Paper Info
Title
Inpainting Attacks Against Visible Watermarking Schemes
Abstract
Visible watermarking schemes are common IPR protection mechanisms for digital images and videos that have to be released for certain purposes but illegal reproductions of them are prohibited. Digital data embedded with visible watermarks will contain recognizable but unobtrusive copyright patterns, and the details of the host data are supposed to exist. The embedded pattern of a useful visible watermarking scheme should be difficult or impossible to be removed unless exhaustive and expensive labors are involved. In this paper, we propose a general attacking scheme against current visible image watermarking techniques. After attackers have selected the watermarked areas, no further user intervention are required. During the attacking processes, image inpainting techniques are adopted to obtain better approximations of the watermarked area. For watermarks with thin patterns, basic inpainting techniques can completely remove the embedded patterns. For watermarks with thick patterns, existing information of the host image within large watermarked areas is utilized to reduce the watermark removing problem to the aforecited thin-pattern case, and then the same inpainting techniques can be applied. Although the proposed scheme cannot guarantee that the embedded patterns will be completely removed for the latter case, but the shape of the embedded pattern will be seriously destroyed and reshaped by the structure of the host images. Thus, a general attacking scheme based on the characteristics of current visible watermarking techniques is obtained. To effectively protect digital contents with visible watermarks, we concluded that the structure of a original image should be taken into consideration when designing the corresponding watermark patterns. If the watermark patterns are independent of the details of the original image it protects, embedded visible watermark patterns will suffer from the proposed attacking scheme, and the robustness of the visible watermarks will be doubtful.
Year
DOI
Venue
2001
10.1117/12.435421
SECURITY AND WATERMARKING OF MULTIMEDIA CONTENTS III
Keywords
Field
DocType
IPR protection mechanisms, visible watermarks, image inpainting, attacks against watermarking
Computer vision,Digital watermarking,Digital protection,Robustness (computer science),Watermark,Digital image,Inpainting,Digital imaging,Artificial intelligence,Digital data,Physics
Conference
Volume
ISSN
Citations 
4314
0277-786X
7
PageRank 
References 
Authors
1.12
0
2
Name
Order
Citations
PageRank
Chun-Hsiang Huang113812.05
Ja-ling Wu21569168.11