A Universal Image Forensics of Smoothing Filtering

A Universal Image Forensics of Smoothing Filtering

Anjie Peng (School of Computer Science and Technology, Southwest University of Science and Technology, Mianyang, China & Guangdong Key Laboratories of Information Security Technology, Sun Yat-Sen University, Guangzhou, China), Gao Yu (School of Computer Science and Technology, Southwest University of Science and Technology, Mianyang, China), Yadong Wu (School of Computer Science and Technology, Southwest University of Science and Technology, Mianyang, China), Qiong Zhang (School of Data and Computer Science, Sun Yat-Sen University, Guangzhou, China) and Xiangui Kang (School of Data and Computer Science, Sun Yat-Sen University, Guangzhou, China)
Copyright: © 2019 |Pages: 11
DOI: 10.4018/IJDCF.2019010102

Abstract

Digital image smoothing filtering operations, including the average filtering, Gaussian filtering and median filtering are always used to beautify the forged images. The detection of these smoothing operations is important in the image forensics field. In this article, the authors propose a universal detection algorithm which can simultaneously detect the average filtering, Gaussian low-pass filtering and median filtering. Firstly, the high-frequency residuals are used as being the feature extraction domain, and then the feature extraction is established on the local binary pattern (LBP) and the autoregressive model (AR). For the LBP model, the authors exploit that both of the relationships between the central pixel and its neighboring pixels and the relationships among the neighboring pixels are differentiated for the original images and smoothing filtered images. A method is further developed to reduce the high dimensionality of LBP-based features. Experimental results show that the proposed detector is effective in the smoothing forensics, and achieves better performance than the previous works, especially on the JPEG images.
Article Preview

2. The Proposed Method

In this section, we first analyze the statistical differences between original images and smoothing filtered images in the high frequency residual domain. Then we employ autoregressive model and local binary patterns to extract the fingerprints left behind by the smoothing filtering operations in the residual domain. We finally introduce how to ensemble LBP and AR feature set for smoothing filtering forensics.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 11: 4 Issues (2019): 1 Released, 3 Forthcoming
Volume 10: 4 Issues (2018)
Volume 9: 4 Issues (2017)
Volume 8: 4 Issues (2016)
Volume 7: 4 Issues (2015)
Volume 6: 4 Issues (2014)
Volume 5: 4 Issues (2013)
Volume 4: 4 Issues (2012)
Volume 3: 4 Issues (2011)
Volume 2: 4 Issues (2010)
Volume 1: 4 Issues (2009)
View Complete Journal Contents Listing