Aiming at problems that traditional rail wear detection systems based on image processing are complex and has large amount of computation and difficult to meet the demand of real-time detection

a two-dimensional high precision rail wear detection system is built by laser triangulation method. Rail profile is scanned by lightsectioning method profile coordinates are transmitted to industrial personal computer（ IPC） by ethernet; greyrelational median filtering is applied to preprocess raw profile data. Gauss-Newton nonlinear quadratic fitting based on least squares is conducted to extract coordinates of feature points on rail web center of circle and acquire rail wear with affine transformation

wear value of rail is obtained. Experimental results indicate that system detection precision of vertical and horizontal wear reaches ± 0. 2 mm

which is better than ± 0. 3 mm of traditional image detection method and meets rail standards requirements

the system can be used for real-time rail wear detection and guiding track maintenance work.