An optical flow based approach for motion and shape parameter estimation in computer vision

The authors introduce a dynamical systems approach to machine vision and describe an appropriate generalization of the framework well known in the literature on computer vision for the study of estimation problems based on optical flow. In particular, they show that the problem of motion and shape estimation can be described as an inverse problem associated with a pair of coupled Riccati partial differential equations. Two such pairs of equations, called shape-shading dynamics and shape-isointensity dynamics, have been introduced. A special case is considered for which the shape dynamics is an ordinary differential equation.<
>