文档介绍：Moving Target Classification and Tracking from Real-time Video
Alan J. Lipton Hironobu Fujpproach is one in which classification is based
ing. on simple rules which are largely independent of appear-
The resulting system robustly identifies targets of inter- ance or 3D models. Consequently, the classification metric
est, rejects background clutter, and continually tracks over which is explored in this paper, is based purely on a target’s
large distances and periods of time despite occlusions, ap- shape, and not on its image content.
pearance changes and cessation of target motion. Furthermore, the temporal component of video allows a
temporal consistency constraint [4] to be used in the classi-
fication approach. Multiple hypotheses of a target’s classifi-
1. Introduction cation can be maintained over time until the system is con-
fident that it can accurately classify the target. This allows
The increasing availability of video sensors and high per- the system to disambiguate targets in the case of occlusions
formance video processing hardware opens up exciting pos- or background clutter