Markerless Human Motion Capture Using Hierarchical Particle Swarm Optimisation Vijay John, Spela Ivekovic, and Emanuele Trucco School of Computing, University of Dundee, Dundee, U.K. {vijayjohn,spelaivekovic,manueltrucco}@computing.dundee.ac.uk Abstract. In this paper, we address full-body articulated human motion tracking from multi-view video sequences acquired in a studio environment. The track- ing is formulated as a multi-dimensional nonlinear optimisation and solved using particle swarm optimisation (PSO), a swarm-intelligence algorithm which has gained popularity in recent years due to its ability to solve difficult nonlinear optimisation problems. Our tracking approach is designed to address the limits of particle filtering approaches: it initialises automatically, removes the need for a sequence-specific motion model and recovers from temporary tracking diver- gence through the use of a powerful hierarchical search algorithm (HPSO). We quantitatively compare the performance of HPSO with that of the particle filter (PF), annealed particle filter (APF) and partitioned sampling annealed particle filter (PSAPF). Our test results, obtained using the framework proposed by Balan et al [1] to compare articulated body tracking algorithms, show that HPSO’s pose estimation accuracy and consistency is better than PF, APF and PSAPF. 1 Introduction Tracking articulated human motion from video sequences is an important problem in computer vision with applications in virtual character animation, medical posture anal- ysis, surveillance, human-computer interaction and others. In this paper, we formulate the full-body articulated tracking as a nonlinear optimisation problem which we solve using particle swarm optimization (PSO), a recent swarm intelligence algorithm with growing popularity [3,2]. Because the full-body articulated pose estimation is a high-dimensional optimisation problem, we formulate it as a hierarchical PSO algorithm (HPSO) which exploits the in- herent hierarchy of the human-body kinematic model, thus reducing the computational complexity of the search. HPSO is designed to address the limits of the particle filtering approaches. Firstly, it removes the need for a sequence-specific motion model: the same algorithm with un- modified parameter settings is able to track different motions without any prior knowl- edge of the motion’s nature. Secondly, it addresses the problem of divergence, whereby the filter loses track after a wrongly estimated pose and is unable to recover unless in- teractively corrected by the user or assisted by additional, higher-level motion models [4]. In contrast, our tracking approach is able to automatically recover from an incorrect pose estimate and continue tracking. Last but not least, in line with its ability to recover A. Ranchordas et al. (Eds.): VISIGRAPP 2009, CCIS 68, pp. 343–356, 2010. c  Springer-Verlag Berlin Heidelberg 2010