Multiplicity Matching And Error Processing Method Of High-Speed Railway Track Of Close-Range Photogrammetry

Close-range photography track detection technology uses no contact measurement,which has the advantages of fast measurement speed,simple operation and convenient data storage,and can significantly improve the efficiency of track static smoothness detection.However,this detection method is still in the exploration stage,and there are some problems to be solved.The track regional network of high-speed railway is usually distributed in a strip shape,with large measurement area span,poor reliability,and poor calculation accuracy of track measurement area with long mileage.In addition,the number of three degree homonyms points of the same track based on gray matching is small and there are many mismatches at present,and lacking of elimination methods for gross errors of track image homonyms points.In this paper,close-range photography track detection technology is optimized from three aspects: image homonym point matching,homonym point error elimination and adjustment.In the aspect of image homonyms point matching,on the basis of matching the two degree homonyms points in the images,further obtain the homonym points of multiple degree in the images;The method of coarse elimination and fine elimination are used to gradually eliminate the mismatching homonyms points,and step by step to obtain pure image connection points;The image connection point data obtained by this method participates in the self-checking beam adjustment adjustment with a distance of 120 meters and a distance of 240 meters.In order to improve the accuracy and reliability of orbit detection technology with close-range photography,the research work in this paper is as follows:(1)In this paper,a method of obtaining multiple degree homonyms points in the images is explored.Aiming at the problems of the repeated texture of railway track survey areas and sparse control points,grid is performed in the overlapping areas of the image,on the basis of using ORB algorithm to match a large number of two degree homonyms in the images,according to the information of track number,picture number and image point coordinate value of reference image,search and record on multiple reference images to obtain three degree homonyms points in the images of the same track and multiple degree homonyms points in the images of adjacent tracks.The experimental results show that the number of multiple degree homonyms points is large and evenly distributed,which enhances the connectivity between the orbit images and provides a good data base for the subsequent images adjustment calculations.(2)This paper introduces the error processing method of the homonyms point of the railway track close range image from coarse to fine.In view of the high similarity of railway track close range images,there are many mismatches of the same name points,removing of the homonyms points of duplicate recorded and pole firstly,large and rough points are deleted based on the image-object point matching deviation,select the contact points evenly according to the grid to participate in the adjustment calculations,and the circular points are selected based on the residuals and weights to obtain the pure contact points.The experimental results show that the mean square error of the adjustment solution after error processing is less than one-third of the pixel value,there is no gross error in the obtained contact point,and proving the error processing method of the homonyms points in this paper.(3)The adjustment experiment of self checking beam method is carried out.The distance adjustment of 120 meters and 240 meters track mileage is calculated by using the obtained contact points and other known data.The track edge line is fitted according to the coordinates of the rail top edge points obtained by the adjustment,and perform smoothness on the track gauge,level and super high are detected.The experimental results show that the accuracy of the check points of 240 meters mileage track solution is 1.3 mm,0.7mm and2.0 mm in the three directions of X,Y and Z,respectively,at the same time,the elevation fluctuation of the photography center is consistent with the elevation change trend of the existing control points,and the distance between the left and right orbit photography centers remains unchanged,which effectively verifies the reliability and applicability of the static smoothness detection results of the railway track.