Pedestrian detection based on the combination of HOG and background subtraction method

Human detection plays an important role in Intelligent Video Surveillance. Real-time and accuracy are the two important capabilities in the human detection algorithm. The pedestrian detection method based on HOG feature draws many researchers´ attention for its excellent results. However, due to HOG-based detection method requires large amounts of computing, it is difficult to meet the actual demand in real-time applications. This paper proposes a pedestrian detection method that is a combination of HOG detection method and Background Subtraction detection algorithm and achieves the accuracy and real-time demand. Experimental results show correctness and effectiveness of the method proposed in this paper.