【作者】 张维克； 孔祥维； 尤新刚；

【Author】 Zhang Weike1 Kong Xiangwei1 You Xingang1,2(1 Information Security Research Center, Dalian University of Technology, Dalian 116024, China)(2 Beijing Institute of Electronic Technology and Application, Beijing 100091, China)

【机构】 大连理工大学信息安全中心；

【摘要】 为了对经过图像处理操作的图像内容进行准确认证,利用图像DCT低频系数的感知不变性生成了安全的哈希序列索引.用标准化后的DCT低频系数矩阵和基于密钥种子的随机数矩阵为数字图像生成哈希序列,研究分析了提出算法具有唯一性、鲁棒性和安全性的特性.实验结果表明,算法可以抵抗内容保持的修改操作,例如格式转换、中度几何变换和滤波失真等,具有较强的鲁棒性.同时算法具有较强的安全性,在同时得到伪随机序列生成器和密钥的情况下,才能获得图像的哈希值.另外,算法可为视觉近似图像生成相同或相近的400比特哈希值,且冲突率降低到10-8数量级.这种安全鲁棒的图像哈希方法可以用于数字图像认证和大量图像的数据库检索.更多还原

【Abstract】 Secure hash vector index is generated using perceptual invariance of image DCT (discrete cosine transform) low-frequency coefficients to authenticate content of image which has been processed. A hash vector is generated for digital image using the technique of combining standardized DCT low-frequency coefficients matrix with key based random matrix. The proposed algorithm is shown to be unique, robust and secure.Experimental results indicate that the proposed algorithm is able to resist the content-preserving modifications, such as format change, moderate geometric and filtering distortions. One can only acquire hash value of image while having both pseudo-random generator and key. In addition, the proposed algorithm generates same or similar hash value for perceptual same images, length of the hash value is only about 400 bits, and the collision rate decreases to 10-8 level. The proposed robust and secure algorithm can be applied to digital image authentication and retrieval of large image database.更多还原