Open Access
Liu, Sha
Graduate Program:
Computer Science and Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
November 03, 2017
Committee Members:
  • Sencun Zhu, Thesis Advisor
  • Wang-Chien Lee, Committee Member
  • mobile cameras
  • privacy protection
  • Bluetooth
  • BLE broadcast
  • face matching
With the growing popularity of mobile devices with built-in cameras, especially wearable devices such as Google glasses, capturing images has become a trivial job for ordinary people. Compared with legacy cameras, portable mobile devices require us only to press the “shutter” button when we are ready to take a photo, no matter when or where we are. As a result, a growing number of people like to record everything in their daily lives and then share these mementos with their friends online. However, these gathered digital images may violate the consent of the photographed persons, which may lead to concerns about privacy. Unfortunately, the current methods cannot perfectly solve these concerns. Thus, in this thesis, we propose our solution for a privacy-respecting system for mobile phones. In order to meet the requirements of efficiency and usability, on the one hand, we choose Bluetooth wireless communication, which has been widely deployed in most electronic devices, as the main approach to transmit privacy preferences of users. In addition, we try to improve the compatible of our system with the mainstream devices. Therefore, we have designed two different methods to transmit privacy policies: classical Bluetooth communication and BLE (Bluetooth Low Energy) broadcast. On the other hand, we use a simple feature extract algorithm, perceptual hash algorithm, which only uses 8 bytes to represent face of a specific person. This algorithm can largely reduce the data size of the face feature vector and further reduce the computation overhead so that it can be deployed in most of mobile phones. We have implemented a prototype based on Android platform, and also made some evaluations based on three main facets: usability, security and performance. All analysis results prove that our system has a high usability and good performance with acceptable security.