Communications of the ACM
Hong Kong Researchers Develop 3D Tool For Surveillance Footage
Yu Yizhou (left) and Calvin Fong (right) led a research team in the Department of Computer Science at the University of Hong Kong that developed software for reconstruction of past events in three-dimensional space from surveillance videos.
Credit: University of Hong Kong
Researchers at the University of Hong Kong have developed a digital system designed to reduce the amount of time it takes investigators to search through surveillance camera footage for suspects.
The system merges footage taken from different angles to create a three-dimensional reconstruction of the scene, which enables investigators to simultaneously see different angles and more easily identify individuals of interest. Once investigators identify potential suspects, the system scans available surveillance footage and uses the color and texture of the suspects' skin, clothes, personal belongings, and gait to identify them in any available video.
During testing, the system had an accuracy rate of about 90 percent. However, the researchers say frequently crowded areas will reduce the accuracy to around 80 percent, although they say the use of more cameras could increase that percentage.
The researchers also note the system has yet to be tested using footage below the standard definition quality or on video taken outdoors. They plan to continue making improvements, including adding more ways for the system to identify an individual and reducing the time it takes for the system to scan footage.
From South China Morning Post (Hong Kong)
View Full Article
Abstracts Copyright © 2014 Information Inc., Bethesda, Maryland, USA
No entries found