The University of Edinburgh -
Division of Informatics
Forrest Hill & 80 South Bridge

MSc Thesis #9534

Title:Periodic Activity Detection
Date: 1995
Abstract:This paper describes an investigation into the feasibility of a recently proposed activity detection scheme. The scheme detects an activity by applying a periodicity measure to the resulting signatures left in the spatio-temporal solid. The investigation focused on two different methods for extracting the signatures caused by simple periodic activities. At first, motion was detected and tracked by considering the difference between successive frames in a sequence. The second method involved implementing Horn and Schunck's optical flow algorithm to locate and track motion through an image sequence. The performance of the two methods and the application of the periodicity measure was then investigated by using synthetic and real image sequences. Although the signatures produced were very noisy, the periodicity measure was still able to identify the signatures which corresponded to periodic motion within the spatio-temporal solid.

[Search These Pages] [DAI Home Page] [Comment]