Progressive contour models

A progressive contour model is developed based on the idea of deforming the contour from an initial shape as a source of prior knowledge by minimizing a defined contour energy to extract a desired contour from images. This model differs from active contour models (or snakes) in that the internal component of the contour energy is used to impose the smoothness constraints not on the shape of the contour but on the displacements of deformation, and the external component of the contour energy is used to locate the correspondence for the contour through a specified local correspondence mapping. A sequence of deformations is determined by repeatedly deforming and updating the initial contour. It is shown that the contour deformed by this sequence will smoothly and progressively approach a well-defined contour. Finite-element methods, multigrid algorithms, and unconstrained optimization methods are employed to implement this model. This approach offers several attractive advantages including a good convergence rate, the adaptation of the smoothness constraints and the adoption of a globally convergent algorithm. Experiments are conducted on real images to evaluate the performance of a progressive contour program, and a computational complexity in the order of O (lnN) is verified.