Cryptography Reference
In-Depth Information
9
Intelligent Video Event Detection
for Surveillance Systems
Hong-Yuan Mark Liao
1
, Duan-Yu Chen
1
, Chih-Wen Su
1
, and Hsiao-Rong
Tyan
2
1
128, Section 2, Academy Road, Institute of Information Science, Academia
Sinica, Nankang, Taipei, Taiwan
liao, dychen, lucas
@iis.sinica.edu.tw
2
200, Chung Pei Rd., Chung Li, Department of Information and Computer
Engineering, Chung-Yuan Christian University, Taiwan
tyan@ice.cycu.edu.tw
Summary.
In recent years, real-time direct detection of events by surveillance sys-
tems has attracted a great deal of attention. In this chapter, we present solutions
for video-based surveillance systems in the spatial domain and in the compressed
domain, respectively. In spatial domain, we propose a new video-based surveillance
system that can perform real-time event detection. In the background modelling
phase, we adopt a mixture of Gaussian approach to determine the background.
Meanwhile, we use color blob-based tracking to track foreground objects. Due to
the self-occlusion problem, the tracking module is designed as a multi-blob tracking
process to obtain similar multiple trajectories. We devise an algorithm to merge
these trajectories into a representative one. After applying the Douglas-Peucker al-
gorithm to approximate a trajectory, we can compare two arbitrary trajectories.
The above mechanism enables us to conduct real-time event detection if a number
of wanted trajectories are pre-stored in a video surveillance system. In compressed
domain, we propose the use of motion vectors embedded in MPEG bitstreams to
generate so called motion-flows, which are applied to perform quick video retrieval.
By using the motion vectors directly, we do not need to consider the shape of a mov-
ing object and its corresponding trajectory. Instead, we simply link the local motion
vectors across consecutive video frames to form motion flows, which are then anno-
tated and stored in a video database. In the video retrieval phase, we propose a new
matching strategy to execute the video retrieval task. Motions that do not belong
to the mainstream motion flows are filtered out by our proposed algorithm. The re-
trieval process can be triggered by a query-by-sketch (QBS) or a query-by-example
(QBE). The experimental results show that our method is e
cient and accurate in
the video retrieval process.