Комбінований метод виявлення підроблених зображень

Abstract

The paper presents the features of creating neural network technology for detecting fake images. The relevance of applying a combined method that combines color anomaly analysis and spectral analysis of images using fast Fourier transform (FFT) is justified. The principles of feature creation are described, including the use of RGB channels and their pairwise differences, as well as obtaining a frequency representation in the form of a normalized spectrum. By combining these approaches, images can be analyzed in both color and frequency space, which creates the conditions for improving the reliability of counterfeit detection. The results of the study can serve as a basis for further improvement of automatic analysis and monitoring systems for media content.