A new hierarchical method for markerless human pose estimation

Abstract

We present a system for markerless human motion capture through a hierarchical method from multiple camera views. In the absence of markers, the task of recovering the human pose is challenging and requires strong image features and robust algorithm. We propose a solution which integrates the 2D posture information and the volumetric reconstruction. Firstly, the model's initia posture is obtained through the method of segmenting silhouette. After that, we track the human pose by using a hierarchical method, which is divided into three steps: head detection, torso prediction and limb matching. In order to gain the robust results, we discard the interior voxel data, use the middle voxel data for motion tracking, and use the surface voxel data for global optimization. The experiment results show that the method is valid and robust. © 2013 Springer-Verlag.