Research Of Failure Recognition Algorithm For The Handle Bar And Draw Bar In The Bottom Of Freight Car

With the fast development of China’s railway trucking industry, the traditional trucksafety detection way of static and artificial can no longer meet the requirement, because ofits time-consuming and low efficiency. Therefore, the Ministry of Railways is promoting aset of Running Trouble of Freight Car Detection System (TFDS), to improve theefficiency and quality of the truck safety detection through the way of dynamic andmachine.Currently, TFDS is still in the operating mode of human-machine combination, thatmeans human identify the truck images taken by machine. To further improve thedetection efficiency and ensure the stability of the detection quality, and promote TFDSchanging into the operating mode of all-machine control, this paper designs and realizesthe recognition algorithm of the two typical failure of middleware in truck’ bottom, withusing digital image processing technology and combining the theory of computer visionand pattern recognition.In the course of the study of this paper, I use two different solutions to locate thepotential failure areas because the obtaining of the images of middleware in truck changes.One of them is based on template matching, which includes template matching based ongrayscale image, template matching based on edge image, and template matching basedon discrete point sampling image. The other location solution is based on the structuralfeatures of the target, which means using the digital image processing technology, such asgray projector, edge detection, region segmentation and so on, to locate the potentialfailure areas. In addition, this paper states the algorithm of how to determine the failure, tocomplete the whole identification process of the two truck failure.In the end of this paper, the effectiveness and efficiency of the recognition algorithmare proved to correspond to the target of the subject basically, by testing a large of truckimages. Meanwhile, this paper points out the deficiencies of the current algorithm and thedirection and ideas to improve its performance, according to the experience andexperience during the research.