Computer vision based collision detection system for ground vehicles

• Main Authors: Ching-Yu Cheng, 鄭景瑜
• Other Authors: Mau-Tsuen Yang
• Format: Others
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/33790990144499473615

碩士 === 國立東華大學 === 資訊工程學系 === 101 === Computer vision is mainly on the Intelligent Transportation Systems (ITS) to enhance transportation safety, so for collision detection systems in the visual are limited the ground vehicles must be in the simple environment, low-speed, and the direction of movement with the object non-parallel to the ground vehicle, but in an unknown environment, it is difficult to ensure that the information about the direction of movement, area, speed and distance of all objects, so it’s a very limited in the applications of the collision detection system, the visual collision detection system has not yet developed completely. In this thesis, the main concepts of computer vision to implement this system, optical flow (OF) and two kind of focus of expansion (FOE) generated by objects in the unknown environment to analyze and discuss the fixed objects and moving objects, and the full method of collision detection is published, it can avoid detecting miscarriage of the previous paper published, to resolve many traditional constraints, the various type or the size of the area of obstacles、the various relative position between ground vehicles and the obstacles、the parallel and non-parallel obstacles in the 3D space, and detecting in the complex background for example, reducing equipment costs and let the use level more popularization.