Forward Collision Warning system based on vehicle detection and tracking

2016 ◽  
Author(s):  
Zhipeng Di ◽  
Dongzhi He
Keyword(s):  
Warning System ◽  
Vehicle Detection ◽  
Collision Warning ◽  
Detection And Tracking ◽  
Vehicle Detection And Tracking ◽  
Forward Collision Warning

A Fast Forward Vehicle Detection Method by Multiple Vision Clues Fusion

2014 ◽  
Vol 543-547 ◽  
pp. 2647-2651
Author(s):  
Tai Qi Wu ◽  
Ye Zhang ◽  
Bin Bin Wang ◽  
Jia Heng Yu ◽  
De Wei Zhu
Keyword(s):  
Detection Method ◽  
System Development ◽  
Warning System ◽  
Vehicle Detection ◽  
Detection Methods ◽  
Vehicle Body ◽  
Detection Accuracy ◽  
Collision Warning ◽  
Moving Vehicle ◽  
Forward Collision Warning

With the development of intelligent vehicle technology, vehicle detection based on vision analysis has become an research hotspot in forward collision warning system development. Aiming to solve the existing problems in the current vehicle detection methods, for example, the detection accuracy is sensitive to the variation of illumination and object angle, we propose a forward moving vehicle detection method according to multiple vision clues fusion. Firstly, we locate the rough position using vehicle bottom shadow detection. The shadow is detected using an adaptive threshold image segmentation approach twice. Secondly, the symmetry of vehicle body and the perspective of camera field of view are both referenced to remove the inaccurate location in the first stage. The proposed method has been tested on several videos recorded in real urban conditions. Experimental results show that our method achieves 93.67% average detection accuracy in daytime, and its processing speed is more than 25fps. The proposed method has certain application prospects for improving the vision based forward collision warning system performance.

Sign in / Sign up

Export Citation Format

Share Document