A clustering and quadratic programming based POCS algorithm for point matching

This paper proposes a new projection onto convex set (POCS) algorithm for enforcing two way constraints originated from point matching, which is based on clustering and quadrate programming. Via point clustering, the original POCS problem where the convex set is described by point correspondence' constraints is converted to the POCS problem where the convex set is described by cluster correspondence's constraints. As a result, a lower computational complexity is achieved. Then a numerical quadratic programmming (QP) technique is employed to solve the POCS problem, which, in practice, shows to be capable of achieving better performance than existing successive POCS (SPOCS) algorithm. Simulation results show that the algorithm has satifactory accuracy and computational save.