A method for multiple image interpolation employing self-organizing map

Morphing is one of the computer graphics operations that generate the intermediate image between images to express the changing from a shape to another shape in animation. As a typical example of real applications, the morphing operation can be used to produce a continuous sequence of sign languages from a set of sign-language-words. In general, to use morphing operation in such applications, it is needed to prepare all image changing paths that emerges in all possible sequence of states. To avoid such time-consuming preprocessings, we propose a novel method for image interpolation using Self-Organizing Map (SOM). The SOM can interpolate among data using their proximity relationships in data space by the topological mapping property. The proposed SOM-based image interpolation makes it possible to generate ungiven feasible paths from the corresponding points in each image unless any intermediate image among them are given. The effectiveness and possibilities of the proposed method are demonstrated through some toy problems.