Description of images using a configuration equivalence relation

An approach to constructing a description of data and images based on the search for an optimal configuration (permutation) of their components (pixels, regions, feature vectors, etc.) is proposed. The quality criterion of the configuration, which may be selected in accordance with the application, determines the concept of optimal configuration. With specific configurations, the whole set of analyzed data / images is broken down into equivalent subclasses characterized by identical descriptors. Issues of invariant description, robustness of the proposed presentation, and the relationship of the proposed approach with the existing ones (Local Binary Patterns (LBP) and image representation by sign data) are considered. By way of illustration, an applied problem is solved using the proposed approach.