Based On The Key Technologies Of 3d Precision Measurement Of Non-ballasted Track Of Machine Vision Systems

The development of China's high-speed railway propose higher request to the line maintenance and repair work. At present, although there is a portable track detectors to evaluate track smooth degree,it only contents of the relatively track smooth,not including the line's absolute coordinates information. Therefore it can not be directly used to guide laying operations.The existing rail inspection instrument does not meet the needs of high-speed measurement in construction and after completion and acceptance. In this situation, we proposed the issues of developing a portable non-ballast track 3D detection system.Considering the crafts requirements of non-ballast track,The papers based on the original track inspector GJY-T4,increase the 3D coordinate measurement function. Compared the existing 3D measurement system usually,which uses total station to measure the tracks'external geometry parameters,the paper innovatively proposed the means of using optical photo grammetry,machine vision and digital image processing technology instead of total station to achieve the fast,accurate measurements of railways' three-dimensional coordinates.The measurement system constitutes of a pair of high-precision CCD, image acquisition card and a computer.Through shooting CPIII points, processing the collected image pairs picturing at the same time from different locations, combining CPⅢcoordinates, we calculate the absolute coordinates of measuring point.The essence of the paper:(1) the camera calibration algorithmThe method mentioned in this essay sets the target a assistant plane and establishes coordination on it. The process is divided into two steps. Compared with traditional camera calibration methods,the method mentioned here takes full account of radial and tangential distortion to improve the location accuracy. It is simple and fast to solve linear equations by using least squares in both two steps.(2) feature extraction and matching algorithmThe pyramid algorithm is applied in normalized cross-correlation template matching algorithm in the paper. This method greatly reduces computing time, guaranteeing the same accuracy and works well in tracking.(3) three-dimensional coordinates reconstruction algorithmIn accordance with measurement data of established track inspection car GJY-T4,CPⅢgeodetic coordinates,image measurements data,camera calibration results,the paper projecte a set of three-dimensional coordinate algorithm. The algorithm takes the advantages of designing sophisticated, calculating less, almost taking no CPU-time. It can meet the demand of real-time measurement.The three-dimensional coordinate measuring system takes the place of traditional system. It has features of high precision,fast measurement, on-line analysis and full-function. Research results can be used both in stages of construction and checking. It has a wide range of applications.