Summary
International Technical Conference on Circuits/Systems, Computers and Communications
2016
Session Number:P3
Session:
Number:P3-16
Gaussian Filtering Detection of Digital Image Forensic
Se Hwan Park, Kang Hyeon Rhee ,
pp.1053-1056
Publication Date:2016/7/10
Online ISSN:2188-5079
Summary:
For a design of the Gaussian filtering (GF) detection (GFD) in the altered digital images, this paper presents a new feature vector that is formed from the autoregressive (AR) coefficients by AR model of the gradients of the horizontal and vertical lines in an image. In the proposed algorithm, AR coefficients are computed from the gradients of the lines. Subsequently, the defined 20-dim. feature vector is trained in a SVM (Support Vector Machine) for the GFD in the forged images. In the experiment, two kinds test items are the area under curve (AUC), and a minimal average decision error. The performance is excellent both at GF (3ﾃ・) vs. median filtering (3ﾃ・), and GF (5ﾃ・) vs. original and JPEG (90) on the DFD. However, in the measured performances of the AUC by the sensitivity (TP: True Positive rate) and 1-specificity (FP: False Positive rate) is above 0.9. Thus, it is confirmed that the grade evaluation of the proposed algorithm is rated as "Excellent (A).”