earticle

영어

Crowded pedestrian detection and density estimation are very useful and important under transportation environment. In this paper, we present a novel method for crowded pedestrian detection and density estimation through a weighting scheme of bag of visual words model which characterizes both the weight and the relative spatial arrangement aspects of visual words in depicting an image. Firstly, we analyze the visual words generation process. We give each visual word a weight by counting the number of images through which each visual word is clustered and computing the cluster radius of each visual word. To be more specifically, the co-occurrences of visual words are computed with respect to spatial predicates over a hierarchical spatial partitioning of an image. We validate this method using a challenging ground truth pedestrian dataset Pascal VOC 2007. Our approach is shown to be more accuracy than a non-weighting bag-of-visual-words one. The algorithm’s cost is also more efficient than the competing pairs.