Keywords

morphing, image registration, shape blending

Abstract

An algorithm is presented for morphing two images, often with little or no user interaction. For two similar images (such as different faces against a neutral background) the algorithm generally can create a pleasing morph completely automatically. The algorithm seeks the minimum work to deform one image into the other, where work is a function of the amount of warping and recoloration. A hierarchical method for finding a minimal work solution is invoked. Anchor point constraints are satisfied by imposing penalties on deformations that disobey these constraints. Good results can be obtained in less than ten seconds for 256 x 256 images.

Original Publication Citation

P. Gao and T.W. Sederberg, "A work minimization approach to image morphing," The Visual Computer, (1998) 14:39-4.