两种基于压缩感知的图像数字水印方法  

Two Image Watermarking Methods Based on Compressive Sensing

作　　者：苗依迪 吕军 李秀梅[1] 

机构地区：[1]杭州师范大学杭州国际服务工程学院,浙江杭州311121

出　　处：《系统仿真学报》2017年第11期2649-2655,2662,共8页Journal of System Simulation

基　　金：国家自然科学基金(61571174);浙江省自然科学基金(LY15F010010)

摘　　要：压缩感知作为一个新兴的采样理论,由于其突破奈奎斯特采样定律而引起了广泛的关注。基于压缩感知技术分别针对载体图片和水印图片进行压缩感知测量,实现了两种不同的数字水印嵌入及提取方法,并对载水印图片进行了高斯噪声、椒盐噪声、滤波、压缩、裁剪等不同形式的水印攻击测试。实验表明,两种不同的图像数字水印方法在流程上虽有不同,但都能保证嵌入数字水印的鲁棒性和安全性。As an emerging sample theory, compressive sensing attracts wide attention because it breaks through the Nyquist sampling theorem., Two different methods of watermark embedding and extraction are presented by measuring the carrier image and watermark image respectively based on compressive sensing. Moreover, the attack tests, such as the Gaussian noise, pepper and salt noise, filtering, compression, and cropping, are implemented to watermarked images. Experiment results show that although the two different methods for image watermarking have different processing procedure, both can guarantee the robustness and security of embedded digital watermark.

关 键 词：压缩感知 数字水印 水印攻击 鲁棒性 安全性 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]