Method for morphing n-dimensional data using a scattered data transformation

Abstract

An improved “scattered data transformation” method for warping/morphing color data is provided to transform n-dimensional data so that only selected regions of the n-color space are modified. The method uses a set of “input points” that are chosen to represent the color gamut for an input color space. In addition, a set of “training points” having associated “training vectors” are chosen in the input color space, and the training points are used to define the transform output for the input points in the color input space. Some (or all) of the input points may also be training points, but it is not necessary that any of the input points be training points. The vector displacement effect of the preferred embodiment is both exponentially and inversely proportional to the distance of each of the input points to one or more of the training points so that, the closer a training point is to an input point, the greater the effect of that training point on that input point. If a training point is located beyond a pre-determined distance from an input point, then it will have no effect at all on that input point. A refinement of the present invention is to weight the vectors for the training points, so that certain training points can become much more “important” than others when modifying the input color space to the output color space.