Digital Watermarking Using Variable Embedding Intensity Based on Degrees of Complexity of Image Blocks

概要

論文の詳細を見る
When embedding information in an image with digital-watermarking techniques, positions of embedded information influence the quality of the image and robustness against attacks. Human visual perception is not sensitive to change in complicated domains but is sensitive to change in flat domains. Thus, the current method divides an image into flat and non-flat parts by standard deviation, and setting the embedding intensity for each domain. However, since the edge area and the complicated areas are not distinguishable, if embedding intensities were made high, degradation of an embedded image in an edge area appeared. So, we propose a method of dividing an image into three domains, i.e., flat, edge, and complicated areas by introducing degrees of complexity of image blocks used in steganography techniques in addition to standard deviation used in the current method. Then, embedding intensities arc set for each domain and watermark information is embedded. This mitigates degradation of the embedded watermark information. We present evaluations of the quality of an embedded image and its robustness for attacks using the proposed method.
社団法人電子情報通信学会の論文
2003-01-14