The Research And Application Of Railway Fasteners Detection Based On 3D Imaging Technology

With the sixth speed-up of Chinese railways,the development of high-speed railways has entered a period of fast growth.As railway patrol inspection is an important guarantee of the safety of train operation,intelligent patrol inspection technology has been the focus and hotspot of the patrol system research.The technology aims at achieving the rapid and accurate identification of problems such as surface abrasion,defect,fracture and contour deformation that maybe exist in railway tracks,fastener clips,track beds and other targets.Based on the requirements of the intelligent inspection vehicle project proposed by China Academy of Railway Sciences Co.Ltd and the current development of optical imaging technology,this paper proposes a target detection method for railway equipment inspection using three-dimensional imaging technology on the basis of the line structured light,and concentrates on the core researches such as system design,system calibration and data processing relying on the detection of the fastener clips.The research results of the paper are as follows:(1)A hardware platform is constructed and an application software is designed and developed for three-dimensional image data acquisition.Based on laser triangulation method,a 3D image data acquisition hardware platform is built with a camera,a line laser and an electronically controlled sliding track as the core components.Based on Visual Studio 2010,the corresponding application software named LP-S2-3D is designed and developed to realize the functions of three-dimensional image data acquisition,three-dimensional image reconstruction,etc.(2)A calibration method is discussed for 3D image acquisition system based on the line structured light.With regard to the calibration of three-dimensional imaging system based on the line structured light,a sawtooth calibration method with high accuracy is proposed on the basis of the principle of line structured light triangulation.During the calibration process,the original method that extracts the center line of laser stripes by using Gauss fitting algorithm is improved with fusing the direction template algorithm.According to the characteristics of the sawtooth images,an improved piecewise linear fitting algorithm based on the least square method is used to extract the center line of the laser stripes accurately.Finally,the high-precision system calibration results are obtained.(3)A self-adaptive region-growing method is studied for the detection of clips.In order to realize the state detection of the clips,an image segmentation method based on self-adaptive region-growing method is proposed.Through analyzing the characteristics of the three-dimensional image data,a preprocessing procedure is required to improve the identification of the strip boundary.The research uses self-adaptive region-growing algorithm,combining the edge extraction algorithm based on Sobel operator and region-growing method,to automatically select the initial seed point.Besides,the appropriate region-growing standards with the characteristics of depth images are used to achieve the automatic segmentation and extraction of the three-dimensional images of clips.The research results have been applied preliminarily in Infrastructure Inspection Research Institute,China Academy of Railway Sciences Co.Ltd,which indicates that the proposed three-dimensional imaging detection method based on the line structured light has strong validity and good prospects in practical applications.