Lane detection system based on software and hardware codesign

Abstract

In this paper, we present a lane detection system (LDS) based on software and hardware codesign. In combining both hardware and software designs, it can achieve a real time lane detection within a processing time of less than 50ms. The hardware implemented by FPGA chip captures the lane image from CCD camera within a time less 10ms, which is faster than by software captures. In software design, the global edge detection is able to transfer the gray level image to binary pattern and show the edge of the object. Then, using this binary pattern find out the traffic lane location with following algorithm like the peak-finding and grouping, edge connecting, lane segment combination, lane boundaries selection. The lane departure warning algorithm detects the vehicle whether in traffic lane and judges whether sends out the warning. Experimental results demonstrate a quite good accuracy in lane detection whether at day or night condition.