Top Banner
Inferring the kernel: multiscale method Input image Loop over scales Variation al Bayes Upsample estimates ulti-scale approach to avoid local minim Initialize 3x3 blur kernel Convert to grayscale Remove gamma correction User selects patch from image
22

Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Mar 27, 2015

Download

Documents

Chloe Howe
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Inferring the kernel: multiscale method

Input image

Loop over scales

Variational

Bayes

Upsample

estimates

Use multi-scale approach to avoid local minima:

Initialize 3x3 blur kernel

Convert tograyscale

Remove gammacorrection

User selects patch from image

Page 2: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Image ReconstructionInput image

Full resolutionblur estimate

Non-blind deconvolution

(Richardson-Lucy)

Deblurred image

Loop over scales

Variational

Bayes

Upsample

estimates

Initialize 3x3 blur kernel

Convert tograyscale

Remove gammacorrection

User selects patch from image

Page 3: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:
Page 4: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Results on real images

Submitted by people from their own photo collections

Type of camera unknown

Output does contain artifacts– Increased noise

– Ringing

Compares well to existing methods

Page 5: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Original photograph

Page 6: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Blur kernel Our output

Page 7: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Original photographMatlab’s deconvblind

Page 8: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Original

Our output

Close-up of garland

Matlab’sdeconvblind

Page 9: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

A submitted photograph

A small list of the reasons why we didn’t attempt this photograph:

• Most of the features of interest are saturated.

• Different blur kernels for different lights (compare lantern streaks with sky light streaks and different than the water reflection streaks, and the car streaks below bridge and the streaks to left of bridge.)

• The objects reflected by flash are stationary and have no motion blur.

Page 10: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

We don’t handle subject motion blur

Page 11: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Failure mode

Page 12: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Original photograph

Page 13: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Matlab’s deconvblind

Page 14: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Photoshop sharpen more

Page 15: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Our output Blur kernel

Page 16: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:
Page 17: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Original photograph

Page 18: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Our output

Blur kernel

Page 19: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Original photograph

Page 20: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Our output

Blur kernel

Page 21: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Matlab’s deconvblind

Page 22: Inferring the kernel: multiscale method Input image Loop over scales Variational Bayes Upsample estimates Use multi-scale approach to avoid local minima:

Original photograph