People number estimation in the crowded scenes using texture analysis based on gabor filter
A method for estimating the number of people in surveillance video is presented in this paper. The people number is calculated by a mathematical relationship between the global texture features of crowded scene and number of people in the scene. A set of well-established 2-D Gabor filters are used to extract the global texture features, which can effectively solved the problems of overlap among crowd members and perspective distortion. The LS-SVM method is utilized to learn the mathematical relationship mentioned above. The proposed method has been assessed and compared with other algorithms using the same PETS 2009 dataset. Experiment documents that the proposed method is more accurate and robust.