Method of Probabilistic Identification of Images in Solving Problems of Applied Photogrammetry

In this paper, we propose an image identification method used in solving problems of calculating the number and geometric parameters of objects. Thus, for the analysis of geometric features in order to unambiguously identify the compared images, it is necessary and sufficient to simultaneously use the coefficients of the mean value, standard deviation, asymmetry, and kurtosis, splitting the original image into its constituent components. The proposed identification method is isomorphic with respect to the scaling and orientation of images. The use of a database for storing, along with the images themselves, and their numerical probabilistic characteristics, makes it possible to significantly speed up the process of identifying images when solving problems of applied photogrammetry. The method of obtaining high-resolution images proposed by the authors eliminates the need for such pre-processing operations as noise filtering and post-processing of images to improve their quality.