将3d矩阵转换为特征向量

Question

So I have data in shape (100,100,5000).

Basically, it is a 100 by 100 pixel image which each (x,y) pixel having some spectrum vector.. So, the data is format

     [ [ [ 0, 0.2.....],[0.1,0.3.....].. and so on]]

When we do d[0][0][0] corresponds to (0,0) pixel in image and the spectrum value is 0

Now, i want to run kmeans algorithm in it.. Thats the background.

To run kmeans, I want to convert it into feature vector.. as

[0,0,0]
[0,0,0.2]
and so on..

Is there a "python" way to do this conversion. I am guessing there should be some numpy reshape method which can easily achieve this, rather than writing the three for loops which I am trying to avoid.

Answer 1

Use numpy reshape http://docs.scipy.org/doc/numpy/reference/generated/numpy.reshape.html eg

a.reshape((m,n))

where m and n are the desired coordinates of your array a