PARAMETRIC ESTIMATION OF MULTI-DIMENSIONAL HOMEOMORPHIC TRANSFORMATIONS

摘要

In object registration and recognition based on a set of known templates, the tremendous set of possible transformations that may relate the template and an observed signature makes any detection and recognition problem ill-defined unless this variability is taken into account. The present invention estimates the deformation that transforms some pre-chosen representation of an object (template) into the current observation. The method employs a set of non-linear operators to replace a high dimensional problem by an equivalent linear problem, expressed in terms of the unknown parameters of the transformation model. The solution is applicable to any homeomorphic transformation regardless of its magnitude. In the special case where the transformation is affine the solution is shown to be exact.