1. CS-467 Image processing and Computer Vision
Course Project 8
Goals:
1) to learn how to restore blurred images
Use VEGA program
1. Choose an image ( , )f x y from a collection of blurred images
(see classdataIMAGESBLUR EXAMPLES)
degraded by Gaussian, motion, rectangular, and defocus blur.
2. Restore all 4 versions of this blurred image in VEGA using an image restoration tool. First, use a
Restoration“Blur Recognition” tool and then, after a type of blur and its parameters are recognized,
use Restoration “Image Restoration” tool to restore an image. Use Wiener filter, Tikhonov
regularization, and Inverse Filter. Compare their efficiency. Try to find a better parameter for the
recognized type of blur, if you are not satisfied with the restoration results. Estimate the restoration
quality using PSNR as a measure.
3. Prepare a brief report based on the measured PSNRs.
Bonus (50 % extra credit). Design Matlab functions for measuring BSNR and ISNR (see slide 27 of
the Lecture-11 presentation) and apply them along with PSNR to evaluate your results. In BSNR, M
(noise variance) shall be used as one of input parameters), but in your experiments you may set 1σ = ,
since images in the “blurry” collection are not noisy. Include BSNR and ISNR in your report and make
your conclusions based on all of PSNR, BSNR, and ISNR.
Put your resulting images and the report in the subfolder Project 8 (you need to create it) located in the
designated folder (named by your last name) in the folder
sfs01classesCS 467 001Class Data (The folder sfs01classes is mapped from all the lab computers,
so you can easily find it through File Explorer (Computer) in Windows 7. A shortcut to the Classes
folder is also available on the desktop of the lab computers.