PRIVACY PRESERVATION IN SOCIAL MEDIA BY IMAGE PROCESSING

Photo sharing is a captivating feature which popularizes Online Gregarious Networks (OSNs). Lamentably, it may leak users’ privacy if they are sanctioned to post, comment, and tag a photo liberatingly. In this paper, we endeavor to address this issue and study the scenario when a utilizer shares a photo containing individuals other than him/her (termed co-photo for short). To obviate possible privacy leakage of a photo, we design a mechanism to enable each individual in a photo be cognizant of the posting activity and participate in the decision making on the photo posting. For this purport, we require an efficient facial apperception (FR) system that can apperceive everyone in the photo. However, more authoritatively mandating privacy setting may limit the number of the photos publicly available to train the FR system. To deal with this dilemma, our mechanism endeavors to utilize users’ private photos to design a personalized FR system concretely trained to differentiate possible photo co-owners without leaking their privacy. We additionally develop a distributed consensus predicated method to reduce the computational intricacy and for fend the private training set. We show that our system is superior to other possible approaches in terms of apperception ratio and efficiency. Our mechanism is implemented as a proof of concept Android application on Facebook’s platform.照片分享是一项极具吸引力的功能，推动了在线社交网络（Online Gregarious Networks，OSNs）的普及。遗憾的是，若用户被允许自由发布、评论及标记照片，则可能造成用户隐私泄露。本文致力于解决该问题，并针对用户分享包含他人的照片（以下简称合影）这一场景展开研究。为避免照片可能存在的隐私泄露风险，我们设计了一种机制，使照片中的每位主体均能知晓该照片的发布行为，并参与照片发布的决策过程。为此，我们需要一套能够识别照片中所有人物的高效面部识别（Facial Recognition，FR）系统。然而，若对隐私设置进行更为严格的强制管控，则可用于训练FR系统的公开照片数量将会受限。为解决这一困境，我们的机制旨在利用用户的私人照片构建个性化FR系统，该系统可在不泄露用户隐私的前提下，精准区分照片中可能存在的共同所有者。此外，我们提出了一种基于分布式共识的方法，以降低计算复杂度并保护私人训练数据集。实验结果表明，相较于其他可行方案，我们的系统在识别率与效率方面均更具优势。我们还在Facebook平台上实现了该机制的概念验证安卓（Android）应用程序。