Abstract | ||
---|---|---|
Higher frame-rates promise better tracking of rapid motion, but advanced real-time vision systems rarely exceed the standard 10–60Hz range, arguing that the computation required would be too great. Actually, increasing frame-rate is mitigated by reduced computational cost per frame in trackers which take advantage of prediction. Additionally, when we consider the physics of image formation, high frame-rate implies that the upper bound on shutter time is reduced, leading to less motion blur but more noise. So, putting these factors together, how are application-dependent performance requirements of accuracy, robustness and computational cost optimised as frame-rate varies? Using 3D camera tracking as our test problem, and analysing a fundamental dense whole image alignment approach, we open up a route to a systematic investigation via the careful synthesis of photorealistic video using ray-tracing of a detailed 3D scene, experimentally obtained photometric response and noise models, and rapid camera motions. Our multi-frame-rate, multi-resolution, multi-light-level dataset is based on tens of thousands of hours of CPU rendering time. Our experiments lead to quantitative conclusions about frame-rate selection and highlight the crucial role of full consideration of physical image formation in pushing tracking performance. |
Year | DOI | Venue |
---|---|---|
2012 | 10.1007/978-3-642-33786-4_17 | ECCV (7) |
Keywords | Field | DocType |
fundamental dense whole image,better tracking,tracking performance,high frame-rate,frame-rate selection,cpu rendering time,camera tracking,physical image formation,real-time camera tracking,application-dependent performance requirement,image formation | Computer vision,BitTorrent tracker,Computer science,Shutter,Motion blur,Image formation,Robustness (computer science),Artificial intelligence,Frame rate,Rendering (computer graphics),Computation | Conference |
Volume | ISSN | Citations |
7578 | 0302-9743 | 43 |
PageRank | References | Authors |
2.68 | 14 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ankur Handa | 1 | 479 | 26.11 |
Richard A. Newcombe | 2 | 3003 | 111.68 |
Adrien Angeli | 3 | 291 | 11.87 |
Andrew J. Davison | 4 | 6707 | 350.85 |