A real-time vision system for crowding monitoring

In this paper, the approach used to estimate people number for planning purposes in DIMUS (ESPRIT project P-5345) is described. Crowding estimation is based on the image-processing and inference phases applied to acquired data. Images come from a set of visual b/w camera oriented towards a zone to be monitored. Some significant features extracted from each acquired image are related to the number of people present in the monitored scene by using non-linear models obtained by means of Dynamic Programming in an off-line training phase. The present approach, employing previously obtained estimates, both improves accuracy of estimation, with respect to an evaluation based only on present available data, and can predict crowding values without new dala, between two successive acquisitions. Results obtained alter an extended test phase in a station of Genova's underground are reported.