Video Stabilization Algorithm from Low Frame Rate Video for Hyperlapse Applications
(2015) In Master’s Theses in Mathematical Sciences FMA820 20151Mathematics (Faculty of Engineering)
- Abstract
- There are several methods that one can use to visualize image sequences. One such method, called timelapse, is based on synthesizing a video from the image sequence. One sub category of timelapses is the so-called hyperlapse, which is defined as a timelapse with a camera movement over great space. A problem with combining camera movement with speeding up the frame rate per second is that camera shakes appear magnified. One way to minimize this problem is to stabilize the video, using estimated relative camera movement. Such estimates can be obtained using computer vision methods based on epipolar geometry. Choosing how to compensate for camera shakes and calculate a new, more smooth camera path is essential to the video stabilization... (More)
- There are several methods that one can use to visualize image sequences. One such method, called timelapse, is based on synthesizing a video from the image sequence. One sub category of timelapses is the so-called hyperlapse, which is defined as a timelapse with a camera movement over great space. A problem with combining camera movement with speeding up the frame rate per second is that camera shakes appear magnified. One way to minimize this problem is to stabilize the video, using estimated relative camera movement. Such estimates can be obtained using computer vision methods based on epipolar geometry. Choosing how to compensate for camera shakes and calculate a new, more smooth camera path is essential to the video stabilization algorithm. One aim of this thesis is to create such a video stabilization algorithm. Another aim is to examine how performance degrades with decreased frame rate for the input sequence. Along with this thesis we have collected a set of benchmark image sequences. Several different video stabilization algorithms have been developed in the project. These have all been tested on the benchmark data sets and evaluated with promising results. (Less)
- Popular Abstract (Swedish)
- I dagens samhälle är vi alltmer ivriga att dokumentera och dela våra upplevelser och vår vardag med andra genom sociala medier. Ett nytt sätt att göra detta har utvecklats av Narrative som med sin smidiga kamera, vilken kan fästas på dina kläder, erbjuder dig ett verktyg att dokumentera händelser utan att du behöver anstränga dig. Men om man vill presentera bilderna som en video, går det? Det är frågan som har legat bakom vårt examensarbete.
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/5462251
- author
- Hansson, Björn LU and Tengbom Zetterman, Kim LU
- supervisor
-
- Karl Åström LU
- organization
- course
- FMA820 20151
- year
- 2015
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- computer vision, mathematics, hyperlapse, video stabilization, epipolar geometry
- publication/series
- Master’s Theses in Mathematical Sciences
- report number
- LUTFMA-3274-2015
- ISSN
- 1404-6342
- other publication id
- 2015:E14
- language
- English
- id
- 5462251
- date added to LUP
- 2015-06-18 12:05:43
- date last changed
- 2015-06-18 12:05:43
@misc{5462251, abstract = {{There are several methods that one can use to visualize image sequences. One such method, called timelapse, is based on synthesizing a video from the image sequence. One sub category of timelapses is the so-called hyperlapse, which is defined as a timelapse with a camera movement over great space. A problem with combining camera movement with speeding up the frame rate per second is that camera shakes appear magnified. One way to minimize this problem is to stabilize the video, using estimated relative camera movement. Such estimates can be obtained using computer vision methods based on epipolar geometry. Choosing how to compensate for camera shakes and calculate a new, more smooth camera path is essential to the video stabilization algorithm. One aim of this thesis is to create such a video stabilization algorithm. Another aim is to examine how performance degrades with decreased frame rate for the input sequence. Along with this thesis we have collected a set of benchmark image sequences. Several different video stabilization algorithms have been developed in the project. These have all been tested on the benchmark data sets and evaluated with promising results.}}, author = {{Hansson, Björn and Tengbom Zetterman, Kim}}, issn = {{1404-6342}}, language = {{eng}}, note = {{Student Paper}}, series = {{Master’s Theses in Mathematical Sciences}}, title = {{Video Stabilization Algorithm from Low Frame Rate Video for Hyperlapse Applications}}, year = {{2015}}, }