A New Algorithm for Human Motion Capture via 3D Active Contours

Abstract

Motion capture is one of the most challenging problems in computer vision. In this paper, we propose a new algorithm for markerless human body motion capture. We compute volume data (voxels) representation from the images using the method of SFS (shape from silhouettes). Then we match a predefined human body model with pose parameter to the volume data, and the calculation of this matching is transformed into energy function minimization. In minimizing the energy function, we use a method of 3D active contours to solve this problem. In the process of curving surface evolution, the curving surface will drive the human model close to the visual hull. On the other hand, when the human model is superposed with the human real pose, the curving surface can create a 3D human body reconstruction based on the visual hull and human model. Promising results on real images demonstrate the potentials of the presented method.