Output details
15 - General Engineering
University of Sussex
Adaptive Sample Count Particle Filter
A method of adaptively controlling the sample count of a particle filter was developed. Additionally, an active area of support around the tracked object is employed which allows a more accurate colour histogram calculation. These methods are novel and reported here for the first time. Extensive experimental assessment is reported using thousands of frames of video from different scenarios. Tracking accuracy comparable to existing methods is achieved but the filter can be calculated up to five times faster. This contributed significantly to the author’s DSTL funded tracking research; rated highly upon completion: Contact Dr Peter Harvey, Tel: 01980616407, E-mail: prharvey@mail.dstl.gov.uk.