Vision-based System
 | Video Surveillance | VISUAL SURVEILLANCE SYSTEMS WHICH CAN UNDERSTAND OF HUMAN ACTIONS INTELLIGENTLY IN REAL-TIME | |
|---|
Anti-Terrorism has been a global issue and video surveillance has become more and more popular in public places such as elevators, banks, airports and squares. The fundamental problem is the understanding of human actions intelligently in real-time. We present a low-cost PC based real-time surveillance system to model and analyze human actions based on learning by demonstration. By teaching the system the differences between normal and abnormal human actions, the computational models built inside the trained machines can automatically identify whether the newly observed behaviors require security interference.
In our project, we present two approaches to detect abnormal behaviors. The first one employs Principal Component Analysis for feature selection and Support Vector Machine for classification of human behaviors. The feature selection method is based on the border information of four consecutive blobs. The second approach computes optical flow to obtain the velocity of each pixel for determining whether a human behavior is normal or not. Both algorithms are successfully implemented in crowded environments for detecting the human abnormal behaviors, such as:
a) running people in a crowded environment;
b) bending down movement while most are walking or standing;
c) a person carrying a long bar;
d) a person waving hand in the crowd.
| Key Investigators: Prof. Yangsheng Xu, Weizhong Ye, Xinyu Wu, Zhi Zhong, Xi Shi |
| |
|
