| Summary: | 碩士 === 義守大學 === 工業工程與管理學系碩士班 === 95 === The rapid growing transportation has a critical affect on economic development. However, the growing volume of traffic cause large amount of traffic accidents and serious social loss. Consequently, driver assistance system (DAS) designed for safer automobile travel becomes universally sought. The research develops a prototype of DAS by utilizing machine vision techniques. Main function of the system is to find the collision avoidance path by obstacle detection and lane detection. In this paper, we propose a novel approach for obstacle detection in front of moving cars, using linear stereo vision. To speed up the processing time, the research transferred colored images to gray images. The gradient information is utilized to search the candidate features. Followed by a k-means-like method, the features can be extracted. To obtain the physical three-dimensional coordinates of the objects, the crucial problem is the so-called “correspondence problem” which consists in matching features in two stereo images. The research proposed a new matching algorithm according to the similarity matrix of the two images. We also address a modified edge detection method to quickly track the lane. Finally, a collision-free vehicle guiding path is determined. The system has been evaluated with experiments on outdoor road environments. Experimental results demonstrate the effectiveness of the proposed approach.
|
|---|
