彩色图像的高保真可逆信息隐藏算法研究

项目名称： 彩色图像的高保真可逆信息隐藏算法研究

项目编号： No.61502160

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 自动化技术、计算机技术

项目作者： 欧博

作者单位： 湖南大学

项目金额： 20万元

中文摘要： 近年来，随着社交网络、共享平台快速兴起，以及彩色图像本身所具有的内容直观、表现力丰富等优势，彩色图像迅速成为日常生活中最受欢迎的一种知识传播和信息共享的媒介。然而现有的可逆信息隐藏研究主要关注灰度图像的理论研究，应用面较窄，针对彩色图像的算法还比较少。不同于灰度图像，彩色图像三通道之间既存在较强的相关性，也有差异。直接应用传统算法没有充分考虑颜色通道间的相关性和差异性，因而难以达到高保真的嵌入性能要求。为此，本课题将研究颜色通道间的相关性和差异性，基于自适应嵌入的思想，提出适合彩色图像的高保真可逆信息隐藏技术，主要包括：基于通道内容自适应分配容量的可逆信息隐藏、基于区域划分的自适应嵌入和利用启发式分类与特征融合的彩色图像快速嵌入算法，并构建基于人眼视觉特性的彩色图像可逆嵌入失真模型。项目的研究成果可以广泛应用于彩色图像的快速可逆嵌入和提取，具有重要的理论意义和实用价值。

中文关键词： 可逆信息隐藏；高保真；自适应嵌入；彩色图像

英文摘要： In the past few years, with the development of social networking and sharing platform, color image has become the most popular medium for communication and information sharing in daily life, as it can convey the information in a more intuitive and expressive way. However, current reversible data hiding (RDH) researches mainly focus on the gray-scale image, but seldom pay attention to the color one. Different from the gray-scale image, in a color image, a vector consisting of red (R), green (G) and blue (B) components is assigned to a pixel. Usually, RGB channels are not only highly correlated with each other, but also have different properties to some extent. Directly applying the current gray-scale RDH methods cannot provide an optimal solution for the high-fidelity RDH on color images, as they ignore the correlations and differences among channels. To remedy this drawback, an adaptive embedding criterion for RDH on color images is proposed by considering both the correlations and differences among channels. The key ideas behind this topic includes channel-dependent payload partition, region-segmentation based adaptive embedding, heuristic classification, feature fusion, and construction of embedding distortion model by considering human visual system. The proposed research has important theoretical significance and practical value, and can be widely used for RDH on color images with a low computation complexity.

英文关键词： Reversible data hiding ;High-fidelity;Adaptive embedding;Color image