基于SVD-GA的量化指纹优化算法  

Quantization Fingerprinting Optimization Algorithm Based on SVD-GA

作　　者：程格平[1] 徐德刚[1] 王新颖[1] CHENG Ge-ping;XU De-gang;WANG Xin-ying(School of Mathematical and Computer Sciences,Hubei University of Arts and Science,Xiangyang 441053)

机构地区：[1]湖北文理学院数学与计算机科学学院,襄阳441053

出　　处：《现代计算机（中旬刊）》2018年第6期60-64,67,共6页Modern Computer

基　　金：湖北省教育厅科学技术研究项目(No.B2017156)

摘　　要：为了获得指纹检测概率和视觉质量的平衡性能,提出基于SVD-GA的量化指纹优化算法。首先对载体图像的离散余弦变换域进行奇异值分解,然后根据视觉模型调节量化嵌入的强度,在原始图像的最大奇异值分量嵌入指纹信息,最后利用遗传算法优化指纹的量化嵌入参数,以权衡指纹算法的共谋抵抗性能和视觉保真度。实验结果表明,提出的指纹算法具有较好的抗共谋攻击能力和视觉保真度。Proposes a fingerprinting optimization algorithm based on SVD-GA to balance the performance between detection probability and perceptual quality.The singular value decomposition is firstly accomplished in the discrete cosine transform domain of cover image,then the quantization embedding strength is adjusted by the visual model and the largest singular values from the original image are modified to embed the fingerprinting information.Finally the genetic algorithm is used to optimize the quantization embedding parameter to achieve the tradeoff between collusion resistance and visual fidelity of the fingerprinting algorithm.Experimental results show that the proposed fingerprinting algorithm has better collusion-resistance and perceptual fidelity.

关 键 词：数字指纹 视觉模型 量化嵌入 奇异值分解-遗传算法(SVD-GA) 

分 类 号：TP309.7[自动化与计算机技术—计算机系统结构]