Video Sensor Network Laboratory at UCR Receives Federal Funding
Posted on April 11, 2006 at 12:26:15 amWireless video networks – the kind of technology useful in disaster scene analysis, homeland security and search and rescue operations – may soon be in development at the University of California, Riverside thanks to a $250,500 award from the National Science Foundation.
The proposed Outdoor Video Sensor Network Laboratory facility, called Video Web, would be the first of its kind in the nation, according to Electrical Engineering Professor Bir Bhanu, who led the proposal effort with Bourns College of Engineering colleagues, Computer Science & Engineering Professor Chinya Ravishankar and Electrical Engineering Assistant Professor Amit Roy-Chowdhury. The two-year National Science Foundation grant took effect March 1 and expires in 2008.
“Suppose President Bush comes to UCR for a daylong visit? The (wired) video security and surveillance needed would cost hundreds of thousands of dollars,” Bhanu said. “After he leaves, that money is essentially wasted, but the wireless systems we’re talking about could be very flexible and assembled when and where needed, then used again.”
-ADVERTISEMENT-Bhanu and his colleagues are not just researching large networks of wireless cameras that would collect images, but they are developing algorithms and systems that would automatically analyze such images for activities and detect problems or possible threats, then warn its users.
“There is no technology available to detect and recognize people, vehicles or objects that could pose problems in an environment of large distributed wireless network of video sensors,” Bhanu said. “Or to control and manage animals in wilderness parks or large farms using not only the non-imaging sensors but also the network of videos. This requires continuous data collection and analysis, which is what we propose to do.”
The technologies needed require the expertise of faculty from many areas of engineering, such as computer vision, machine learning, image processing, pattern recognition and data mining, communication and control, network security, database and artificial intelligence. The goal is to develop integrated techniques that can work in a distributed environment and account for dynamic tradeoffs between processing and communication.
The laboratory will consist of 80 pan-tilt zoom video cameras that can be accessed over a network using IP addresses. Each camera would be connected to a computational unit that takes care of processing at the camera site. It will compress and transmit the data and identify the activities/objects. The whole system will be triggered by acoustic, seismic and vibration sensors. Fixed sensors are powered conventionally, while mobile ones will be powered by solar energy.
While the facility will provide a test-bed for many ongoing research projects, according to Bhanu, he foresees its work being applicable to faculty in multiple disciplines, including Electrical Engineering, Computer Science, Psychology and Biology.
“Had such technology existed when the Sago Mine disaster took place, the trapped miners could have established a video lifeline to the outside world,” Bhanu said.