Laser Spot Center Location of Track Measuring Device

Abstract

On the basis of the existing rail laser measuring device, the paper proposes a method of laser spot center calculation that is suit for the in-site survey base on the common algorithms’ comparison. And the morphological cascade filter and Gaussian filter are constructed in the denoising and smoothing pre-processing of the laser spot image. In order to make full use of the intensity distribution of laser spot, the threshold range is determined by the Otsu algorithm and maximum light intensity, which is non-uniformly segmented by exponential function to obtain a series of binary spot images. After the center of light spot is initially determined by the centroid method, a series of center samples are obtained by Hough transform on the edge extraction image, and then the clustering center is obtained by K-means algorithm, at last the actual coordinate of the laser spot center is obtained through the relationship of the coordinate functions fitted by BP neural network. The proposed algorithm is tested on the coordinate screen of the track trolley, and its error can meet the requirements of the in-site survey.