Skip to content

Noiseprint

Definition

A CNN-based camera-model fingerprint extractor by Cozzolino and Verdoliva. Applied to deepfakes, it reveals inconsistency between the camera fingerprint in the genuine background and the absent or different fingerprint in the synthesised face region.

Method
CNN-based camera-model fingerprint extraction from noise residual
Detects
Copy-move splicing and deepfakes via fingerprint inconsistency
Inventors
Cozzolino and Verdoliva

Common questions

What does Noiseprint actually do?+

Noiseprint analyzes the noise residual in a digital image to extract a camera-model fingerprint. Every camera has a unique fingerprint from its sensor and processing pipeline. If parts of an image have fingerprints that don't match the dominant pattern, those regions are flagged as potentially manipulated.

How does Noiseprint detect deepfakes?+

In a deepfake, the synthesised face region typically lacks the camera fingerprint present in the genuine background. Noiseprint spots this inconsistency. The fake face either has no fingerprint or a different one than the rest of the image, signaling that it wasn't captured by the same camera.

Is Noiseprint reliable for copy-move detection?+

Noiseprint flags regions where the noise fingerprint breaks the dominant pattern, which is the key signature of copy-move splicing. However, its accuracy depends on the image quality, camera model coverage in the training data, and whether post-processing or compression has degraded the noise pattern.

Related terms

Affine Transform Estimation
The recovery of the geometric transformation (rotation, scaling, shear) relating a copied region to its source, using matched keypoint pairs. Geometrically consistent...
Block-Matching
A copy-move detection strategy that divides the image into overlapping fixed-size blocks, computes a compact feature vector per block, sorts vectors, and...
Chromatic Aberration
Colour fringing at high-contrast edges caused by differential refraction of light wavelengths by a lens. The pattern is lens-specific and inconsistent across...
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...
Copy-Move Forgery
An intra-image manipulation in which a patch copied from one location within the image is pasted over another location in the same...
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 Analysis
Detection approach that transforms image patches into the frequency domain (DCT or FFT) to expose periodic artifacts introduced by upsampling layers in...
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...
Localisation Map
The output of a forgery detection algorithm that marks, at pixel or block resolution, which regions of the image are identified as...

Explained in these topics

Your journey to becoming a forensic professional starts here.

Practice with mock tests, learn from structured notes, and get your questions answered by a global forensic community, all in one place.