With the proliferation of cameras in handheld devices that allows users to capture still images and videos, providing users with software tools to efficiently manage multimedia content has become essential. In many cases users desire to organize their personal media content using high-level semantic labels. In this paper we will describe low-complexity algorithms that can be used to derive semantic labels, such as "indoor/outdoor," "face/not face," and "motion/not motion" for mobile video sequences. We will also describe a method for summarizing mobile video sequences. We demonstrate the classification performance of the methods and their computational complexity using a typical processor used in many mobile terminals.

Date Published: 10 February 2006
PDF: 12 pages
Proc. SPIE 6074, Multimedia on Mobile Devices II, 607408 (10 February 2006); doi: 10.1117/12.643246

Published in SPIE Proceedings Vol. 6074:
Multimedia on Mobile Devices II
Reiner Creutzburg; Jarmo H. Takala; Chang Wen Chen, Editor(s)