Research on Digital Watermarking for Video Data and Text Documents

(1)图像和视频数字水印方法的发展：我们提出了一种基于小波变换的静止图像和视频数字水印方法。首先，我们已经将所提出的方法应用于静止图像的水印，并已表明，这种方法是鲁棒的有损压缩，如JPEG，添加噪声，降低灰度级，和StirMark攻击。接下来，我们将此方法应用于视频水印。实验结果表明，水印是从MPEG压缩的视频中提取出来的。(2)文本数字水印方法的发展：我们提出了一种基于形态信号处理技术的文本数字水印方法。该方法将信息数据（水印）嵌入到文本文档中的字符和/或图形等二进制模式中。我们已将此方法应用于英文字母表的字母，并实现了一个英文文档的水印技术。使用所提出的技术，我们已经证实，从数字实验中嵌入的数据可以检测到水印的字母与噪声和打印在纸上。(3)在频域数字水印算法性能评价方面：重点研究了基于量化控制的频域数字水印算法，提出了水印嵌入和检测系统的模型。其次，利用该模型对水印算法进行了分析，并提出了水印算法的性能评价方法。因此，我们可以知道有多少位能够被嵌入，以及应该设置多大的嵌入强度。此外，我们还讨论了基于离散小波变换和离散余弦变换的数字水印算法的性能差异。

(1) Development of digital watermarking method for images and video :We have proposed a watermarking method for both still images and video using the wavelet transform. First, we have applied the proposed method to the watermark for still images and have shown that this method is robust against lossy compression like JPEG, adding noise, reduction of grayscale level, and StirMark attacks. Next, we have applied this method to the watermark for video. As a result, we have confirmed that the watermark is extracted from the watermarked video compressed with MPEG.(2) Development of digital watermarking method for text documents :We have presented a watermarking method for text documents using the morphological signal processing technique. This method embeds information data (watermark) into binary patterns such as character and/or figure patterns in text document. We have applied this method to the letters of the English alphabet and realize a watermark technique for English documents. Using the proposed technique, we have confirmed from numerical experiments that the embedded data can be detected from watermarked letters with noise and printed on paper.(3) On the evaluation of performance of digital watermarking method in the frequency domain :We have focused on the watermarking method in the frequency domain, in which watermark is embedded using a controlled quantization process, and suggested the model of watermark embedding and detecting system. Next, using this model, we have analyzed the watermarking method and proposed how to evaluate its performance. As a result, we can know how many bits are able to be embedded and how large the embedded intensity should be set. Furthermore, we have discussed the difference of the performance between the DWT- and DCT-based watermarking methods.