Frequency-Domain Analysis
Definition
Detection approach that transforms image patches into the frequency domain (DCT or FFT) to expose periodic artifacts introduced by upsampling layers in GAN architectures. GAN-generated images commonly show spectral peaks at intervals corresponding to the upsampling stride, which real camera images do not.
- Field
- Deepfake and GAN-image detection
- Transforms used
- DCT or FFT
- Signal exploited
- Spectral peaks from GAN upsampling stride
- Comparator
- Real camera images lack these periodic peaks
Common questions
Why do GAN-generated images produce spectral peaks that real photographs do not?+
GAN decoders build images through repeated upsampling layers such as transposed convolutions, and each upsampling step introduces a regular, periodic checkerboard-like pattern at a spatial frequency tied to the stride. A camera sensor's optical formation process has no equivalent periodic step, so its spectrum stays comparatively smooth.
How well does frequency-domain analysis hold up against image compression or resizing?+
Only moderately. JPEG compression and resizing can suppress or shift the characteristic peaks, which is why detection pipelines often combine spectral features with spatial-domain CNN features rather than relying on the frequency signature alone, especially for images that have been shared through social platforms.
Related terms
- CLIP-Based Detection
- Detection approach using OpenAI CLIP or similar vision-language foundation models as a feature extractor. The broad pre-training enables generalisation to generation methods...
- CNN Residual Detector
- A convolutional neural network trained on the high-frequency residual image, the difference between the original and a de-noised version, to classify whether...
- Detection Generalisation
- The capacity of a trained detector to correctly identify deepfakes produced by generators not seen during training. Low generalisation is the central...
- FaceForensics++
- A video dataset released by Rossler et al. (2019) containing 1000+ YouTube videos manipulated by four methods: DeepFakes, Face2Face, FaceSwap, and NeuralTextures....
- Frequency-Domain Artefact
- A periodic or statistical anomaly in the Fourier spectrum of an image or audio signal introduced by the generation pipeline's upsampling, filter,...
- Generalisation Gap
- The drop in detection accuracy when a classifier trained on one generation method is applied to a different method. Caused by learning...
- Noiseprint
- A CNN-based camera-model fingerprint extractor by Cozzolino and Verdoliva. Applied to deepfakes, it reveals inconsistency between the camera fingerprint in the genuine...
- Physiological Signal
- A biological process visible in video, such as eye blinking, rPPG (remote photoplethysmography), and head micro-motion from the cardiac cycle, that deepfake...
- Remote Photoplethysmography (rPPG)
- A technique that detects the pulse-driven skin-colour variation in a face video without contact sensors. In real video, this signal is present...
- rPPG
- Remote photoplethysmography. A technique for measuring heart rate from subtle periodic colour changes in facial skin caused by blood-volume pulses. Authentic video...
- XceptionNet
- A depthwise-separable convolutional architecture proposed by Rossler et al. as the baseline binary classifier in FaceForensics++. Trained to distinguish real from manipulated...