ABSTRACT
With the advances in technologies, autonomous cars/self-driving cars are now-a-days gaining more demand due to the increment in mortality rate by road accidents caused due to human errors. Detecting obstacles on a road is one of the biggest challenges in autonomous vehicle/self-driving navigation system. In this paper, we have proposed the real-time smart lane detection and object detection system (SLODS) which captures the real-time road traffic using two cameras, one in the front and the other one at the back of the car. The front one detects the lane while the other one detects if any other vehicle is approaching while changing the lanes, ensuring safe lane change. Region of interest (ROI) determines object and lane detection. The performance of the edge detection algorithms like Roberts, Sobel, Prewitt’s, and Canny edge detectors, are evaluated based on precision, recall, F1 score, and peak signal to noise ratio (PSNR) values. For PSNR, Canny is outperforming others by the difference of -39dB with Sobel, -14 dB with Prewitt, and -48 dB. Further the proposed system also calculates the speed of the approaching vehicle.
