subsampling a numpy array?

Question

Im relatively new to numpy but have started using it to read and write from and to h5 files. I have image data on which I have computed some zonal statistics, reading each pixel value in a given zone into a h5 file. However, I have a lot of pixel values (possibly tens of millions) and wanted to subsample this data so that I am able to cut down the data size but keep the general distribution of the data.

I was wondering if there was a simple way of sampling every 200th value of an array?

I would put up what code I have already but my code only goes as far as to read in my existing data - Im completely stuck as to how I might subsample it so have nothing to show so far.

Thanks

your question is a little unclear; is your concern with the size on disk or the size in memory? if you don't care about disk space; you can read a sliced view from a h5 file. this will still be slow though, since you effectively still need to read everything from disk. however, subsampling is better done by summing over all pixels; otherwise you might get nasty aliasing artifacts. if disk space is no objection, you could store a whole mipmap in your h5 file. that will give best performance and quality, but will increase rather than decrease your disk space use. — Eelco Hoogendoorn
– Eelco Hoogendoorn, Commented Feb 10, 2014 at 20:20

wim · Accepted Answer · 2014-02-10 18:26:10Z

8

You can use an array slice:

>>> import numpy as np
>>> a = np.eye(1000)
>>> a[::200, ::200]

array([[ 1.,  0.,  0.,  0.,  0.],
       [ 0.,  1.,  0.,  0.,  0.],
       [ 0.,  0.,  1.,  0.,  0.],
       [ 0.,  0.,  0.,  1.,  0.],
       [ 0.,  0.,  0.,  0.,  1.]])

answered Feb 10, 2014 at 18:26

wim

368k114 gold badges681 silver badges817 bronze badges

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

Collectives™ on Stack Overflow

subsampling a numpy array?

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related