The Segmentation Of Sleeper In Railway’s Images Based On Gabor Transform

Rail transportation is an important part of Transportation Mode in the world. In recent years, with the increasing investment on the urban light rail transit and high-speed rail passenger line, and the significantly raise of rail transit status, its security issues have become increasingly prominent. As a critical infrastructure of the railway line, the working state of sleeper has an important impact on the safety of the railway line. With the development of the non-contact detection technology based on image processing technology, it has been gradually applied to the detection of the state of the railway line. The purpose of this paper is to provide a precise and high reliable segmented image of sleeper for the subsequent detection of sleepers. To achieve this goal, we mainly do the following three aspects of the work:(1) The Current Situation of the sleeper segmentation in the railway image at home and abroad was introduced, and the necessity and importance of the research of sleeper segmentation in the railway image were analyzed combined with the existing railway line inspection method;(2) A nonlinear method of extraction texture features based on Gabor filters transform was proposed according to the characteristics of the railway image and combined with biological research, the relevant parameters of the Gabor filter bank are designed, then the direction selectivity-. rotation invariance、translation invariance and scale invariance of Gabor filters were discussed;(3) Nonlinear analysis was adopted to extract and analyze texture features of the railway image based on the Gabor filters transform. Split the cluster center selection rules was made according to the shortcomings of the classic K-means clustering segmentation random to determine the cluster center and can’t use the pixel space constraints, then space constrained K-means clustering segmentation method for sleeper was proposed. Experimental results show that the improved K-means clustering segmentation algorithm for sleeper in this paper is outperform the classic K-means segmentation algorithm, segmentation accuracy rate is higher, it has a certain application value.