CYBERSECURITY OF OPTICAL IMAGES IN TASKS OF RECOGNIZING OBJECTS BY UNMANNED UNDERWATER VEHICLES

Abstract

The approach to cybersecurity in the information technology of optical image identification proposed in this paper involves the construction of subroutines for selecting objects in a scene image, creating reference images, and comparing the object with the reference. Object recognition tasks for unmanned underwater vehicles require new approaches and solutions. The purpose of the study was to optimize the identification of optical images in the tasks of recognizing objects of unmanned underwater vehicles. The research methods were established by optimization methods of identification in the tasks of recognizing objects of unmanned underwater vehicles.

The dual problem generates a combination of images, and the direct problem is related to the information approach to the identification problem. When using the information approach, the flow of information from the external environment Iw and the flow of information from the reference generator Iw* are considered, and the mutual information Iw/w* is used as a measure of their proximity, which defines the method as a method of compensating for information flows. One of the significant advantages of this method is the simplicity of assessing the system's performance based on the principle of compensation - the system is operational on the alphabet W under any disturbances if its reference generator can reproduce any image from W

The issues of optimizing the identification of optical images in the tasks of recognizing objects of unmanned underwater vehicles considered in this paper lead to the following conclusions: the criterion of minimum mutual information between the object in the scene and the reference is double to the criterion of minimum control costs of the reference. Thus, one of the most difficult tasks to be solved in ensuring the robustness of the information technology for identifying optical images to perturbations in the space of objects is the task of constructing an algorithm for generating reference images.