Study On Initiative Safety System Of Smart Vehicle

Aiming at the safety problem of vehicle driving, this thesis' research focuses on such aspects as structure, modeling, real-time information acquisition and anti-collision, vehicle initiative safety system as main object, using information processing technologies and smart agent technologies.At the aspect of vehicle and safety, problems related to the vehicles safety are summarized briefly. The intelligent traffic system and the safety system to anti- collision are discussed also. In addition, the research content of vehicle initiative safety system is analyzed.This paper presented the theoretical structure of vehicle initiative safety system based on steering behavior. Adopting the concurrence theory, the initiative safety system based on multitask and multithreading is proposed, which can perform the whole process from dynamic data acquiring in real-time to decision making of expert system by the cooperation of divided module.Based on computer vision theory, main algorithms of the technology are introduced. Combined with the research object, 2D-3D move estimate theory based on video frequency is presented. At last, the main applications of computer vision theory in the fields of ITS are briefly referred.Derived from the base theory of wavelet analysis, the application of this analysis in engineering and the fast algorithm are briefly presented. Because of the important value of odd detection of wavelet in image edge, combined with feature of this project, edge search algorithm based on Bubble wavelet is proposed. Using area expand and minimal dual-multiplication mainly, a kind of road detection and prediction is proposed, and a kink of road curve direction distinguishing method as well. Based on the theory of segmenting video frequency image, this paper discusses the method of dual-step segmentation: firstly estimating region briefly, secondly exactitude calculating position based on several elaborate algorithm. It figures out the target image from scene image effectively, and manage to segment the target obstacle. Single-eye distance detection algorithm based on structuralized road model is adopted in this paper. On the road modeling, according to road geometry restriction and industry standards, 3D road model is reconstructed and the distance of obstacle is calculated.