Research About Railway Tracking And The Identification Of Railway Signal Light

Railway transportation is an important part of transportation in China. In recentyears, with the rapid development of science technology, train speed and traffic densityhave increased many times. There comes a constant concern on how to ensure the safetyof railway transportation. In order to reduce railway accidents, artificial intelligencetechnology which is based on image processing and computer vision has been graduallyapplied to railway transportation.This thesis has proposed a multi-template weighted based on the infrared imagehistogram similarity adaptive update the relevant tracking technology to track the rail.Infrared image which has strong spatial correlation is a grayscale image. If the similarityof adjacent grayscale distribution is high, the correlation of neighboring areas of graylevel histogram is high. We can use the similarity of infrared image gray histogram todetermine the trace of the target area. With the constant movement of the locomotive,the background environment is changing and the method of multi-template-weightedadaptive can update template. Effective template revised and updated, greatly improvingthe stability of the railway track.To improve the stability of railway tracking and the safety of railwaytransportation, this thesis has proposed a multi-template weighted based on the infraredimage histogram similarity adaptive update the relevant tracking technology to track therail. Since infrared image is a grayscale image with strong spatial correlation, thesimilarity of adjacent grayscale distribution is high. Therefore, the tracking target areacould be determined by the similarity of infrared image gray histogram. With theconstant movement of the locomotive, the background environment is changing at thesame time. The method of multi-template-weighted adaptive used in this paper couldadaptively modify and update the background environment template effectively, thusfurther improve the stability of the railway tracking.Considering current railway signal lamp is small, and it is easily interfered by otherlights and environments. And also the regional centers of railroad signals of color imageacquisition are saturated. All these factors will increase the difficulty of the recognition of railway signal light. To solve this situation, this paper studies the color space and itsconversion. We use the method of HSI color space threshold segmentation to split theedge of railroad signals and background environment, extract its contours. Based on thegeometry characteristics of the railway signal (circular), we take advantage of themethod of circular Hough transform detection to detect railway signal lights the circleon the binary contour image, improve the accuracy of identification, finally fills color ofthe railway signal on the original images, and combine the infrared image trackingtechnology to detect and identify left side lights of the railway.This paper proposes railway tracking technology and railroad signals recognitiontechnology based on image processing, using the camera in the front end of locomotiverailway to shot road ahead. The real-time display can help drivers to take preventivemeasures to reduce railway accidents rate when facing emergency, thereby increasingthe safety of the railway transport.