A system and method for authenticating the origin of visual content through spectral analysis and AI to distinguish real captures from screen reproductions and synthetic media, enhancing security, forensic reliability, and protection against misinformation.

The present invention relates to a system and method for authenticating the origin of visual content, including still images and video sequences, by means of spectral analysis combined with artificial intelligence. The invention addresses the technical challenge of distinguishing between three categories of media: (i) organic captures originating from real-world environments under natural or artificial light; (ii) reproductions of content displayed on electronic screens such as LCD, LED or OLED devices; and (iii) synthetic content generated by artificial intelligence, such as deepfakes or computer-rendered imagery.

The system acquires digital media through standard input devices, extracts luminance and chrominance matrices, and applies spectral decomposition techniques such as Fast Fourier Transform (FFT) or Wavelet Transform to identify frequency-domain patterns characteristic of each class. These patterns may include periodic flicker signals associated with display refresh rates, moiré or aliasing artifacts caused by pixel grids, or the spectral uniformity and absence of natural noise often found in synthetic renderings. The extracted features are processed by a machine learning classifier, such as a convolutional neural network or transformer model, trained on labeled datasets of organic, screen-based, and synthetic samples.

The output of the system is a probabilistic classification of the content’s origin, which may be labeled as “organic,” “screen reproduction,” “synthetic,” or “inconclusive.” To ensure reliability and integrity, the result is coupled with a cryptographic hash and timestamp, optionally anchored in a blockchain ledger for traceability and legal validity.

Applications of the invention include biometric authentication resistant to spoofing attacks, forensic validation of digital evidence, verification of media authenticity in journalism and social networks, protection of copyrighted material in streaming platforms, and monitoring of security camera footage to prevent tampering.