Robust multiple features improve pedestrian detection

Pedestrian detection is a key problem in computer vision with a number of applications including robotics, automotive and surveillance safety. Most of existing approaches use only feature for pedestrian detection. However, a single feature is not sufficient to represent objective content. In this paper, we present a novel pedestrian detection algorithms, the basic idea to design simple and computationally efficient features by means of a SVM ensemble. Therefore, we employ multi-features, such as HOG (Histograms of Oriented Gradients) and LBP (Local Binary Pattern), in this paper, we made improvement for HOG and LBP, so that no extra computational cost is needed with respect to a holistic method. It is observed from the experimental results on INRIA pedestrian datasets that our method has the high level of accuracy in the pictures.