COMPUTER VISION BASED LANE DETECTION SYSTEM USING 8MP WEBCAM CAMERA RESOLUTION 640X480 PIXEL

• Main Author: Hafizh Al-Fauzi, Ahmad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/41889
• Institution: Institut Teknologi Bandung

In this book, we will explain one of the subsystems on AssystDrive products, which is a lane detection system using computer vision. Lane detection system is very useful when cars pass through highways. The development of a lane detection system is a challenge because several factors faced by lane detection systems such as the unclear pattern of lane lines, perspective consequences, line lines that have low visibility, shadow, brightness and light reflection. Gradient and color thresholding is the central part of detecting line lines. Applying gradients and color thresholding to identify the lines of the image that have been converted into binary images first. Lane detection is obtained by using a sliding window search technique that visualizes the line lines on the road and uses least square method to get the coefficients on the line. The performance of the proposed path detection system is evaluated by video input and in realtime on the Pasteur-Cimahi Toll road at 2:00 p.m. to 3:00 p.m. with the condition of the line's visibility clearly visible, no thick shadows covering the lane line, brightness and light reflection still tolerated by the camera we use. The lane detection system can also work well at night at 8:00 p.m. to 9:00 p.m. due to the help of light from the car's lights. The sensitivity value is 90% and the average specificity is 97.16%.