Automatic Video Based Surveillance System for Abnormal Behavior Detection

Security and surveillance are important issues in todays world. Any behavior which is uncommon in occurrence and deviates from normally understood behavior can be termed as suspicious. This model aims at automatic detection of abnormal behavior in surveillance videos. We have targeted to create a system for the recognition of human activity and behavior, and extract new information of interest for end-users in highly secured indoor surveillance system. The objective for this project is to design a model for detection of abandoned objects and track abnormal human behaviors. The multi-object detection is done by background subtraction with the help of appropriate model and then recognizing the person using HOG feature and SVM classifier. Face of a detected person can be captured using ViolaJones algorithm. Finally, anomaly detection is done for tracking persons based on their individual appearance using Mean Shift Technique. For change detection, Mean-ratio and Log Ratio operators are used.