Publication de 12 numéros par an
ISSN Imprimer: 0040-2508
ISSN En ligne: 1943-6009
Indexed in
GEOMETRICAL INVARIANT FEATURES PECULIAR FOR THE METHODS OF STRUCTURAL CLASSIFICATION OF IMAGES
RÉSUMÉ
New feature spaces modeled on the set of the coordinates of characteristic features of images have been studied. Geometrical invariant features are mapped a spatial arrangement of some single segments of a video-object. The results of development of modified methods that provide a reduction of computational consumptions through the description compression ensuring simultaneously high level of reliability have been given. Experimental results of image classification and analysis based on introduced methods have been discussed.
-
Rachkovskij Dmitri A., Representation of spatial objects by shift-equivariant similarity-preserving hypervectors, Neural Computing and Applications, 2022. Crossref