Color based multiple people tracking

In this paper, we address the probleni of tmcking multiple people in a cluttered environment had on thorough usage of color infomintion. We introduce color invariants to the proposed conihineci color spxtial model, which combines the low-level arid nlid<ilelevel color spatial infomiation togetlier. This formalization gives ils a way to coiiibine low-level ~ n d middle-level color spatial distribution to I,iiild t i person model. In multiple targets cluttered environment, color invariant and color sjmtial information play an important role in robust surveillance s)steni. We propme a moditied mean shift nlgorithm tn track multiple people based on color invariants niid color spatial model. The interactions of multiple people have also been studied in order to guide the t r ih ing under occlusion.