Live RGB-D camera tracking for television production studios

► A novel low-cost tool for camera tracking in broadcasting studio environments. ► Driftless tracking with keyframes. ► Real-time performance using a GPU. ► Allows moving actors in the scene while tracking. ► Comparison with Kinfu. In this work, a real-time image-based camera tracker is designed for...

Full description

Saved in:
Bibliographic Details
Published in:Journal of visual communication and image representation 2014-01, Vol.25 (1), p.207-217
Main Authors: Tykkälä, Tommi, Comport, Andrew I., Kämäräinen, Joni-Kristian, Hartikainen, Hannu
Format: Article
Language:English
Subjects:
Augmented reality
Automatic
Cameras
Computer Science
Computer Vision and Pattern Recognition
Cost function
Dense 3D reconstruction
Dense tracking
Drift
Engineering Sciences
GPU
Image registration
Parallel computing
Real time
RGB-D
Robotics
Signal and Image processing
Static models
Studios
Television
Tracking
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:► A novel low-cost tool for camera tracking in broadcasting studio environments. ► Driftless tracking with keyframes. ► Real-time performance using a GPU. ► Allows moving actors in the scene while tracking. ► Comparison with Kinfu. In this work, a real-time image-based camera tracker is designed for live television production studios. The major concern is to decrease camera tracking expenses by an affordable vision-based approach. First, a dense keyframe model of the static studio scene is generated using image-based dense tracking and bundle adjustment. Online camera tracking is then defined as registration problem between the current RGB-D measurement and the nearest keyframe. With accurate keyframe poses, our camera tracking becomes virtually driftless. The static model is also used to avoid moving actors in the scene. Processing dense RGB-D measurements requires special attention when aiming for real-time performance at 30Hz. We derive a real-time tracker from our cost function for a low-end GPU. The system requires merely a RGB-D sensor, laptop and a low-end GPU. Camera tracking properties are compared with KinectFusion. Our solution demonstrates robust and driftless real-time camera tracking in a television production studio environment.
ISSN:1047-3203
1095-9076
DOI:10.1016/j.jvcir.2013.02.009