摘要 |
FIELD: information technology. ^ SUBSTANCE: method involves converting an image to an equal number of rows and columns; selecting feature space for the first and second images; generating feature matrices, each element of which is a vector of pixel feature values; a plurality of elements is selected from the feature matrix of the first image; for each element, a plurality of elements is selected from the feature matrix of the second image; images are formed, which are matrices of scalar elements; a matrices of estimate values are formed; the matrices are approximated; the spatial shift of the second image relative the first image is determined. ^ EFFECT: high accuracy of determining spatial shift of images of the same scene. ^ 5 cl, 12 dwg |