Detection On Train's Cutout-Cock Fault And Cross-Bar Fault Based On Digital Image Processing

Running Trouble of Freight Car Detection System (TFDS) is a set of automatic security system that Ministry of Railways promotes. This system includes the automatic detection on Cutout-Cock fault and Cross-Bar fault. Cutout-Cock fault means the handle of semen which lies in the middle of railway carriages'subface is turned off, and Cross-Bar fault means the cross bar which lies in the ends of railway carriages'subface is bending. Once these two faults happen, train travel will be very dangerous.The whole work of this issue is to realize the automatic detect function of TFDS Cutout-Cock fault and Cross-Bar fault, which means to design and realize two sets of fault automatic detection algorithm in order to realize the detection of faults by Computer Vision not eyes which can improve the safety of train travel.Throughout the whole design and realization process, this issue flexibly uses a lot of classical algorithms of Digital Image Processing, such as, Grey-Level and Gradient Projection, Thresholding, Edge Detection based on gradient information, and Hough Transform. These two algorithms achieve the precise location and segmentation of the fault location and the accurate determination of the fault type. They achieve both 90% accuracy. Meanwhile, this issue uses Window Grey-Level Projection first time, this algorithm is improve on classical Grey-Level Projection algorithm of Image Processing, it can locate feature with higher accuracy and has certain value and meaning in some regions of image segmentation.At the end of the issue, two sets of fault automatic detection algorithm are examined, and their defects are described and analyzed in detail. At the same time, the speed of the algorithms is optimized from three parts which makes the algorithms more coincident with engineering requirements.