A COST-EFFECTIVE PEOPLE-COUNTER FOR PASSING THROUGH A GATE BASED ON IMAGE PROCESSING

作者：Chen Tsong Yi^*; Chen Thou Ho; Wang Da Jinn

来源：International Journal of Innovative Computing Information and Control, 2009, 5(3): 785-800.

• People counter
• Color image processing
• Object segmentation
• Object tracking

Based on area estimation and color analysis, this paper proposes a cost-effective bi-directional people counter for the pedestrian flow passing through a gate or a door. Firstly, the passing people are roughly counted by the area of people image captured by a zenithal video camera. To refine the initial count, the pedestrians will be recognized by tracking their image patterns with intersection of their bounding-boxes or color vector matching. Besides, both inherent problems of image pattern overlapping and merge/split phenomenon will be also overcome. Experimental results show that 100% accuracy of bi-directional counting can be achieved if the pedestrians walk in a normal speed and the people number of an image pattern is less titan three. When compared to other reported methods of using a zenithal camera setting scheme, the proposed method can obtain higher count accuracy over 95%.

• 出版日期2009-3