项目名称: 图像感兴趣区域提取及加密研究
项目编号: No.61462032
项目类型: 地区科学基金项目
立项/批准年度: 2015
项目学科: 计算机科学学科
项目作者: 温文媖
作者单位: 江西财经大学
项目金额: 44万元
中文摘要: 网络的开放性给图像数据的安全传输带来了很大挑战,图像完全加密技术一直以来是解决该问题的关键手段之一。然而,在很多商业多媒体应用中,需有意对用户感兴趣的图像区域重度加密,此时用户需要根据部分可视化区域来判断是否应该购买相应的服务,这种便捷的交互模式是完全加密技术力所不及的。本项目围绕这类特殊的应用需求研究图像感兴趣区域提取及加密技术。通过对图像特征数据的分析和处理,整合图像高效特征来构建面向感兴趣区域提取的变分模型,利用能量泛函的极小化来提取感兴趣区域;根据感兴趣区域的物理特点,充分利用光学变换并行处理的能力以及混沌系统初始敏感的特性来设计适配感兴趣区域的加密算法;同时,拟通过常用的攻击方法去验证或者利用可证明安全去证明感兴趣区域加密方案的安全性。本项目的研究实现了图像处理与图像加密的融合,其取得的研究成果将为未来数字多媒体安全应用提供了一种新的思路。
中文关键词: 图像加密;感兴趣区域提取;变分模型;通信安全
英文摘要: The openness of the network has brought a great challenge for image-data security transmission, and full encryption has been one of key technologies for such a problem. However, in many commercial multimedia applications like pay-per-view, they require part of multimedia content preview, where the regions of interest (ROI) are encrypted heavily and part of multimedia content is visualized or degraded to some poor quality so that it can be previewed free of charge to determine whether should pay for the corresponding service, which are beyond full encryption. In order to meet these kinds of multimedia applications, this project makes a research on ROI extraction with application to image encryption. We analyze and handle image feature data, and then integrate the important feature to construct some variational models for ROI extraction, and such regions can be detected by minimizing the energy functional. According to the physical characteristics of ROI, some adaptive encryption schemes are designed by the significant properties of optical transform and chaotic system, such as parallel processing, and sensitivity to control parameter and initial condition. The security and availability of image encryption scheme based on ROI will be verified by common attacks methods or provable security. This completion of this project will integrate image processing and image encryption and a new means of the future digital multimedia security applications will be achieved.
英文关键词: Image Encryption;Regions of Interest (ROI) Extraction;Variational models;Communication Security