Intelligence Image Processing Based Railway Obstacle Detection On-Vehicle Technology

With the increase of quantity of passengers and freights, more non-occupation railway accidents have taken place. Railway obstacles severely affect safety of train operation. The method depending on drivers' vision and conventional detecting technologies is difficult to be efficiently applied to detection of railway obstacles.The obstacle detection method based on computer vision provides an important approach to solve the problem of railway obstacle detection. According to the characteristics of railway, this paper does some research which provide assistant-driving information to drivers on in-vehicle intelligent detection method of railway obstacle based on computer vision. It offers effective technological method to ensure safety of the railway operation.In this paper, by using the method based on analyzing video image sequence which is captured by a in-vehicle black and white vidicon, the author investigate railway obstacle detection technology. Firstly, video image sequence is preprocessed. The processing of smooth filtering of image can wipe off high frequency noise, the processing of image edge detection, image binaryzation and image morphology can extract edge information of railway. For narrowing search area of stationary obstacle, image window is established based on the edge information of railway.Secondly, the detection method of moving obstacle is presented. The method is based on optical flow field of the adjacent frames. Compensation for image dithering caused by vehicle vibration and vehicle running was achieved. Moving obstacle can be detected by optical flow field which is compensated.Finally, the detection method of stationary obstacle is presented. Stationary obstacle can be limited in a relatively narrow search area by using image window. In order to make a prejudgment whether an obstacle is existence, the author adopts the method of analyzing the characteristic parameters of gray scale histogram. The processing of Canny image edge detection and image binaryzation can obtain binary image. The binary image is projected to the horizontal and vertical directions. According to the result of projection, computing the weighted average to obtain the threshold value of projection curve. Based on the threshold value of projection curve of the horizontal and vertical directions, the upper and lower edge and. the left and right edge of stationary obstacle is derived. Then, the location of stationary obstacle is determined from the edge of obstacle. According to the test results, moving obstacle in the image window and stationary obstacle can be detected by the same method. The simulation experiments indicate that the method proposed in this paper is effective for railway obstacle detection.