Research On Machine Vision Inspection Algorithm For High-speed Railway Track Surface Defect

Along with the High-speed Railway, heavy haul train, city subway and light rail increase, which require higher reliability and safety. The traditional railway track inspection depends on manual inspection, which has been unable to adapt to the current demands and the high-speed railway surface defect detection by machine vision technology is one development trend of railway maintenance. Based on machine vision technology, this thesis studies a high speed high-speed railway surface defect detection system, which detects the high-speed railway surface defects rapidly and accurately.Firstly, this thesis not only introduces the background and significance of high-speed railway track defect inspection, but also sketches the necessary of the detection and analysis of the current situation and developmental trend of the domestic and abroad high-speed railway surface defect detection.Second, machine vision system which is commonly used is researched. The design requirements and performance indicators of platform for high-speed railway surface defects is detailed, and the plan of platform is designed. The properties and requirements of lighting project, light source, camera, lens and storage scheme are analyzed and the type of these equipment are decided. The platform for high-speed railway surface defects which can get qualified images is confirmed by experiment.Thirdly, this thesis designed the algorithm of rapid detection of high-speed railway surface defects. The thesis also proposed overall process and steps of the high-speed railway surface area extraction, image preprocessing, regional positioning of defects, and OTSU based on the experience threshold value and threshold interval. The vertical projection method for regional positioning of defects and OTSU based on the experience threshold value and threshold interval are focused on the thesis. The experimental results verify that this algorithm is suitable for different image quality and could meet the requirements of detection system with high efficiency and real-time.Then, algorithm of fuzzy matching for redundancies of high-speed railway surface images is designed. The sources of redundancies of high-speed railway surface images and the necessity to remove these redundancies are proposed. At the beginning, the location information of defects is obtained through the external rectangle of defect. Then the morphological information of the defects is achieved by the perception of hashing algorithm. At last, the defect location information and morphological information are matched based on the improved fuzzy matching algorithm. The experimental results verify that this algorithm can effectively identify the redundancy information of image.Finally, based on machine vision, the thesis designs the high speed detection system software of high-speed railway surface defects. The thesis also summarizes system development and operation environment, and analyzes the software design target. This thesis introduces image acquisition software, defect detection software, image and defect information display software and thesis generation software.