Steganalysis using higher-order image statistics

By Siwei Lyu, Hany Farid. IEEE Transactions on Information Forensics and Security, 2006

464 citations recorded by the source corpus.

Techniques for information hiding (steganography) are becoming increasingly more sophisticated and widespread. With high-resolution digital images as carriers, detecting hidden messages is also becoming considerably more difficult. We describe a universal approach to steganalysis for detecting the presence of hidden messages embedded within digital images. We show that, within multiscale, multiorientation image decompositions (e.g., wavelets), first- and higher-order magnitude and phase statistics are relatively consistent across a broad range of images, but are disturbed by the presence of embedded hidden messages. We show the efficacy of our approach on a large collection of images, and on eight different steganographic embedding algorithms.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/a911b55b474c658105307893ec1ed9e120f9a62e

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

Digital image steganography: Survey and analysis of current methods (2010, unknown): “’’ Using the 2nd letter from each word the secret message reveals: ‘‘Pershing sails from NY June 1’’ [2,6,7].”
Steganalysis of LSB matching using differences between nonadjacent pixels (2016, methodology): “Lyu and Farid [23] decomposed the image with wavelet transformation, and calculated high-order magnitude and phase statistics as features.”
Image Quality Assessment for Fake Biometric Detection: Application to Iris, Fingerprint, and Face Recognition (2014, unknown): “steganalysis [26], [27] in the forensic field.”
Statistically undetectable jpeg steganography: dead ends challenges, and opportunities (2007, methodology): “Based on the comparison in [29], blind steganalysis constructed using this feature vector achieves performance that is one of the best compared to currently available blind steganalysis classifiers capable of detecting stego content in JPEG images [3, 2, 33, 1, 25, 10, 39, 9, 29, 12].”
Image forgery detection (2009, methodology): “In both cases, the statistical model is used to detect everything from basic image manipulations such as resizing and filtering [1] to discriminating photographic from computer-generated images [29] and detecting hidden messages (steganography) [30] .”
Region Duplication Detection Using Image Feature Matching (2010, background): “[35] can also be used to reveal image tampering.”
Benchmarking for Steganography (2008, background): “), and a functional assigning a single numerical value to the ROC curve (total minimal decision error [16, 38], probability of detection for fixed false alarm rate [26], or false alarm for probability of detection 50% [20], accuracy [13], etc.”
Low-Complexity Features for JPEG Steganalysis Using Undecimated DCT (2015, background): “Alternatively, the classifier can be trained that recognizes only cover images and marks all outliers as suspected stego images [26], [28].”