Title: | Real-Time Dense and Accurate Parallel Optical Flow using CUDA |
Authors: | Marzat, Julien Dumortier, Yann Ducrot, Andre |
Citation: | WSCG '2009: Full Papers Proceedings: The 17th International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision in co-operation with EUROGRAPHICS: University of West Bohemia Plzen, Czech Republic, February 2 - 5, 2009, p. 105-112. |
Issue Date: | 2009 |
Publisher: | Václav Skala - UNION Agency |
Document type: | konferenční příspěvek conferenceObject |
URI: | http://wscg.zcu.cz/WSCG2009/Papers_2009/!_WSCG2009_Full_final.zip http://hdl.handle.net/11025/10895 |
ISBN: | 978-80-86943-93-0 |
Keywords: | zpracování obrazu;monokulární vidění;optický tok;paralelní zpracování;grafický procesor |
Keywords in different language: | image processing;monocular vision;optical flow;parallel processing;graphic processing |
Abstract: | A large number of processes in computer vision are based on the image motion measurement, which is the projection of the real displacement on the focal plane. Such a motion is currently approximated by the visual displacement field, called optical flow. Nowadays, a lot of different methods are commonly used to estimate it, but a good trade-off between execution time and accuracy is hard to achieve with standard integrations. This paper tackles the problem by proposing a parallel implementation of the well-known pyramidal algorithm of Lucas & Kanade, in a Graphics Processing Unit (GPU). It is programmed using the Compute Unified Device Architecture from NVIDIA corporation, to compute a dense and accurate velocity field at about 15 Hz with a 640 480 image definition. |
Rights: | © Václav Skala - UNION Agency |
Appears in Collections: | WSCG '2009: Full Papers Proceedings |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
Marzat.pdf | Plný text | 4,02 MB | Adobe PDF | View/Open |
Please use this identifier to cite or link to this item:
http://hdl.handle.net/11025/10895
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.