Home // VISUAL 2019, The Fourth International Conference on Applications and Systems of Visual Paradigms // View article

Model-Based 3D Visual Tracking of Rigid Bodies using Distance Transform

Authors:
Marios Loizou
Paris Kaimakis

Keywords: Object 3D tracking; Model-based; Gauss-Newton optimization; Distance Transform

Abstract:
The core idea of model-based 3D tracking is that of continuously estimating the pose parameters of a 3D object throughout a sequence of images, e.g., a video feed. Here, we present an edge-based method for achieving 3D object tracking, via Gauss-Newton optimization. We rely on natural features observations, like edges, for the detection of interest points and by using the 3D pose of the object in the previous frame, we correctly estimate its new 3D position and orientation, in real-time. There is also a C++ implementation of the visual tracking system, with the use of the OpenCV library, which can be found in our GitHub repository.

Pages: 39 to 46

Copyright: Copyright (c) IARIA, 2019

Publication date: June 30, 2019

Published in: conference

ISSN: 2519-8645

ISBN: 978-1-61208-724-5

Location: Rome, Italy

Dates: from June 30, 2019 to July 4, 2019