【作者】 王巍；

【导师】 衣治安；

【作者基本信息】 大庆石油学院 ， 计算机技术， 2008， 硕士

【摘要】 随着信息时代的到来,特别是Internet的普及,信息的安全保护问题日益突出。另一方面,多媒体技术已被广泛应用,需要进行加密、认证和版权保护的声像数据也越来越多。传统的信息安全技术基本上都以密码学理论为基础,但随着计算机处理能力的快速提高,这种方法变得越来越不安全。数字水印是一种在其它媒介上通过嵌入一些重要标识信息以保护媒介的隐私、版权等的安全技术。论文主要应用数字水印技术和指纹识别技术,将要持证件人的指纹图像作为水印嵌入到要作证件的相片中;然后采集证件使用人的指纹和证件里的指纹水印进行相似度比较来辨别真伪。本文以DCT快速变换、图像压缩编码技术、人类视觉系统特性为基础,利用图像本身特征和人的视觉特性确定水印的嵌入位置,设计了一种算法。该算法将图像进行8×8分块后对各块图像进行DCT变换,选择宿主图像的纹理子块作为嵌入水印的地方,然后利用置乱的方法将水印嵌入到这些子块的DCT中高频分量中,水印的提取和检测不需要原始图像。实验结果表明:文中算法不仅具有较好的透明性和鲁棒性,而且即使在含水印图像遭受不同程度的压缩、剪切、滤波等攻击后仍能有效地检测到水印的存在。更多还原

【Abstract】 Information security protection technology is now attracting more and more attention in this Information Age. The demand of copyright protection, secure transmission and confidential communication is increasing dramatically as the multimedia technology widely spreaded. Most of conventional information security techniques are based on the theory of Cryptology. It is obvious that with the calculating ability of modern computers rapidly improved, there is less possibility to keep security. Digital watermark is a security technique which used to protect some information such as privacy, authority through embedding watermark into target mediators.In this thesis we applied digital watermark technique and fingerprint identification technique in ownership verification of certificates. The image of owner’s fingerprint was embedded as watermark into the photo which is to be used in certificate making. After that we compare the watermark with the fingerprint we got.A novel algorithm was proposed in our work by discrete cosine transformation (DCT), still image compression coding technique, and human visual system (HVS) to calculate the position embedded. The DCT is applied in blocks of 8×8 pixels in this approach. Texture blocks of the host image are selected as target positions, and then the watermark was inserted in the DCT high and middle frequency of those parts with scrambling technique. The extraction and detection of the watermark does not need the original image. The experiment results show that the algorithm we use meets the demand of transparency and robustness effectively: the watermark can be detected even the watermarked image was compressed, clipped or filtered greatly.更多还原