Optimization Techniques on Pixel Neighborhood Graphs for Image Processing

A class of image processing problems is considered from the standpoint of treating them as those of co-ordinating the local image-dependent information and a priori smoothness constraints. Such a generalized problem is set as the formal problem of minimization of a separable objective function defined on an appropriate pixel neighborhood graph. For attaining a higher computation speed, the full pixel lattice is replaced by a succession of partial identical neighborhood trees. Two versions of a high-speed minimization procedure are proposed for, respectively, discretely defined and quadratic objective functions.