Computational intelligence-based real-time lane departure warning system using Gabor features

Lane detection and lane departure warning are crucial parts of Advanced Driver Assistance Systems (ADAS), which is designed to increase general safety on the road. This paper proposes a novel approach for lane detection, which boosts the accuracy of lane departure warning system, specifically on highway and urban road under sunny condition, using Gabor Filter and other image processing algorithms. Gabor Filter is implemented to enhance the directionality of lane marking patterns. This filter automatically eliminates shadows, road marker and other non-related objects due to lack of directionality. Canny edge detection is then applied to extract the edges of lane marking and enhance the lane marking pattern. Lastly, Probabilistic Hough Transform (PHT) is applied to identify the correct left and right lane candidates on the road. We have also implemented lane departure warning to alert the driver when the vehicle is veered out of the lane. We showed that our framework is capable of realtime implementation using a Raspberry pi 3B to achieve 93% for lane detection and 95% for lane departure warning with 20 frames per second (fps) and only 75% Central Processing Unit (CPU) utilization.