Using physics engines to track objects in images

This paper tackles the problem of tracking multiple articulated objects undergoing frequent contacts in a video sequence. Conventional tracking methods usually fail to distinguish objects during contact, implying the use of disambiguation techniques to recover the identity of each object. Moreover, such methods do not provide detailed shape information at the articulation level. We address these limitations by proposing a novel approach to track multiple articulated objects using the mean-shift technique and physics engines. By defining a model of each object using geometrical primitives and physical constraints, we exploit the physics engine force solver as a control layer of the mean-shift process and follow the model and its deformation along the sequence. The method is applied to track mice observed with a webcam and two sporozoites observed in reflection interference contrast microscopy, highlighting the flexibility and genericity of the framework.