Questions of image comparison in computer vision systems are considered. The measures which are taking into account distortion and false component occurrence in the structural description are suggested. The analysis of properties of methods is shown, results of computer modeling are described. It is first shown how formalize and apply metrical approaches for comparison of structural objects in associative space, the features of the known approaches are studied and the synthesis of likenesses is discussed with new properties.
structural methods of recognition, keypoints, similarity of sets, descriptors, components, a measure of affinity, the metrics, the analysis of conformity, the set of relations