How to efficiently compute over specified index of array nested in numpy ndarray?

Question

I know versions of this question appear frequently on here, but I was not able to find a solution that works for me. For some background on the problem, I have an RGB image that is divided into chunks of NxN pixels. I want to compute the average of each color channel separately for each chunk. I know numpy is best used by leveraging vectorized operations, but the level of higher-dimensional slicing and indexing required here is beyond me. Essentially I need the following functionality:

 for row in tiles:
     for col in row:
         rsum = 0;
         gsum = 0;
         bsum = 0;
         for n in col:
             for vec in n:
                 rsum += vec[0]
                 gsum += vec[1]
                 bsum += vec[2]
         col[..., 0] = rsum/n.shape[0]**2
         col[..., 1] = gsum/n.shape[0]**2
         col[..., 2] = bsum/n.shape[0]**2

Where the shape of my ndarray is:

tiles.shape = (138, 84, 100, 100, 4)

A 138x84 matrix of 100x100 matrices, where each element is a length-4 vector. Is there a way to do this without any loops? Should I reshape my ndarray? Any guidance is much appreciated.

orlp · Accepted Answer · 2020-11-10 20:43:11Z

1

Simply pass the axes you want to average over to np.mean:

avg = np.mean(tiles, axis=(2, 3))

answered Nov 10, 2020 at 20:43

orlp

119k39 gold badges226 silver badges324 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Hans Musgrave Over a year ago

I don't think that's quite right yet. He's leaving the 4th channel alone.

Collectives™ on Stack Overflow

How to efficiently compute over specified index of array nested in numpy ndarray?

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related