Entropy of an image

C++
,
1

I want entropy applied on the entire image.

I tested it quickly using Scikit Image Entropy

I want implement the same in C++ using opencv, I found a solution on the stack overflow entropy for a gray image in opencv

Scikit Image implements it on the entire image where as the stackoverflow is giving a float value.

Are both the same?
If same/different, How to I implement the scikit implementation of entropy in opencv using C++?

3

what is your “use-case” here ? what do you want to achieve with it ?

  • the c++ version calculates a single entropy value for an image
  • the scipy one calculates entropy values for a pixel neighbourhood (a subimage), and sets the pixel value to the entropy of that subimage.
4

The scipy version gives higher values for edges in the entire image which is of interest for me.

I would like to implement with the same method for entropy as done on scikit image, entropy values for a pixels neighborhood(a subimage) and set the pixel value to the entropy of that subimage. Using OpenCV C++.

5

(did you mean: edges inside the image, or the image borders ?)
you will need a “sliding window”, and there’s a border condition

(and, how do you know the difference, if you dont have a working c++ impl ?)

[edit] oh wait, did you mean, it highlights the edges in your image ?
again, what is this for ? do you need an “edge detector” (like Canny or Sobel) ?