Passive Image Manipulation Detection Using Wavelet Transform and Support Vector Machine Classifier
In this paper, blind global contrast enhancement detection method is proposed using wavelet transform-based features. Wavelet subband energy and statistical features are computed using multilevel 2D wavelet decomposition. Mutual information-based feature
- PDF / 154,494 Bytes
- 9 Pages / 439.37 x 666.142 pts Page_size
- 30 Downloads / 190 Views
Abstract In this paper, blind global contrast enhancement detection method is proposed using wavelet transform-based features. Wavelet subband energy and statistical features are computed using multilevel 2D wavelet decomposition. Mutual information-based feature selection measure is employed to select the most relevant features while discarding the redundant features. Experimental results are presented using grayscale and G component image database and SVM classifier. Simulation results prove the effectiveness of the proposed algorithm compared to other existing contrast enhancement detection techniques. Keywords Image forgery detection ment detection Wavelet transform
Passive authentication Contrast enhance-
1 Introduction Advances in digital imaging technology resulted in sophisticated and cheap, yet powerful tools that enable the generation and manipulation of digital images without leaving any traces of tampering. As digital images are frequently used in military, printing media, medical records, and in court as legal photographic evidence, verifying its authenticity and integrity is important in order to restore trustworthiness of the candidate image. Image forgery detection algorithms are categorized as (1) active methods and (2) passive (blind) methods [4, 10]. G.K. Birajdar (&) Department of Electronics and Communication, Priyadarshini Institute of Engineering and Technology, Nagpur 440019, Maharashtra, India e-mail: [email protected] V.H. Mankar Department of Electronics and Telecommunication, Government Polytechnic, Nagpur 440001, Maharashtra, India e-mail: [email protected] © Springer Science+Business Media Singapore 2016 S.C. Satapathy et al. (eds.), Proceedings of International Conference on ICT for Sustainable Development, Advances in Intelligent Systems and Computing 408, DOI 10.1007/978-981-10-0129-1_47
447
448
G.K. Birajdar and V.H. Mankar
In active method, known information is embedded using preprocessing such as digital watermark or signature at the time of creating the image. Passive or blind techniques work on the assumption that digital image manipulation may not leave any visual traces of tampering. They may alter the underlying statistics or inconsistency of an image. Image forensic algorithms are of two types: (1) forgery detection and (2) forgery localization. In this article forgery detection is investigated. Contrast enhancement image processing operation is widely used by the attacker to conceal the cut and paste and spicing forgery in doctored images. In this paper, a blind method is proposed for global contrast enhancement detection based on energy and statistical features extracted using multilevel wavelet decomposition. Contrast enhancement operations can be considered as nonlinear pixel mappings which introduce inconsistencies into the statistical properties of an image. Mutual information-based criterion is employed to choose the most informative features from the original large feature space and removing redundant features. This technique is applicable to large ra
Data Loading...
