Road Detection System Based On HSV Color Segmentation

In recent years, the development of driver assistance technology has become a major focus in the automotive industry, particularly in enhancing road detection systems with informative features about the driving environment. These systems aim to provide navigation and improve driving safety. It is crucial for these systems to accurately recognize and understand road environments, especially marked and unmarked roads. Therefore, the objective of this research is to detect marked and unmarked roads using HSV (Hue, Saturation, Value) color space segmentation, and to produce information such as the direction of the road (straight or turning), the vehicle's position based on lane position (Center Offset), and visual warnings when the vehicle changes lanes. This system uses a single camera sensor mounted on the car's front dashboard. The proposed main method in this project involves using HSV color space segmentation to handle variations in object colors caused by camera sensor limitations. Perspective transform is then applied to change the viewpoint of an object or image, making it appear as though viewed from a specific angle. This allows the system to detect both straight and curved roads. Sliding Window is used to identify parts of the road image with lane markings (such as white or yellow lines) and those without markings. The results of this research demonstrate the system's ability to identify lane markings on marked roads and to differentiate between road areas and non-road areas. The system achieved an accuracy of 90% in tests on marked roads in clear weather and 86% in tests on unmarked roads. This system is expected to provide a solution to assist drivers and has the potential to enhance mobility in restricted access driving scenarios.