Whats included:
Datasets:
Codes used in Coded Exposure:
A. Optimal codes for coded exposure for code length up to 100. Each of these codes are 50% on/off. For smaller code lengths (<=24), all possible codes can be tested to find the optimal code. For larger code lengths, we randomly search 1 million codes. The criteria of optimality is to maximize the minimum of the DFT of zero padded codes.
B. Matlab/C build to search for optimal code of length n for coded exposure.
Coded Exposure Deblurring (Matlab Code)
The matlab code shows the correct way of deblurring coded exposure images. Note that when the blur size is larger than the code length, deblurring does not result in deconvolution artifacts. Only the minimal resolvable blur size is increased. One should not see any ghosting or deconvolution artifacts.
We hope these datasets would be useful to students and researchers working in this field. Using high speed videos, several cameras can be simulated easily. For example, a traditional camera with a finite exposure can be simulated by simply averaging frames. A coded exposure camera can be simulated by adding frames according to the code. Camera motion can be simulated to a large extend by shifting the images according to camera motion before averaging. The datasets include high speed videos of a moving ISO resolution chart, which will be useful to evaluate the quality of deblurring algorithms/capture procedures. Noise analysis can be done using the homogeneous parts of the resolution chart. (See our CVPR 2009 paper for more details).
Dataset A: High Speed Videos
(Click on images to download .zip file of frames and associated matlab code)
| Moving ISO
Resolution Chart 1000 fps Motion: 0.28 pixels/frame, horizontal, left to right |
 |
| Moving Hairnet
Box 1000fps Motion: ~0.5pixels/frame, horizontal, right to left |
 |
| Outdoor Car Sequence |
Dataset B: Coded Exposure Images using Canon Camera and Ferro-Electric Shutter
(Download complete matlab code and all four input files)
| Indoor Toy
Train (118 pixel blur) |
 |
| Outdoor Green
Taxi (235 pixel blur) |
 |
| Outdoor White
Taxi (66 pixel blur) |
 |
| Outdoor Car
License
Plate (60 pixel blur) |
 |
Dataset C: Coded Exposure Images using Pointgrey Dragonfly2 Camera (Trigger mode 5)
| Moving Lena Poster | |
Dataset D: Varying Exposure Images using Canon Rebel XT (AEB mode)
(Download images directly)
| Face (Person
standing up) (Images already aligned) |
 Photo 1
|
 Photo 2
|
 Photo 3
|
| Outdoor Truck |
Dataset E: Varying Exposure Images using PointGrey Flea2 Camera
| Outdoor Car | |
| Indoor Toy Car with Occlusions |
Matlab/C Codes
1. Matlab Code
for Deblurring Coded Exposure Images in
SIGGRAPH 2006 paper
2. Matlab Code (Resolution
Chart, Moving
Box)
showing how to simulate a coded exposure image from
above high speed videos. The code also shows that blurred
image should
be resampled to code size before deblurring.
3. Matlab/C Code to
search for optimal
code of length n
Joint work with Yi Xu, Ramesh
Raskar and Jack
Tumblin
Related Papers in Motion/Focus Deblurring from our
group
SIGGRAPH 2006 Coded exposure for motion
deblurring
SIGGRAPH 2007 Coded
aperture
and
a new theory of light
field capture
SIGGRAPH 2009 Invertible
Motion
Blur
in Video
CVPR 2007
Simultaneous
motion
deblurring and super-resolution
CVPR 2009
Optimal
Single
Image Capture for Motion Deblurring
CVPR 2009
New
codes for coded
exposure deblurring to help in PSF Estimation
Copyright 2006-2010 Mitsubishi Electric Research
Laboratories All
Rights
Reserved.
Permission to use, copy and modify softwares linked from this page and its documentation without fee for educational, research and non-profit purposes, is hereby granted, provided that the above copyright notice and the following three paragraphs appear in all copies.
To request permission to incorporate this software into
commercial
products contact: Vice President of Marketing and Business
Development;
Mitsubishi Electric Research Laboratories (MERL), 201
Broadway,
Cambridge, MA 02139 or
IN NO EVENT SHALL MERL BE LIABLE TO ANY PARTY FOR DIRECT,
INDIRECT,
SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING
LOST PROFITS,
ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS
DOCUMENTATION, EVEN IF
MERL HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
MERL SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
BUT
NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
FITNESS FOR A
PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS ON
AN "AS IS"
BASIS, AND MERL HAS NO OBLIGATIONS TO PROVIDE MAINTENANCE,
SUPPORT,
UPDATES, ENHANCEMENTS OR MODIFICATIONS.