Article Preview
TopIntroduction
The widespread availability of Internet, together with the rapid popularity of digital cameras has resulted in rich digital images in our daily life. However, digital images can be easily manipulated by various editing software such as Adobe Photoshop. This has created an environment where the authenticity of digital images is often in doubt. To prevent the use of forged/tampered images in news report and court evidence, digital image forensic has attracted great research interests. Especially, since passive image forensics does not need any auxiliary data such as watermark or signatures, a variety of passive image forensic schemes have been developed (Li & Li, 2012). The most representative works about passive image forensic can be divided into two categories. The first class of forensics methods concentrate on identifying those content-changing image manipulations, including image splicing (Weir & Lau, 2010), and copy-move (Ferrara & Bianchi, 2012; Peng & Nie, 2011), which change the image content both visually and semantically. These content-changing manipulations are usually malicious forgery operations worthy of serious attention from the public, since they will influence people’s understanding towards the content of digital image. For image copy-move, it can be localized via fine-grained analysis of color filter array (CFA) artifacts or compound statistical features. And an evaluation of popular copy-move forgery detection approaches is summarized in (Christlein & Riess, 2012). The second class of forensics methods focus on those image manipulations which do not change the image content, such as resampling (Mahadian & Saic, 2008), JPEG compression (Fan & Queiroz, 2003), contrast enhancement (Stamm & Liu, 2008; Stamm & Liu, 2010), median filtering (Cao & Zhao, 2010; Kirchner & Fridrich, 2010) and image sharpening (Cao & Zhao, 2009; Cao & Zhao, 2011). However, those content-preserving image manipulations are more easily to be performed since it does not need sophisticated skills by professional users. They are often utilized as post-processing after image tampering to conceal its visual trails. For example, image sharpening is commonly applied as a retouching tool.
Unsharp masking (USM) sharpening is a widely used image sharpening method. The most popular image editing software Adobe Photoshop supports this function for local contrast enhancement. The principle behind USM is exaggerating the light-dark contrast between the two sides of an edge. Therefore, USM might be utilized to hide the tampering trails such as blurriness. In recent years, two blind forensics schemes have been proposed to detect the possible image sharpening. The clues for blind forensics are histogram aberration /ringing artifacts (Cao & Zhao, 2009), and the overshoot artifacts (Cao & Zhao, 2011) occurred around side-planar edges in sharpened images, respectively. Desirable results are obtained for the detection of USM sharpening. It is reported that even for small size images undergone USM sharpening operation, overshoot artifacts could be considered as a dominating feature for the forensics of the USM sharpening operations.