Model Driven Segmentation of Articulating Humans in Laplacian Eigenspace

We propose a general approach using Laplacian Eigenmaps and a graphical model of the human body to segment 3D voxel data of humans into different articulated chains. In the bottom-up stage, the voxels are transformed into a high-dimensional (6D or less) Laplacian Eigenspace (LE) of the voxel neighborhood graph. We show that LE is effective at mapping voxels on long articulated chains to nodes on smooth 1D curves that can be easily discriminated, and prove these properties using representative graphs. We fit 1D splines to voxels belonging to different articulated chains such as the limbs, head and trunk, and determine the boundary between splines using the spline fitting error. A top-down probabilistic approach is then used to register the segmented chains, utilizing their mutual connectivity and individual properties. Our approach enables us to deal with complex poses such as those where the limbs form loops. We use the segmentation results to automatically estimate the human body models. While we use human subjects in our experiments, the method is fairly general and can be applied to voxel-based segmentation of any articulated object composed of long chains. We present results on real and synthetic data that illustrate the usefulness of this approach.

1 Introduction

Human motion capture and analysis has applications in different fields such as kinesiology, biomechanics, surveillance, human-computer interaction, animation, and videogames. There is a correspondingly large body of literature on human motion analysis and pose estimation from video data. However, the requirements in terms of the detail of pose parameters and accuracy in estimation vary from application to application, as does the form of the available input data. Surveillance applications, for instance, usually require just the location of the subject or an approximate estimate of human pose from a single video stream, whereas biomechanical applications require accurate pose estimates of different joint angles from images obtained using multiple video cameras. The most common methods for accurate capture of 3D human movement require attachment of markers, fixtures, or sensors to body segments. These methods are invasive; i.e., they encumber the subject, hinder movement, and require subject preparation time. Biomechanical and clinical applications [1], [2] require the accurate capture of normal and pathological human movement without the artifacts associated with current state-of-the-art marker-based motion capture techniques. A markerless motion capture system using multiple video streams therefore possesses several advantages over marker-based systems and is highly desirable.