See-Through Vision: A Visual Augmentation Method for Sensing-Web

Abstract

Many surveillance cameras are being installed throughout the environments of our daily lives because they effectively maintain safety and offer security to ordinary people. On the other hand, they can also cause serious discomfort and anxiety to the same people. This paper proposes a novel framework called See-Through Vision that utilizes surveillance cameras as a public sensing device by exploiting a state-of-the-art mixed-reality technique. Some advanced systems developed by us for See-Through Vision are introduced, and we discuss their advantages and how they actually maximize the benefits of surveillance cameras. Since See-Through Vision is a powerful tool that may violate the privacy of other people, we propose a sophisticated solution that can strike a good balance between the benefits and the drawbacks of such an approach. We propose privacy-safe See-Through Vision and demonstrate the system at a shopping mall in Kyoto.