The research leading to these results has received funding from the European Union’s Seventh Framework Programme (FP7/2007-2013) under grant agreement n° 248138 Format-agnostic Approach for Production, Delivery and Rendering of Immersive Media Felix Lee HPC I 28.10.2013
The research leading to these results has received funding from the European Unions Seventh Framework Programme (FP7/2007-2013) under grant agreement n°
Mar 30, 2015
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
The research leading to these results has received funding from the European Union’s Seventh Framework Programme (FP7/2007-2013) under grant agreement n° 248138
Format-agnostic Approach for Production, Delivery and Rendering
of Immersive Media
Felix Lee
HPC I28.10.2013
© FascinatE Consortium - 2 - www.fascinate-project.eu
Real-time Live Panoramic VideoOMNICAM
real-time acquisition
warping & stitching
blending and rendering of 6 HD video streams
one panoramic video of 7k x 2k resolution
on a single PC exploiting GPU processing
© FascinatE Consortium - 3 - www.fascinate-project.eu
Scene Acquisition (1)
OMNICAMARRI Alexa
HD Broadcast camera
© FascinatE Consortium - 4 - www.fascinate-project.eu
Scene Acquisition (2)
Soundfield® Stereo pair
Gun mics
Eigenmike®
© FascinatE Consortium - 5 - www.fascinate-project.eu
Football Premier League Game
Chelsea FC – Wolverhampton Wanderers (London, October 2010)
Dance Performance Carmen Suite
Berlin Philharmonic Orchestra conducted by Sir Simon Rattle & Dance Compagnie Sasha Waltz (Berlin, May 2012)
Concert BBC Proms (London, August 2012)
Three Demo Productions (so far)
© FascinatE Consortium - 6 - www.fascinate-project.eu
Saliency Map Center-surround model
inspired from biological models of human visual sensory system Center-surround features
Pixel intensity
Color chromatic opponents RG and BY
Saliency map for still images
© FascinatE Consortium - 7 - www.fascinate-project.eu
Saliency Detection
Image pyramid based saliency map Create image pyramid: blur and shrink the
image Build the image differences Add the image differences
© FascinatE Consortium - 8 - www.fascinate-project.eu
Spatio-temporal Saliency Histograms
Spatiotemporal saliency model Split image into grid cells (cell size: e.g. 20x20 pixels for HD) Compute histogram of pyramid differences for each cell Compare histograms
Current histogram vs. average of histograms from past second Distance function: earth mover distance
Region of interest (ROI) Convert high salient regions into rectangles
Application: automatic view selection
Demo video
Related Documents
