Rich Models for Steganalysis of Digital Images

By Jessica J. Fridrich, Jan Kodovský. IEEE Transactions on Information Forensics and Security, 2012

1,994 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/043c62d35243e8aa7b8dfd98cf43367cfaf2482f

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

Citations in the corpus

Deep residual learning for image steganalysis (2018, methodology): “As a result, statistical descriptions to the filtered image become more compact and robust [8].”
Low-Complexity Features for JPEG Steganalysis Using Undecimated DCT (2015, methodology): “Figures 4, 5 and 6 show the detection error EOOB for J-UNIWARD [14], ternary-coded UED (Uniform Embedding Distortion) [12], and nsF5 [11] achieved using the proposed DCTR, the JPEG Rich Model (JRM) [20] of dimension 22,510, the 12,753-dimensional version of the Spatial Rich Model called SRMQ1 [10], ”
FaceForensics: A Large-scale Video Dataset for Forgery Detection in Human Faces (2018, methodology): “In the absence of compression, all methods, including [55] based on handcrafted features, achieve a relatively high performance.”
DeeperForensics-1.0: A Large-Scale Dataset for Real-World Face Forgery Detection (2020, background): “The benchmark includes six image-level face forgery detection baselines [6, 8, 12, 13, 15, 32].”
Multi-task Learning for Detecting and Segmenting Manipulated Facial Images and Videos (2019, methodology): “The noise-based method proposed by Fridrich and Kodovsky [12] is considered one of the best handcrafted detectors.”
Noiseprint: A CNN-Based Camera Model Fingerprint (2018, methodology): “This is the same path followed by the popular rich models, which were proposed originally in steganalysis [38], and then applied successfully in image forensics for the detection and localization of various types of manipulations [22], [26], [39], [40].”
Constrained Convolutional Neural Networks: A New Approach Towards General Purpose Image Manipulation Detection (2018, methodology): “Furthermore, we show that we can significantly improve the performance of our CNN-based detector by using very large amounts of training data that are computationally prohibitive for forensic detectors based on the SRM and its associated ensemble classifier.”
Autoencoder with recurrent neural networks for video forgery detection (2017, methodology): “Here, we use features inspired by work in steganalysis [33, 34].”
Deep learning for deepfakes creation and detection: A survey (2019, background): “and how the results are reached. Machine learning and AI algorithms can be used to support the determination of the authenticity of digital media and have obtained accurate and reliable results, e.g. [146]–[148], but most of these algorithms are unexplainable. This creates a huge hurdle for the appl”
MVSS-Net: Multi-View Multi-Scale Supervised Networks for Image Manipulation Detection (2021, background): “Zhou et al. [14] develop a two-stream Faster R-CNN, coined RGB-N, which takes as input the RGB image and its noise counterpart generated by the spatial rich model (SRM)[29].”
Learning Rich Features for Image Manipulation Detection (2018, methodology): “Recently, local noise features based methods, like the steganalysis rich model (SRM) [15], have shown promising performance in image forensics tasks.”
Thinking in Frequency: Face Forgery Detection by Mining Frequency-aware Clues (2020, methodology): “] have been proposed to detect the forged media. A series of earlier works relied on hand-crafted features e.g., local pattern analysis [21], noise variances evaluation [47] and steganalysis features [13,24] to discover forgery patterns and magnify faint discrepancy between real and forged images. D”
Steganalysis of least significant bit matching using multi-order differences (2014, background): “Fridrich proposed the Rich models on the basis of the noise component of image to train the ensemble classifiers, which could deal with the high-dimension feature vector well [13].”
A new cost function for spatial image steganography (2014, methodology): “And the performances are evaluated by steganalyzers using 34,671- dimensional feature set SRM [9], and 6,000-dimensional feature set MCNUQ [11], with ensemble classifiers [16].”
Media Forensics and DeepFakes: An Overview (2020, background): “Therefore, it is not by chance that the most popular such features, known as rich models [119], have been”