Relative entropy between 2 images.

2 Ansichten (letzte 30 Tage)
Rohini
Rohini am 20 Mai 2011
Bearbeitet: Jan Valosek am 24 Mär. 2020
Hello!!
I have taken an image. I found the entropy value using 'entropy' the inbuilt function. Next I altered the image by converting to blocks and shuffling. I found the entropy of this altered image. The value was found to be the same. Since the blocks are shuffled, the entropy has to increase. How do I find the changed entropy? The inbuilt function 'entropy' is giving the same value for both the original and altered image.
Please Help!

Antworten (2)

Andrew Fowler
Andrew Fowler am 20 Mai 2011
The entropy is calculated from the histogram of the image, so if you have the same pixel values, it doesn't matter if you shuffle the image, your entropy will stay the same because the pixel distribution is the same. What you could do, however, is find the entropy for each of your blocks, and measure the changes in entropy in different regions of the image as you shuffle blocks into and out of the region.

Jan Valosek
Jan Valosek am 24 Mär. 2020
Bearbeitet: Jan Valosek am 24 Mär. 2020
Hi,
I am dealing with similar task - finding some proper metric for quantification of difference between original and shuffled image. If entropy is not right way, can you recommend me another approach?
I thought about comparison of spectrums after fft2. Could it be the right approach?
Thanks!

Tags

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by