Abstract:
A structure of an artificial neural network for assessing the similarity of a pair of images, comparing the quality of its
work with other criteria for image similarity are presented. Based on the proposed artificial neural network, a methodology for
estimating the adequacy of mathematical image model by the estimation of its similarity to real images has been developed. A
comparison of the estimation of the adequacy of the mathematical image model by the classical method and using the proposed
methodology on the example of a Gaussian mathematical image model has ben conducted.
Keywords:image modeling, model adequacy estimate, artificial neural networks, similarity estimate.