Premium
Multiple visual targets tracking via probability hypothesis density filter and feature measurement
Author(s) -
Lu Xiaofeng,
Izumi Takashi,
Teng Lin,
Wang Lei
Publication year - 2014
Publication title -
ieej transactions on electrical and electronic engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.254
H-Index - 30
eISSN - 1931-4981
pISSN - 1931-4973
DOI - 10.1002/tee.22026
Subject(s) - filter (signal processing) , feature (linguistics) , artificial intelligence , local binary patterns , computer science , computer vision , tracking (education) , bayesian probability , particle filter , pattern recognition (psychology) , frame (networking) , eye tracking , gaussian , monte carlo method , posterior probability , fuse (electrical) , mathematics , image (mathematics) , histogram , statistics , engineering , psychology , telecommunications , pedagogy , philosophy , linguistics , physics , quantum mechanics , electrical engineering
The probability hypothesis density (PHD) filter is a practical alternative to the optimal Bayesian multiple‐targets filter based on random finite sets (RFS). It propagates the posterior intensity of the random sets of targets. In this paper, we apply the PHD filter to track a random number of moving targets in visual sequences. The PHD filter is implemented using a Gaussian mixture. Obtaining the PHD only for one visual frame at a time remains a challenge. To meet this challenge, we propose a method to approximate the posterior intensity using feature measurement. To improve the representability of tracking target, we adopt an adaptive weight to fuse the color and local binary pattern (LBP) features which are extracted by the Monte Carlo method. Experimental results demonstrate the effectiveness of our method. © 2014 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.