Real-time vehicle detection using Haar-SURF mixed features and gentle AdaBoost classifier
Shujuan Sun, Xu Zhize, Xingang Wang, Huang Guan, Wenqi Wu, De Xu · 2015
On-road vehicle detection is one of the key techniques in intelligent driver systems and has been an active research area in the past years. Considering the high demand for real-time and robust vehicle detection method, a novel vehicle detection method has been proposed. This paper presents a real-time vehicle detection algorithm which uses cascade classifier and Gentle AdaBoost classifier with Haar-SURF mixed features. We built up a large database including vehicles and non-vehicles for training and testing. A pipeline is then presented to solve the detection problem. Firstly, lane detection is employed to reduce the search space to a ROI. Secondly, the cascade classifier is applied to generate some candidates. Finally, the single decision classifier evaluates the candidates and provides the target vehicle. The experiments and on-road tests prove it to be a real-time and robust algorithm. In addition, we demonstrate the effectiveness and practicability of the algorithm by porting it to an Android mobile.