New feature presentation of transition probability matrix for image tampering detection

Authors:
Luyi Chen;Shilin Wang;Shenghong Li;Jianhua Li
Affiliations:
Department of Electrical Engineering, Shanghai Jiaotong University, Shanghai, China;School of Information Security, Shanghai Jiaotong University, Shanghai, China;Department of Electrical Engineering, Shanghai Jiaotong University, Shanghai, China;Department of Electrical Engineering, Shanghai Jiaotong University, Shanghai, China
Venue:
IWDW'11 Proceedings of the 10th international conference on Digital-Forensics and Watermarking
Year:
2011

Citing 6
Cited 0

Sum and Difference Histograms for Texture Classification

IEEE Transactions on Pattern Analysis and Machine Intelligence
A natural image model approach to splicing detection

Proceedings of the 9th workshop on Multimedia & security
A Markov process based approach to effective attacking JPEG steganography

IH'06 Proceedings of the 8th international conference on Information hiding
Effective image splicing detection based on image chroma

ICIP'09 Proceedings of the 16th IEEE international conference on Image processing
Steganalysis by subtractive pixel adjacency matrix

IEEE Transactions on Information Forensics and Security
LIBSVM: A library for support vector machines

ACM Transactions on Intelligent Systems and Technology (TIST)

Quantified Score

Hi-index	0.00

Visualization

Abstract

Extraction of discriminative feature is crucial to machine learning approach of image tampering detection. The state-of-the-art Markov transition probability feature is extended in this paper. We show that correlation between adjacent elements on the difference array of block DCT coefficients can be theoretically calculated and provides little information to the classification problem. We propose to decorrelate the variables and use the marginal distribution as feature in image tampering detection. The framework is applied to 1st and 2nd order Markov transition probability feature. Our experiment result shows the new presentation of the feature has competitive performance and greatly reduced dimensionality.