Three dimensional image processing for rail profile analysis in railways

Depending on increase of speed of trains and frequency of use, security at railways and comfortable rail transport has importance each passing day. It is required to identify the impairment of road and rails to provide security and decrease operating cost and maintenance cost. Contact measurement techniques are used for error detection on railways tracks. Because low level of sensitivity and accuracy of the contact sensing systems do not meet the requirements of today's modern railway, the non-contact sensing techniques are used in conjunction with advanced technologies at the railways. In this thesis, the cameras and laser technologies are used, and a three-dimensional image processing algorithms are developed and measurement of non-contact rail profile is made. The images taken from rail is processed with the laser camera, and this guarantees detection of faulty regions, and whether or not there is fault with distance and coordinate data taken with laser camera. The distance obtained with laser technology and refraction of laser beams according to the shape of rail with coordinate data gains third dimension to the two dimensions obtained with camera. The depth knowledge of rail is obtained in this way. The noises for image in realized applications are applied Gaussian and median filtering techniques and are eliminated. For obtainment of beneficial data from images purified from noise, conversions like Watershed conversion apply Edge detection methods such as Roberts, Canny, Prewitt, and Boundary detection techniques such as Boundary Detection are applied. Depth data and image data are classified with intelligent computing techniques such as fuzzy logic and artificial neural networks. In this thesis, the image processing technique based ray profile analysis is made for non-contact monitoring of the rails despite the traditional methods of monitoring and control methods. When considering high spent costs for the measurement of rail line in today's climate, an initial step in making rail profile analysis with lower costs is provided in this thesis. This thesis provides the formation of knowledge about measurement systems in rail, guides dissertations and other studies, and at least partly reduces costs which are extremely expensive in railway measurement systems.