway is presented to form the image detection window; In order to make an assistant judgment whether an obstacle is existence, this paper adopts the method of analyzing the characteristic parameters of gray scale histogram.; The recognition arithmetic for railway and crosstie is also presented, which is used for exact judgment; At last, we carried out the test on samples and test the performance of the method. The experiments show the advantage of the approach.