Indexing in numpy array

Question

Given a lookup table:

colors = [ [0,0,0],\
 [0, 255, 0],\
 [0, 0, 255],\
 [255, 0, 0]]

And an index numpy matrix input 2x2:

a = np.array([[0,1],[1,1]])

How can I map a to a 2x2x3 matrix b, where b[i][j] = colors[a[i][j]]? I want to avoid using a for loop here.

Answer 1

Have you tried:

colors[a]

Here's a full example:

import numpy as np
colors = np.array([[0,0,0],
 [0, 255, 0],
 [0, 0, 255],
 [255, 0, 0]
 ])
a = np.array([[0, 1], [1, 1]])
new = colors[a]
new.shape
# (2, 2, 3)
new
# array([[[ 0, 0, 0],
# [ 0, 255, 0]],
#
# [[ 0, 255, 0],
# [ 0, 255, 0]]])

• agstudy: Note that you redefine colors here using numpy array otherwise you get an error with "OP colors".
• Bi Rico: @agstudy, very true. If colors is a list of lists you can only index it using slices or integers.