一种高度容错和精确识别的数字水印算法  

A Digital Watermarking Algorithm of Error Tolerance and Precise Recognition

作　　者：尹德辉[1] 李炳法[2] 唐燕[3] 

机构地区：[1]宜宾学院计科系,四川宜宾644007 [2]四川大学计算机学院,四川成都610064 [3]四川电视大学,四川成都610073

出　　处：《四川大学学报（工程科学版）》2005年第6期107-112,共6页Journal of Sichuan University (Engineering Science Edition)

摘　　要：在传输过程中,由于受到各种信号处理或攻击,一些数字水印算法不能精确提取水印信息。为克服其缺点,通过将版权信息转化为二进制串,伪随机的添加一定的二进制位后用ASCII码表将其转化为英文字母,并对包含该英文字母的图像进行置乱。利用图像融合技术嵌入容量大的特点,将置乱后的图像嵌入到载体图像中,将置乱的矩阵、随机选取的序列、迭代次数作为密钥,提高算法的安全性。提取水印时,根据融合的逆运算和密钥可还原包含英文字母的图像,将英文字母逆向转化为二进制串,去掉添加的二进制位后还原得到精确的原始版权信息。实验证明算法是切实可行,可广泛应用于数字音、像产品的版权保护和军事上重要信息的加密传输。Due to various signal processing and attacking during image transmission, some digital watermarking algorithms cannot detect watermarking information precisely. In order to overcome the shortcoming, an improved algorithm was proposed. The algorithm converts the copyright information into a binary sequence. After inserting certain amount of pseudorandom binary bits, the binary sequence is translated into English letters according to ASCⅡ table. Then, the image containing the English letters is hashed. Exploiting the large embedding volume of image fusion technology, the hashed image is embedded into the carrier image. The hashing matrices, the randomly selected binary sequence and the iteration number are used as secret keys, which improves the algorithm＇s security. When extracting the watermarking, the original image containing the English letters can be recovered using reverse fusing operation and the secret keys. Then, the English letters axe inversely converted into binary sequences. After removing the appended binary bits, the original copyright information can be recovered precisely. Numerical experiments proved that the proposed algorithm is feasible for copyright protection of digital audio and video products, and encrypted transmission of important military information.

关 键 词：数字水印 复合置乱算法 图像融合 

分 类 号：TP309.2[自动化与计算机技术—计算机系统结构]