Modeling dynamic swarms

Modeling Dynamic Swarms

Bernard Ghanem and Narendra Ahuja
"Modeling Dynamic Swarms"
Journal of Computer Vision and Image Understanding (CVIU 2012) (arXiv e-print 2011 arXiv:1102.1292v1 [cs.CV])

Bernard Ghanem and Narendra Ahuja
Swarms, Dynamic textures, Crowd behavior analysis, Spatiotemporal analysis
This paper proposes the problem of modeling video sequences of dynamic swarms (DSs). We define a DS  as a large layout of stochastically repetitive spatial configurations of dynamic objects (swarm elements)  whose motions exhibit local spatiotemporal interdependency and stationarity, i.e., the motions are similar  in any small spatiotemporal neighborhood. Examples of DS abound in nature, e.g., herds of animals  and flocks of birds. To capture the local spatiotemporal properties of the DS, we present a probabilistic  model that learns both the spatial layout of swarm elements (based on low-level image segmentation)  and their joint dynamics that are modeled as linear transformations. To this end, a spatiotemporal neighborhood  is associated with each swarm element, in which local stationarity is enforced both spatially and  temporally. We assume that the prior on the swarm dynamics is distributed according to an MRF in both  space and time. Embedding this model in a MAP framework, we iterate between learning the spatial layout  of the swarm and its dynamics. We learn the swarm transformations using ICM, which iterates  between estimating these transformations and updating their distribution in the spatiotemporal neighborhoods.  We demonstrate the validity of our method by conducting experiments on real and synthetic  video sequences. Real sequences of birds, geese, robot swarms, and pedestrians evaluate the applicability  of our model to real world data.