Studying On Image Feature For Detection Railway Track And Simultaneous Localization Calculation

Railway track is an important part of urban rail transportation and the basic equipment for safe operation of train.However,the spatial linearity of the railway deformed due to effects of the track bear the wheel-rail dynamic for a long time and roadbed settlement and so on which bring danger to train.Track geometry detection method of high precision and high efficiency is the basic guarantee for the safe operation of trains.At present,the detection method of space linearity of railway track is usually to use Track inspection vehicles and total station.This method is less efficient.This paper presents a method based on railway image detection and simultaneous localization calculation improve the non-contact detection efficiency of track.Designing image acquisition program.Building system of machine vision.Obtaining image of railway track.Processing and separating target area.Extracting and matching feature points in the target image region.Finding the relative relationship between the system coordinate system according to the matching point.Achieving Simultaneous localization calculation of track detection which provides the basic data for calculating the linear parameter of railway track.Putting forward a fast and high efficiency algorithm of detection of railway track for ballasted tracks.First of all,splitting out the target area in this method of Canny edge detection and cross projection.Proposing an algorithm of feature extraction of Block Adaptive Threshold based on the principle of Harris detection.First,blocking image and determining a threshold in using the method of iterative step.Finally,obtaining the feature points which are evenly distributed.Matching feature points using the method Combine Scale-Invariant Properties with Consistency Tests.Establishing the scale space of the image and the initial matching.Achieving the exact match point of the same name in the method of Consistency test.Researching on Method of simultaneous localization calculation of track detection on the basis of high-precision matching.Finding the coordinates of the matching point through the visual system.Finding parameters of the coordinate system of relative transformation according to the matching point of continuous image.At last,Achieving simultaneous localization calculation of visual system through the coordinate system iteration.Carrying out experiments of image processing in ballasted track test base in Shanghai Long Yang road.The results show that the proposed algorithm can locate and segment the fastener target region accurately for ballasted track.Comparing with the classical Harris operator,the segmentation adaptive threshold algorithm can extract the feature points more efficiently.The feature point extraction is not only fast but also has a more uniform distribution,which avoids the clustering of feature points effectively.Scale matching and consistency test matching algorithm can accurately and quickly achieve matching point of the same name of track image.Finally,by setting the parameters of the track.The matching points are calculated and analyzed using the machine vision principle.The results show that the error of the parameter information and the set parameters are within the specified range,which shows that the method based on image feature detection and synchronization positioning and has good potential for rapid detection of orbital space.