Home // CONTENT 2016, The Eighth International Conference on Creative Content Technologies // View article

Dynamic 3D Bounding Box Estimation for Video Segmentation from a Non-Stationary RGB-D Camera

Authors:
Naveed Ahmed

Keywords: RGB-D Video; Non-Stationary RGB-D Camera; Video Segmentation, 3D Animation.

Abstract:
We present a new method for video segmentation of RGB-D video data acquired from a non-stationary RGB-D camera. Our method uses a novel method of dynamic 3D bounding box estimation for moving objects that correctly classifies foreground and background elements of a scene even in the presence of high camera motion. Starting from the acquisition of a dynamic object using a non-stationary RGB-D camera, our method finds the mapping between different frames of the video data using the image features. The mapping between the image features derives our novel method of dynamic 3D bounding estimation that correctly segments the moving object in spite of the camera motion. The segmented video data can be employed in a number of applications, e.g., video editing, motion capture, action recognition, visual FX processing, or free-viewpoint video.

Pages: 25 to 29

Copyright: Copyright (c) IARIA, 2016

Publication date: March 20, 2016

Published in: conference

ISSN: 2308-4162

ISBN: 978-1-61208-464-0

Location: Rome, Italy

Dates: from March 20, 2016 to March 24, 2016