Moving Target Monitoring in Video Surveillances
Published:
Intelligent Video Surveillance System is a new research area which attracts many researchers. By analyzing image sequences captured from cameras, we can detect moving objects from its background. The classification technology helps us choose interested target while the tracking technology enables us to analyze its action. By this means, we achieve our our objective of setting up an automatic early warning system for anomalous events.
In this project, we intend to detect anomalous cross-border actions in a real-time surveillance system. The algorithm can be divided into four parts:
- Gaussian Mixture Models: separate the foreground from the background
- Morphology filtering: analyze connected domains
- Camshift tracking algorithm: track area of interest
- Cross-border warning: send an alert if anomalous events happen
Implementation
- Python 2.7
- OpenCV 2.4.0