Robust Motion Deblurring System Ver.2.0 (Robust deblurring of motion) and offers an effective method of assessing quality of grain, based on the use of spatial iiteratsionnogo (recurring) Detection of grain (Iterative Support Detection) and its refinement to restore photos with significant motion blur.
The authors write, Cho their method can evaluate a very large grain blurring (PSFs).
A more detailed description of the proposed method (with all calculations and formulas) can be read by downloading the "Two-Phase Kernel Estimation for Robust Motion Deblurring", format. Pdf, size 2.18 Mb.
Direct Link: _http: / / www.cse.cuhk.edu.hk/ ~ leojia / projects / robust_deblur / robust_motion_deblurring.pdf
Testing.
1. On the main screen shot - examples of the photographs available in the program and website.
2. When the program starts with caution. The program announces that be able to use GPU, but there is no necessary libraries (my card does not support), we use the SPU.
3. Looking for a blurred image (usually immediately delete) and upload photos via built-in wizard. Choose the area to fix. The first attempt to correct the entire image (3264h2448), I stopped - it's a very long process on my computer.
4. The program offers three modes: Small, Medium, Large (Master in each mode is an example of blurred photos). I tried both modes. The results (enlarged fragments):
5. To evaluate the results is difficult because not known from which motion blur? From the title implies that the system reliably removes blur from motion: When shooting a moving object, shooting in motion (jitter) of the camera. But the errors that the lens does not correct the system.
System requirements:
* Processor Intel ® Pentium ® 4 or AMD Athlon ® 64
* 512 MB RAM
* 30 Mb/120 MB free space on your hard disk space for installation
* Screen resolution of 1024x768
* Supports GPU CUDA technology
CLICK HERE FOR DOWNLOAD
Bookmarks