Image Tampering Localization Using Demosaicing Patterns and Singular Value Based Prediction Residue

Abstract

Almost all image sensors measure only one color per pixel through the color filter array. Missing pixels are estimated using a demosaicing process. For this reason, a demosaiced image leaves a particular trace. When an image is manipulated or tampered, the demosaicing trace can be changed. This change can serve as a basic clue for detecting or localizing image tampering. Demosaicing pattern-based tampering localization algorithms require a re-interpolation process, and the prediction residue between the given image and the re-interpolated image is commonly used to localize tampered regions. However, the prediction residue is not always valid because the demosaicing interpolation kernel cannot be known, which deteriorates the localization performance. This paper presents an effective re-interpolation process using singular value decomposition for an unknown demosaicing method. First, the green channel of the given image is decomposed into four sub-images according to the Bayer pattern. For a small block of each subimage, the singular value decomposition is performed. The prediction residue is obtained by reconstructing the image block after removing the largest singular value. The feature to localize the forged regions is extracted by the logarithm ratio of the prediction residue variance. The proposed method does not require any statistical model for the extracted feature, because the prediction residue is more accurate than that of conventional methods. We perform intensive experiments for three test datasets and compare the proposed method with state-of-the-art tampering localization methods, the results of which indicate that the proposed scheme outperforms existing approaches.INDEX TERMS Image tampering localization, demosaicing trace, singular value decomposition, prediction residue, re-interpolation kernel, color filter array, image splicing.

A Contrario Mosaic Analysis for Image Forensics

“…Uncompressed JPEG Q = 95 JPEG Q = 90 Mimic 0.724 0.311 0.196 4Point [7] 0.709 0.307 0.151 AdaCFA [6] 0.692 0.005 0.003 DDem [10] 0.401 0.129 0.093 Shin [35] 0.104 0.001 0.001 Choi [11,4] 0.603 0.156 0.070 Ferrara [18,37] 0.071 0.000 0.000 Dirik [16,37] -0.002 0.000 0.001 Park [33] 0.116 0.001 0.000…”

Demosaicing to Detect Demosaicing and Image Forgeries

“…Results on the Trace [31] database, with the CFA Grid (Grid) and CFA Algorithm (Alg) datasets, are presented in Table I. We compare our method with six demosaicing-based forensic tools, Bammey [12], Shin [19], Choi [10], [11], Ferrara [7], Dirik [27], Park [28], and with two generic forgery detection methods Splicebuster [30] and Noiseprint [29]. The original Choi and Shin methods [10], [19] only detect mosaic patterns, but do not detect forgeries.…”

“…The implementations for Ferrara [7] and Dirik [27] are taken from [32]. Bammey [12], Park [28], Splicebuster [30] and Noiseprint [29] are used directly with the authors' implementations.…”