A real-time anomaly intrusion and theft items detecting system for surveillance videos

This paper proposes an embedded surveillance system for real-time anomaly intrusion detection based on temporal difference algorithm and theft items detection based on accumulated background subtraction algorithm. This design of modified vision algorithm fully utilize the advanced parallelism of Field Programmable Gate Arrays (FPGA) and this hardware implementation realizes time-consumed difference computing with on-chip FIFO and RAM memory. Finally, we fuse these two anomaly detection algorithms in one FPGA and select the algorithm type by user needs. As a result, the detecting validity and robustness of this implementation have been demonstrated through real-time surveillance videos.