An integrated forward collision warning system based on monocular vision

Abstract

Driving assistance system has a significant influence on driving safety, and we introduce an integrated Forward Collision Warning (FCW) system based on monocular vision. In order to reduce the searching region of original image, lane making is presented to establish the ROI firstly. Secondly, hypotheses are extracted using Haar-like feature and Adaboost classifier. To remove false positive detection in the hypothesis verification process, we utilize SVM-based classifier with HOG feature lastly. Using Time-to-collision (TTC), possible collisions trigger the warning, and such Forward Collision Warning(FCW) system has been evaluated in dynamic environment. Experimental results show that the proposed system is robust and useful in practical applications.