A GM-HMM based abnormal pedestrian behavior detection method

Detection of abnormal behavior is an important area of research in computer vision and is also driven by a wide of application domains, such as smart video surveillance. In this paper, we propose an algorithm applied in video surveillance for abnormal pedestrian behavior detection based on Motion-HOG and GM-HMM. The basic idea of our method is to put the features extracted into HMM to model the normal pedestrians´ pattern, while Motion-HOG has the advantage on extracting pedestrians´ motion features and GM-HMM can model the pattern well and truly. In our experiment, we compared different types of features and HMMs, the results indicate that the method we proposed had the highest accuracy up to 0.837, which demonstrated the effectiveness of the proposed approach.