MIT Media Lab (E14-525), Nagashima Conference Room
We envision a public space that is populated with active visual surfaces. Where there is a sign, there can be a display. Similarly, we assume an equally dense population of normal cameras. The work of this thesis is to use that ensemble of equipment, both personal and environmentally deployed, to build rich information exchanges where the visual data is simultaneously useful to human observers and camera devices. The screens are both displays and data transmitters.
Following the software-defined approach, which leverages screens and cameras as general-purpose computers and allows flexibility and development of new methods, we introduce "VR-Codes". In a particular implementation, the codes are represented as rapidly alternating metamers for the background color of a region in the image. The rapid alternation is designed to be invisible to the human observer but detectable and decodable by existing and prospective cameras in consumer and mobile devices. The techniques presented exploit how cameras see differently than human eyes.
Evaluating performance in an information theoretical sense highlights the underutilized temporal, spatial and frequency dimensions available to the interaction designer concerned with human perception. Results suggest that the one-way point-to-point transmission is good enough for extending the techniques toward a two-way bi-directional model with realizable hardware devices. The new visual environment contains a second data layer for machines that is synthetic and quantifiable; human interactions serve as the context.
Host/Chair: Andrew Lippman
Vincent Chan, Gerald Sussman