Active-meshes

This paper describes the implementation of an 'active-mesh' that is to be a utomatically created and configured directly from a single frame of an imag e sequence. The aim of this approach is to use the derived mesh to perform visual tracking in unconstrained motion environments, allowing movement of the camera, the scene and even the inclusion of background-independent movi ng objects. The main problem in initializing this mesh comes from the fact that there is little a priori information about the scene available. The pa per will discuss methods that are currently available for determining the i nitial position of active contour models within images, also suggesting a m ethod of initializing the active-mesh. The approach is further extended to using multiple meshes and region initialized meshes. (C) 2000 Elsevier Scie nce B.V. All rights reserved.