Campus Event Calendar
Event Entry
What and Who
Video Compression using PDEs and Optic Flow
Qi Gao
IMPRS-CS
Master Seminar Talk
IMPRS Master Student
AG 1, AG 2, AG 3, AG 4, AG 5, SWS, RG1, RG2
Public Audience
English
Note: We use this to send email in the morning.
Date, Time and Location
Tuesday, 17 June 2008
13:05
60 Minutes
E1 3 - Hörsaal Gebäude
003
Saarbrücken
Warning: New Building!
Warning: New Room!
Abstract
Video compression/decompression can be done by a variety of methods. Partial differential equations (PDEs), however, have been little explored in this context. We introduce a novel video compression/decompression scheme that makes use of edge enhancing diffusion (EED) based interpolation and high accuracy optic flow based motion compensation. Firstly, we perform B-tree triangular coding (BTTC)for each intra-coding (I) frames. Next, high accuracy optic flow between neighboring frames is calculated and then encoded in a lossy manner.
Finally, from I frames, we estimate the adjacent frame one by one through motion compensation by decoded optic flow, then encode the differences between these estimated frames and their originals. Thus, most time redundancy is removed. When decompressing, we interpolate each I frame using EED and then reconstruct the adjacent frame one by one using decoded optic flow and differences data. Experimental results show superior performance against the method only using BTTC and EED for each frame.
Finally, from I frames, we estimate the adjacent frame one by one through motion compensation by decoded optic flow, then encode the differences between these estimated frames and their originals. Thus, most time redundancy is removed. When decompressing, we interpolate each I frame using EED and then reconstruct the adjacent frame one by one using decoded optic flow and differences data. Experimental results show superior performance against the method only using BTTC and EED for each frame.
Contact
Stephanie Jörg
(06 81) 93 25 225
--email hidden
passcode not visible
logged in users only
Stephanie Jörg, 06/11/2008 11:24 -- Created document.