Industrial Automatic Control Systems and Controllers Annotation << Back A Survey of Diffusion-Generated Image Detection Methods based on Low-Level Feature Analysis Gut N.A., Loktev D.A. This paper provides a survey of modern methods for detecting images generated by diffusion models, with a focus on the analysis of low-level features. Three main directions are considered: frequency-domain analysis, identification of characteristic artifacts related to the architecture of generative models, and the use of the reverse diffusion process. The principles of these approaches are described, along with their applicability and robustness to post-processing and adaptive manipulations. A comparative evaluation is provided in terms of detection accuracy, computational complexity, and generalization to previously unseen data. It is noted that no universal solution currently exists; however, each approach is capable of identifying specifi c indicators of synthetic image origin. As a promising direction, the development of hybrid methods is outlined, aiming to combine the strengths of different strategies to enhance detection reliability under conditions of high visual realism in artificial content. Keywords: diffusion models, synthetic image detection, low-level features, frequency analysis, generative artifacts, reverse diffusion process, robustness to post-processing. DOI: 10.25791/asu.9.2025.1612 Pp. 45-51.