Fast articulated motion tracking using . . .

DMCA

Fast articulated motion tracking using . . .

Cached

Download Links

by Carsten Stoll , Nils Hasler , Juergen Gall , Hans-peter Seidel , Christian Theobalt

Citations:

2 - 0 self

BibTeX

@MISC{Stoll_fastarticulated,
    author = {Carsten Stoll and Nils Hasler and Juergen Gall and Hans-peter Seidel and Christian Theobalt},
    title = {Fast articulated motion tracking using . . . },
    year = {}
}

OpenURL

Abstract

We present an approach for modeling the human body by Sums of spatial Gaussians (SoG), allowing us to perform fast and high-quality markerless motion capture from multi-view video sequences. The SoG model is equipped with a color model to represent the shape and appearance of the human and can be reconstructed from a sparse set of images. Similar to the human body, we also represent the image domain as SoG that models color consistent image blobs. Based on the SoG models of the image and the human body, we introduce a novel continuous and differentiable model-to-image similarity measure that can be used to estimate the skeletal motion of a human at 5-15 frames per second even for many camera views. In our experiments, we show that our method, which does not rely on silhouettes or training data, offers an good balance between accuracy and computational cost.

Keyphrases

eth zurich human body sog model multi-view video sequence good balance consistent image blob 5-15 frame many camera view color model image domain computational cost high-quality markerless motion capture sparse set differentiable model-to-image similarity measure spatial gaussians skeletal motion

DMCA