Road Apperception Technology Research & Implementation Based On Monocular Vision

Modern management methods are needed to administer traffic on account of the economical development and the speedup of urbanization and motorization, which begets the research on Intelligent Transportation System-ITS. Driver Assistance System is important issues with applications to Intelligent Transportation System(ITS). Owing to its intrinsic Pre-Accident Warning mechanism, Driver Assistance System has bright prospects, especially in high quality active automobile safety and accident avoidance measurement. The Driver-Assistance System that employs Computer Vision technology is one of the booming research fields, due to their extensive signal detection range, integrity and excellent "Quality-Cost" ratio. Over the past decades of years, vision-based road apperceive algorithms has been used in driver assistance system. On-board automotive driver assistance system is aiming to alert a driver about driving environments, and other information. So the study of vision-based road apperception algorithms is very important to the vehicle assistant system.The research topic of the thesis is on-board monocular vision-based road apperception technology. It consists of three parts, in the first part, perspective projection model of 3D-road is established. Based on it, the original images can be transformed into bird-eye images, which can fit the lane location results based on bird-eye view, and provide the base for other algorithms such as vehicles,lines detection based IPM image. In the second part, Retinex algorithm is implemented for preprocessing of low-contrast gray images, and has been used in feature extraction methods to improve object detection results. In the last part, after summarizing the current road detection algorithms, an edge-based road detection algorithm is studied, and a chain-based line detection algorithm is improved, then a roadway-lines detection algorithm is designed and implemented, and detect the roadway region. Experiments show that this method could get a satisfactory result and is useful for the fusion of other road detection algorithms in LDWS.