Unsupervised intrusion detection using color images

Authors:
Grant Cermak;Karl Keyzer
Affiliations:
Institute of Technology, University of Minnesota, Minneapolis, MN;Institute of Technology, University of Minnesota, Minneapolis, MN
Venue:
ISVC'07 Proceedings of the 3rd international conference on Advances in visual computing - Volume Part II
Year:
2007

Citing 4
Cited 1

Training Templates for Scene Classification using a Few Examples

CAIVL '97 Proceedings of the 1997 Workshop on Content-Based Access of Image and Video Libraries (CBAIVL '97)
Video Surveillance of Interactions

VS '99 Proceedings of the Second IEEE Workshop on Visual Surveillance
Visual target tracking and extraction from a sequence of images

ICTAI '97 Proceedings of the 9th International Conference on Tools with Artificial Intelligence
MPEG-2 decoding in a stream programming language

IPDPS'06 Proceedings of the 20th international conference on Parallel and distributed processing

Fast background subtraction using static and dynamic gates

Artificial Intelligence Review

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper presents a system to monitor a space and detect intruders. Specifically, the system analyzes color video to determine if an intruder entered the space. The system compares any new items in a video frame to a collection of known items (e.g. pets) in order to allow known items to enter and leave the space. Simple trip-line systems using infrared sensors normally fail when a pet wanders into the path of a sensor. This paper details an adaptation of the mean shift algorithm (described by Comaniciu et al.) in RGB color space to discern between intruders and benign environment changes. A refinement to the histogram bin function used in the tracking algorithm is presented which increases the robustness of the algorithm.