Researchers develop surveillance

CIS professors received an NSF grant to develop mobile 3-D cameras.

Sunil Wattal and Youngjin Yoo talk to Zoran Obradovic via Skype to discuss research. | Allan Barnes TTN
Sunil Wattal and Youngjin Yoo talk to Zoran Obradovic via Skype to discuss research. | Allan Barnes TTN

In an effort to incentivize the development of innovative uses for high-speed computer networks, the National Science Foundation recently offered a grant of nearly $200,000 to the Computer and Information Sciences Department, which is developing mobile surveillance cameras and enhancing Temple’s security network to accommodate the new technology.

The NSF’s U.S. Ignite initiative extends the opportunity for innovation to universities and colleges, particularly those in urban environments. A team including Jie Wu, chair of CIS, is developing a 3-D camera mountable on a police cruiser dashboard, along with improving Temple’s network to allow footage to be viewed by police dispatchers.

Chiu Tan, assistant professor of CIS, said mounted security cameras have a few notable deficiencies.

“Right now, a lot of cameras are used after [an incident],” Tan said. “Dashboard cameras record, then dump the information [to a database], but it’s not used in real time. … We would like to make use of some of this information while it’s being collected.”

Mounted cameras also provide limited coverage of an area, Tan added.

Haibin Ling, associate professor at the College of Science and Technology said a typical camera cannot recognize an image in poor lighting.

3-D cameras function just like regular cameras, but can also detect depth in addition to the three primary colors. The camera is able to detect how far away something is – and in the case of a moving object, how fast it’s approaching.

“An interesting and desirable function about this camera is no matter the lighting, it will always give you this information because it only uses depth,” Ling said. “[The 3-D camera] can detect things regardless of illumination conditions.”

The second component to the project is maximizing the efficiency of the wireless network.

Tan, who is the facilitator of the streaming component, said uploading and streaming video fast enough to be in real time is a challenge because of the large amount of data that videos take up.

The eventual goal of a more effective network is data compression and summarization. This will assist in investigations of crimes or incidents.

“Our system includes two parts: one is outside identification in real time, the second is simple processing with a computer,” Wu said.

Wu also said that the identification component will use pattern recognition, which then matches with existing databases to find the suspect in the case of a crime.

Ling is the coordinator for the recognition component of the project.

The network is facilitated by the WiMax infrastructure, which is the extension of Wi-Fi. It is installed on top of Wachman Hall and covers around a two-mile radius, allowing complete coverage of Main Campus.

“Surveillance videos can be up to hundreds of hours of footage. Most of the time, there’s nothing in the video; it’s static,” Ling said. “We can usually discard 90 percent of the video. With more research and technology, we can easily reduce that to one hour.”

The camera technology and increased network efficiency will be used toward a more proactive approach to campus safety.

“Our initial goal is to install some of these 3-D cameras on campus systems from campus police,” Wu said.

Wu added that there will eventually be around 600 cameras installed to monitors at Temple. Additional cameras will be installed on Temple police cars for dynamic, mobile coverage of events.

GoPro cameras are the current models for testing the 3-D capabilities before prototypes are built.

The project was approved in October and is on a two-year timeline. The approach and detection camera technology will be ready in May or June of this year. The initial data summarization will be completed in January 2016, with an advanced version in May 2016.

Next month, the team will go to Washington D.C. to present its preliminary results.

Lian Parsons can be reached at lian.parsons@temple.edu or on Twitter @Lian_Parsons.

Be the first to comment

Leave a Reply

Your email address will not be published.


*