Lane Detection Using Hybrid Hough Transform for ADAS

Lane Keep Assist (LKA) is a core feature of Advanced Driver Assistance Systems (ADAS) that helps maintain vehicle position within lane boundaries. Achieving reliable LKA performance depends on accurate lane detection across diverse road conditions. Techniques such as Standard Hough Transform, Probabilistic Hough Transform, and RANSAC have been widely explored for this purpose. In this paper, a hybrid vision-based method was developed to improve lane detection performance in ADAS. The proposed approach combines Standard and Probabilistic Hough Transforms to balance detection accuracy and processing speed. Gaussian filtering and Canny edge detection are applied to improve and enhance lane features within a defined region of interest. To improve direction classification, a slope-histogram technique classifies lane orientation as left, right, or straight. The system is designed to handle environmental challenges and is suitable for real-time use on embedded platforms. The proposed method is implemented on a Raspberry Pi and tested in simulated and real-road environments.