Particle-filter-based intelligent video surveillance system

Abstract

This paper designs an intelligent video surveillance system based on the particle filter. In the design, the adaptive Gaussian mixture model is applied to construct the background model. Utilizing the Gaussian mixture background model, the moving objects can be detected by background subtraction. For the moving objects appearing in the margin of the video frame, it is considered as a new unit (person). For the new considered unit, a new particle filter is established and designated to track the new unit. Once the tracked unit leaves the video frame, the corresponding particle filter will be terminated. Moreover, the Kalman filter is applied to track the units when they are occluded. By tracking the units in the video frame, we can obtain some important information, e.g. the number of persons in the area (or having been in the area), hot spots, etc.