A Digital Image Watermarking using a Feature-driven Attention Module

M.C. Park and K.J. Cheoi (Korea)


Visual, Attention, Region,Watermarking, Semantic, Importance


This paper takes a feature-driven attention module for an active vision system into consideration for digital image watermarking. The major feature of active vision is detecting interesting regions from a large image without scanning the whole field of view. A feature driven attention module generates the visual pathways in a natural scene for an active vision system to elicit a fixation point. This module is employed for image watermarking to obtain the region of interest (ROI) or drawing visual attention. Watermark information, an indication of ownership or copyright, is embedded into the obtained region using a feature-driven attention module. Previous studies in digital image watermarking have been focused on the signal processing techniques, especially in waveform coding spreading watermarks over the entire target image. However, we notice that the third party's visual attention is usually concentrated on a few regions in an image but not on all of them. These regions are easy to be the target of attacks, in other words, copying, cropping and so forth. If watermark information is inserted into the region from the beginning, it can be detected with high correlation. Therefore, we decided to embed watermarks into the visual attention area considering semantic importance by psychological factor. We suggest a digital image watermarking method for the first time, which uses a biological system and semantic importance.

Important Links:

Go Back