| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 4957530 | Pervasive and Mobile Computing | 2017 | 18 Pages |
Abstract
A video playing on a television screen emits a characteristic flickering, which serves as an identification feature for the video. This paper presents a method for video recognition by sampling the ambient light sensor of a smartphone or wearable computer. The evaluation shows that given a set of known videos, a recognition rate of up to 100% is possible by sampling a sequence of 15 to 120 s length. Our method works even if the device has no direct line of sight to the television screen, since ambient light reflected from walls is sufficient. A major factor of influence on the recognition is the number of video cuts that change the light emitted by the screen.
Related Topics
Physical Sciences and Engineering
Computer Science
Computer Networks and Communications
Authors
Lorenz Schwittmann, Christopher Boelmann, Viktor Matkovic, Matthäus Wander, Torben Weis,