Research And Implementation For Lane Detection Driving Assisted System Base On SoPC

Lane detection is one of the core technologies in the car driver assistance.With the fast development of intelligent transportation, the research has been widely preceding. But most of the detection methods are based on complex algorithm and great performance of the PC platform which will lead to high cost, large size and limitation of applying range. If the real-time and robustness based on the embedded platform can be realized to satisfy the engineering applications driver assistance system, its usage will undoubtedly increase. However, the performance of the embedded system is lower than the PC system relatively. So how to ensure the accuracy and real-time under the condition of low performance is the key research in this paper.In the respect of video image processing, the paper analysis of existing image processing technology, combined with the features of road images, and to compare various types of classical algorithm, a new algorithm for lane detection with panorama visual which based on color space Collision detection algorithm for the SOPC platform is proposed.Experimental results show that the potential to reach13-15frame/second and97%accuracy with the structure of the road lane markings, and simple background. The algorithm is fast, stable and accurate, while the implementation on SOPC platform makes portable and reusable which meet the engineering requirements well.The innovation in the process of detection algorithm include:panoramic vision of the collision detection algorithm which adapt to the embedded platform based on the application of panorama visual images is proposed, and implementation it based on SOPC platform. These work laid the foundation for the subsequent design of lane departure warning system.At the end of the paper, summary and conclusion of subject involved the deficiencies of the system and follow-up can be improved is put forward.