Generative Tracking of Human Motion by Sequential Clonal Selection Algorithm

High dimensional pose state space is the main challenge in articulated human motion tacking. In this paper, we propose a novel generative approach in the framework of artificial immune model, by which we try to widen the bottleneck with effective search strategy embedded in the extracted state subspace. Firstly, we learn the latent space of pose state and propose a manifold reconstruction method to establish the inverse mapping. Pose analysis in this latent space is more effective and accurate. Secondly, we apply Clone Selection Algorithm (CSA) for human pose estimation. In order to make CSA suitable for pose tracking, we propose a sequential CSA (SCSA) framework by incorporating the temporal continuity information into the traditional CSA. Experimental results show that our method achieves better results than state-ofart methods.