Digital Watermarking for Images and Videos

• Main Authors: Hsu, Chiou_Ting, 許秋婷
• Other Authors: Ja-Ling Wu
• Format: Others
• Language: zh-TW
• Published: 1997
• Online Access: http://ndltd.ncl.edu.tw/handle/33928614185915423842

博士 === 國立臺灣大學 === 資訊工程學系研究所 === 85 === Due to the rapid and extensive growth of electronic publishing industry, data can now be distributed much faster and easier. Unfortunately, engineers still see immense technical challenges in discouraging unauthorized copying and distributing of electronic documents. One potential solution for claiming the ownership is to use watermark, which must be embedded into the media so that the watermarked media is imperceptible by the user. In most of the previous works, the watermark is a symbol or ID number which comprises of a sequence of bits. The verification of the extracted results can only be evaluated by quantitative measurements. In this thesis, we propose several techniques for embedding digital watermarks with visually recognizable patterns into images and video. Robustness is crucial to the success of watermark embedding. Making the watermark indestructible is not a trivial problem. Whatever "hole" in the host data one finds to fill with watermark is likely to be eliminated by data compression. The key is to find holes that are not suitable for exploitation by compression algorithms. In this thesis, the following embedding algorithms are presented. In DCT-based image watermarking, the binary watermarks are embedded into the images by selectively modifying the middle-frequency parts of the images to achieve a robust and invisible result. In multiresolution image watermarking , both the watermark and the host image are composed of multiresolution representations. Since the HVS inherently performs multiresolution decomposition, each decomposed layer of the binary watermark is embedded into the corresponding layer of the image. In case of image quality degradation, the low resolution rendition of the watermark will still be preserved within the corresponding low resolution components of the image By extending the temporal domain, the DCT-based approach could also be applied to the video frames. The multiresolution technique could also be applied to the video frames based on 3-D spatio-temporal decomposition. The experimental results show that the proposed methods result almost imperceptible difference between the watermarked data and the original data, and also robust to common image manipulation and processing operations, JPEG lossy compression, MPEG compression and even multiple-watermarking.