Behavior Monitoring for Assistive Environments Using Multiple Views
By Kosmopoulos, Dimitrios I.; Universal Access in the Information Society, Vol. 10, No. 2, pp. 115-123Publication Date: June 2011
Paper presents a methodology for modeling and understanding human behavior using multiple views in real time. This approach is appropriate for monitoring people in an assistive environment for the purpose of issuing alerts in cases of abnormal behavior. Abnormalities in the established patterns for short-term behavior and motion for a person may be indicative of problems; for example, people with mild cognitive impairment may show increased day to day variability in their activities at home. The proposed system processes video streams from several cameras with overlapping fields of view. The output of multiple classifiers is used to model and extract abnormal behavior from both the target trajectory and the target short term activity, such as walking, running, and abrupt motion. Spatial information is obtained after an offline camera registration using homography information. The proposed approach was verified experimentally in an indoor environment. The experiments were performed with a single moving target; however, the method can be generalized to multiple moving targets which may occlude each other. The multiple cameras allow for efficient handling of occlusions. The presented model is closer to human perception compared to pure image based techniques, due to separate handling of the two information sources. Future steps in the development of the system are discussed.
Published by: Springer Publishing Company (Website:http://www.springerpub.com)
Link to text: http://www.springerlink.com/content/l273816232751727/fulltext.html
