The Rectification And Merging Of Digital Image Of Railway Tracks

Now, static measurement of track linetype usually uses railway track detection device or manual measuring, which is time-consuming and hard sledding. Aiming at actuality of track static measurement technology, considering that precision requirement of orbit state is constantly improving after the speed up of Chinese railway, track linetype, which is an important indicator of track detection, is directly related to safety and comfort of train driving. It would have important actual meanings that a fast and efficient method for detecting the track linetype is studied.Recently, with the popularity of digital camera and constant improvement of digital image processing technology in accuracy, the applications of close range photogrammetry are more and more extensive. It will be worth to study that close range photogrammetry is used to detect the track linetype. The paper uses non-metric digital camera to gain image sequences of track orbit, and studies the methods of treating these images, so as to obtain the overlooking 2-dimensional railway image which can be used in the following track linetype detection. The investigation performed in this thesis and relevant conclusions are outlined as follows:1. Analyzes in detail the feasibility that using digital image detects the track linetype, and studies influence of errors in rectifying and jointing of image mosaic on detecting the track linetype. A formula between overlapping and global errors is given. The picture shows that global error of images jointed with 53% overlapping is the minimum.2. Image distortion is rectified by the method based on straight line characteristic, in which radial distortion parameter is calculated by relative variation of distortion error of the diffracted image and the reflective image taken as the cyclic conditions. The optimal distortion factor is gotten. The error calculated by linear fitting method is 0.16mm through tests, and distorted image is better rectified.3. According to the features of the railway image, its digital image is rectified by the method based on vanishing point geometry and geometric constrains of parallel object lines. The vanishing point and orientation elements may be calculated by the parallel lines in image, and then a model of geometric correction is obtained. The experiments show that this method is the same with rectification of railway digital image, the comparative experiments between practical measurements and image measurements show that their root-mean-square deviation in track orientation is about 0.3mm.4. Putting forward the mosaic method suitable for railway image rectified. The feature points are used to match rectified image, and a track mosaic is obtained by the seamless splicing. A mosaic of about 10m track is carried out in experiment. The whole precision of mosaic image may be analyzed; the maximum error, minimum error, and mean error in track orientation are calculated by fitting. The results in detecting track orientation show that the maximum error is 1.2mm, the minimum is 0.2mm, and the mean error is less than 0.7mm. There are some gaps between requirements and measurements in precision of detecting the linetype of 10m-bowstring.