A DSP-based lane recognition method for the lane departure warning system of smart vehicles

This paper describes a lane recognition method for the lane departure warning system of smart vehicles and the algorithm implemented in a dual core ADI-BF561 600 MHz DSP embedded system to verify the functionality. Since the computing power and memory size of the embedded system are not as good as those on a personal computer, special techniques have been applied in the algorithm to enhance the performance of lane recognition while maintain the reliability of the results. The applied median filter can obtain the median element of 4-by-3 array using only 19 comparison operations. Furthermore, the edge enhancing filter can washout the foreign objects in the region of interest and keeps the lane markings with tilt/slope pattern. Using these two tools as the basis, the applied algorithm can detect the event of lane-departure and alarms the warning to assist drivers for driving safety on the road. The system is implemented and tested on a DSP-based embedded platform which can perform the required processes in real-time under various weather conditions.