Research On High-precision Specular Highlight Removal Algorithm For Metal Rail

With the rapid development of China’s railway industry,the safety of rail transit is becoming increasingly important.By studying the wheel-rail contact attitude under high-speed and analyzing the status of wheel-rail disease,the safety of railway operation can be maintained.For wheel-rail contact measurement in motion,a non-contact measurement method is required.High-precision non-contact measurement usually uses structured light.Due to the high specular reflection characteristics of the metal wheel and rail and the influence of poor lighting conditions,the structure light on the metal wheel and rail will cause over-exposure and missing information,which greatly reduces the measurement accuracy.Therefore,this article aims to improve the image quality of metal wheel-rail structured light.The main research contents are as follows:(1)Aiming at the problem of over-exposed and under-exposed areas in images under uneven lighting conditions,an image adaptive exposure recovery algorithm based on exposure fusion model is proposed.Firstly,the weight matrix is calculated by the scene light map to weigh the proportion of well-exposed pixels.Secondly,a multi-exposure image is generated according to the camera response model,and finally the original image is fused with the multi-exposure image to obtain an enhanced image.Experiments have verified that the algorithm in this paper can adaptively recover the exposure of over-exposed and under-exposed pixels in the image,avoiding the problems of over-enhancement and color distortion of traditional image enhancement algorithms,and compared with other mainstream image enhancement algorithms,the algorithm in this paper performs better on image evaluation indicators such as PSNR and MSE.In addition,the algorithm in this paper only needs a single image to achieve exposure fusion,has high real-time performance,and can be applied to dynamic scenes.(2)Aiming at the problem of over-saturation of pixels in the image caused by high specular reflection on the metal wheel/rail surface,a specular reflection removal algorithm based on chromaticity space is proposed.Based on the two-color reflection model,the diffuse reflection chromaticity is calculated by the wheel-track structured light image,the pixels are clustered in the chromaticity space,the reflection intensity ratio of the clustered pixels is calculated,and the specular reflection component is calculated according to the reflection intensity ratio separate from.Experiments have verified that the algorithm in this paper can remove specular reflections in metal wheel and rail images with high precision,and compared with the results of other specular reflection removal algorithms,the PSNR value and SSIM value obtained by the algorithm processing results in this paper are higher,and the image distortion after removing the specular component is less.(3)In order to solve the problem of large area specular highlight area in metal wheel rail structured light image and the distortion area after removing the specular reflection,an image inpainting algorithm based on boundary pixel matching is proposed.Firstly,the repair priority of edge pixels in the area to be repaired is calculated,and similar pixel blocks are found in the adjacent area for filling repair The center line of the fringe is calculated,and similar blocks are searched along the direction of the center line of the fringe.This algorithm can effectively restore the distorted part of structured light fringe in the image,and improve the accuracy and efficiency by using the spatial structure information of structured light.