A project study for extracting suspicious packages from surveillance video programs

A project for extracting suspicious packages from surveillance video programs is proposed in this paper. Three methods: improved RGB deduction, improved Hidden Markov Model and template matching, were used in this detecting and recognizing process. A simulate environment where a man with a parcel walking at the entrance of subway was set up for testing. A prototype system for unattended-package detection and recognition is implemented based on VC++ and Open CV. Experiment results show that the target package can be extracted successfully under this special circumstance. These results can be used for reference in building intelligent video surveillance systems.