Using noise inconsistencies for blind image forensics

By Babak Mahdian, Stanislav Saic. Image and Vision Computing, 2009

428 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/e4a0a58a91b82a87c6b0228ae5352ef217a55cf8

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Generalizing Face Forgery Detection with High-frequency Features (2021, background): “Given that manipulation operations ruin the consistency of characteristics in the original image, there often leave distinctive traces in the noise space [17, 33].”
Fighting Fake News: Image Splice Detection via Learned Self-Consistency (2018, methodology): “We compared three baseline algorithms, all of which use classic image processing techniques to explicitly model and exploit imaging artifacts: Color Filter Array (CFA) [42] detects artifacts in color pattern interpolation; JPEG DCT [43] detects inconsistencies over JPEG coefficients; and Noise Varia”
SPAN: Spatial Pyramid Attention Network forImage Manipulation Localization (2020, methodology): “into xed resolution of 224x224, for optimal performance. Total run time for a 256x384 image is around 0.11 second. 4.3 Evaluation and Comparison We evaluate and compare SPAN with current SoTA methods [37,43,3,4,19,23,10] on the four benchmarks under dierent setups: (1) pre-training only; (2) pretrai”
Image Splicing Localization using a Multi-task Fully Convolutional Network (MFCN) (2017, background): “The ﬁrst class of splicing localization algorithms exploits noise patterns under the assumption that diﬀerent images have diﬀerent noise patterns as a result of a combination of diﬀerent camera makes/models, the capture parameters of each image, and post-processing techniques [26, 27, 9, 7, 8, 21].”
PSCC-Net: Progressive Spatio-Channel Correlation Network for Image Manipulation Detection and Localization (2021, methodology): “Our baseline IMDL methods include ELA [30], NOI1 [42], CFA1 [15], J-LSTM [3], H-LSTM [4], RGBN [69], ManTra-Net [65], and SPAN [22] where SPAN has reported the SOTA performance on localization.”
Splicebuster: A new blind image splicing detector (2015, background): “In alternative, detection and localization may rely on the different intensity and properties of the noise introduced in the image by different camera sensors [6], [7].”
Noiseprint: A CNN-Based Camera Model Fingerprint (2018, methodology): “From left to right: forged images, ground truth, heatmaps from the method proposed in [65] and [24].”
Information Forensics: An Overview of the First Decade (2013, methodology): “Unlike techniques that identify copy-move forgeries through block matching, SIFT-based approaches can identify duplicate image regions even if the duplicated region has been subject to scaling, rotation, an affine transform, or illumination adjustment [31], [32].”
Media Forensics and DeepFakes: An Overview (2020, methodology): “Local noise analysis has been proposed using statistical tools in the image domain [60], [61] or in the wavelet domain [62].”
Learning Rich Features for Image Manipulation Detection (2018, methodology): “• NOI1: A noise inconsistency based method using high pass wavelet coefficients to model local noise [24].”
ManTra-Net: Manipulation Tracing Network for Detection and Localization of Image Forgeries With Anomalous Features (2019, methodology): “…we compare ManTra-Net’s performance against numbers reported in [55], which provides scores of the classic unsupervised methods: ELA [29], NOI1 [34], CFA1 [22], and the latest DNN-based solutions, MFCN [39] and J-LSTM [7] on the four benchmark datasets, namely NIST 2016 [3], CASIA [2],…”
Hybrid LSTM and Encoder–Decoder Architecture for Detection of Image Forgeries (2019, methodology): “Some of the tamper localization techniques include DCT His-tograms [52], ADJPEG [13], NADJPEG [13], PatchMatch [24], Error level analysis [57], Block Features [49], and Noise Incon-sistencies [60].”