The Encoder
The encoder compresses the frame buffer after processing has been completed by Core B. All the while, the system can output frame buffers modified by Core B through the other PPI to display real-time tracking results. The object tracking algorithm is a real-time function, and thus does not cause a delay in video encoding.
When no moving object is detected, the encoder may work in a state of low code rate and/or low frame rate, or even stop encoding altogether. Once moving objects of a specific classification are detected, the encoder starts working immediately, uploading the compressed code stream and corresponding time to the administrative system through an Ethernet interface.
The system may set up a surveillance region boundary to identify the surveillance area of invasion detection and transmit the area through UART or Ethernet interfaces to the console, and then when a moving object crosses the boundary, the system will send an alarm. The console may send instructions to this smart surveillance endpoint, to vary its functions, from smart tracking to intrusion detection, and then to left-object detection or people counting, etc. The implementation of such sophisticated and numerous functions in an embedded device was once beyond imagination, but that was before the strong processing power and flexibility of a Blackfin processor arrived.
Trends in Smart Video Surveillance
Although video analytics have already improved the functionality of video surveillance, increased complexity and functionality is on the horizon. Gun-fire detection and associated suspect tracking capabilities are targets for the next generation of video surveillance and analysis. Interconnected smart video surveillance systems with combinations of audio/video detection, vision/non-vision imaging, and object tracking/behavior analysis/feature recognition, will be used to identify and track suspects and accomplices with pin-point accuracy. These individual technologies are all currently being developed, and advancing. With time systems will integrate all of this technology, requiring even more intelligent endpoints, high capacity networks and intelligent supervisory systems.
About the authors:
Harry Wei is Senior Technical Application Engineer – DSP/Embedded Processor for Analog Devices, Inc. He can be reached at Harry.Wei@analog.com
Michael L. Long is Product Line Manager – Industrial Video and Imaging Solutions for Analog Devices, Inc. He can be reached at Michael-l.long@analog.com
