Data hiding scheme with geometric distortion correction

Jiang, Xiaoxia (2008) Data hiding scheme with geometric distortion correction. Masters thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (5MB)

Abstract

This work presents a new data hiding scheme with geometric distortion corrections. In a data hiding scheme, robustness of embedded information against geometric distortion is a critical issue, since a decoder can be easily confused when image size or orientation is changed, even to a small extent. One of the approaches to overcome this problem is to rescale and/or rotate the image back to its original size/orientation. In order to do so, the decoder needs to know by how much the image has been scaled and rotated. Such knowledge of the original image would greatly simplify the extraction of embedded information. However, huge quantities of images have to be compared by intelligent agents to locate the original images. Therefore, there is a strong practical reason to seek robust extraction schemes without access the original images. -- In this approach, two separate channels of information are embedded in a typical RGB image by selecting two of its three color planes as two channels: one for synchronization information and the other one for the actual information. Synchronization information of the original image is computed by the locations of feature points extracted by a Harris-Laplace corner detector and embedded as a watermark into the image by a content-based watermarking method. This synchronization information is extracted at the detection stage and compared with the calculated synchronization information of the distorted image so that the scaling factor and rotation angle can be estimated. -- The effectiveness of the presented scheme is evaluated by assessing its robustness against different geometric attacks and transformations. The geometric distortion estimation shows best robustness in certain ranges of scaling and rotation distortions. The proposed scheme exhibits great potential because of the successful detection of synchronization information.

Actions (login required)

View Item