A Lane Detection Using Image Processing Technique for Two-Lane Road

Abstract

Lane detection and tracking technique are commonly used for a vehicle to navigate autonomously on the road. Various techniques have been developed by researchers and it seems image processing from vision sensors appears to be a popular approach. Hence, seeing the relevance of the technique, this research intends to develop the road lane detection technique which comprises OpenCV, Gaussian Blur, Masking, Canny Edge, and the Hough Transform methods. The technique was set to run using an embedded controller that is connected to a vision sensor. They were installed on the dashboard of the car to perform the detection of the two-lane road at different times. Several videos were recorded in real-time with 3-hour intervals starting at 10 am. During the recording, the technique analyzes and segmentizes the images from the video so that the white lanes on the road can be detected and tracked. To observe the performance of the technique, the images of the detected lane were converted to a histogram. Via the histogram value, it shows the best time to attain optimal performance of the lane detection technique. According to the outcomes of the experiment, it appears that at 1 pm., the technique works very well to perform the detection compared to other times. At present, we established a two-road lane detection and tracking technique that can be applied for autonomous navigation. However, there is still improvement that can be made to enhance the technique to carry out lane detection in the presence of shadows and perform at night.