Search

• Effective ML that combines histogram statistics with support vector machine ability. • Explores saturation attainment of gamma-corrected versions to detect contrast change • Encodes pattern as slope of kurtosis and mean skewness displacement features. • Achieves excellent large-margin separability of images with 97% accuracy on Dresden Photo attack and anti-forgery techniques have been on the rise in the internet age. Attackers conceal tampering by applying anti-forgery techniques, such as contrast enhancement. Digital image forensics concerns with detecting traces of these operations subjected upon the image. In this work, a novel Anti-Forensic Contrast Enhancement Detector (AFCED) scheme has been proposed to assess contrast changes from anti-forensic digital images. The proposed model is based on the underlying observation that, with repeated gamma correction, the rate of attaining contrast saturation in normal images is greater than already contrast-improved images. Thus, we employ skewness and kurtosis of the gamma-corrected second-order derivative histograms as features to distinguish normal images from contrast-enhanced images. These features are robustly classified on a Support Vector Machine. The proposed model achieved 97.40% F1-score and 98.18% sensitivity on the Dresden and Wikiart image databases on-par with state-of-the-art results. [ABSTRACT FROM AUTHOR]

Abstract: In this paper, we propose a noise removal algorithm for digital images. This algorithm is based on hypergraph model of image, which enables us to distinguish noisy pixels in the image from the noise-free ones. Hence, our algorithm obviates the need for denoising all the pixels, thereby preserving as much image details as possible. The identified noisy pixels are denoised through Root Mean Square (RMS) approximation. The performance of our algorithm, based on peak-signal-to-noise-ratio (PSNR) and mean-absolute-error (MAE), was studied on various benchmark images, and found to be superior to that of other traditional filters and other hypergraph based denoising algorithms. [Copyright &y& Elsevier]