【作者】 李凯；

【导师】 裴庆祺；

【作者基本信息】 西安电子科技大学 ， 信息安全， 2017， 硕士

【摘要】 随着社交网络个性化、多元化服务的普及,社交网络平台如雨后春笋般出现在人们的日常生活中。越来越多的用户对社交网络平台这种社交方式产生了依赖。庞大的用户规模、海量的信息交互、多元化的系统服务导致了社交网络中产生了巨量的以图像形式为主的共享数据。而由于现有系统针对图像中人脸敏感信息的访问控制粒度较粗,使得共享数据在社交网络中强渗透力地传播的同时,人脸敏感信息泄露问题严峻,甚至引起恶性事件的发生。但是,如果强制性对用户的信息进行全方位的保护,使得用户安全性得到保证的同时,就会大大限制社交网络中用户之间的社交体验感。因此,针对社交网络中细粒度人脸隐私保护的方案研究具有重要意义。为了解决图像中非敏感信息在社交网络中的传播和共享,便捷用户社交的同时,对图像中人脸敏感信息进行细粒度隐私保护的博弈性难题,本文提出一种基于用户间细粒度社交关系,对图像中人脸敏感信息进行细粒度隐私保护的方案。本文在确定系统的访问控制单元上,将图像资源细分到人脸图像块和非人脸图像块,一方面将图像中的人脸敏感区域与非敏感区域分离,为系统细粒度访问控制奠定基础;另一方面也增强了系统的防爬虫性能。在确定访问控制依据和策略上,本文融入社交的主动性和时间属性塑造了细粒度的用户间社交关系,而且,由于访问控制单元细化到了人脸敏感块,使得方案不仅在访问者与发布者之间拥有细粒度的关系刻画,同时,也构造了访问者与访问资源之间以及发布者与访问资源之间细粒度的关系,使得系统的访问控制依据更加细粒度。之后,结合更加人性化和细粒度的访问控制策略,制定了一个细粒度的访问控制方案。而且在访问控制策略实现方式上,本文采取AES加密和属性基加密结合的方式,既极大程度保证了数据传输过程中敏感隐私数据的安全性,又真正意义上地实现了高效、细粒度、支持动态改变的访问控制策略。同时,本文结合分布式人脸识别方案,基于Ucenter Home针对上述方案进行二次开发,实现并且验证了系统方案的可行性以及在社交网络中针对图像中的人脸敏感信息的细粒度隐私保护的能力。更多还原

【Abstract】 With the increasing popularity of individualized and diversified social network services,social network platforms have sprung up everywhere in people’s daily lives.In social networks,the large scale of users,the interaction of mass information and the diversity of services lead to a large amount of shared data,which is mainly formed by images.The coarse-grained access control scheme for sensitive facial information of images makes the shared data spread penetratively in existing social networks.This kind of scheme also causes severe problems in revealing facial sensitive information,and even leads to vicious incidents.However,in the event that users’ information was protected comprehensively and compulsively,the security of users would be guaranteed,which will sharply reduce the sense of social experience among users in social networks.Therefore,research on the fine-grained privacy preserving scheme of facial sensitive information in social networks is of great significance.In order to solve a gaming problem that non-sensitive information of images spreads and shares strongly,and facial sensitive information of images is more fine-grained protected,we propose a fine-grained facial sensitive information preserving scheme.The core of the scheme depends on the fine-grained relationships among users.This scheme divides images into facial and non-facial patches closely at the stage of defining access control cell.On the one hand the facial sensitive and non-sensitive areas of the images are separated for fine-grained access control,and on the other hand the performance of the anti-crawler is raised.At the stage of building the access control strategy,we establish fine-grained relationships among users,which contains social initiative and time attributes.In addition,our scheme establishes fine-grained relationships among publishers,visitors and accessing resources,aiming to a more fine-grained access control evidence.At the stage of establishing implementation of the access control strategy,we combine the AES encryption policy with the ABE encryption policy,which not only assures sensitive data security in the process of data transfer,but also achieves changed,fine-grained access control efficiently and dynamically.At the same time,by the distributed face recognition method,this paper launches the secondary development based on UCenter Home,demonstrating the feasibility and the fine-grained privacy preserving capability of facial sensitive information in social networks.更多还原