Lane detection based on spiking neural network and hough transform

In the field of the unmanned automobile and the automobile auxiliary driving system, the real-time and accurate detection of the lane is very important. Based on the previous research on the lane detection, the paper introduces the spiking neural network with the parallel mechanism to detect the lane. Firstly, the region of interests (ROI) is set on the origin image that collected by a vehicle on-board camera. In order to reduce processing time, areas outside the road are excluded in the ROI. Then the image preprocessing is applied to the ROI, including RGB to grayscale, gray stretch and median filtering to eliminate noise. Edge detection of the lane is the key to determine whether the Hough transform can detect the lane. In this paper, the spiking neural network is used to detect the edge of the lane. Finally, Hough transform is used to detect the lane. Experimental results show that this method is more accurate and robust than other methods.