Matplotlib displaying RGB pixels weirdly


I have a set of RGB pixels I gathered from an image, but when I try to save them and display them, they come out very weirdly, as 3 separate colors. It seems like it’s not reading my array as a list of RGB values.

I have an abbreviated version of my code, with a short list of RGB values:

import matplotlib.pyplot as plt
import numpy as np

img = np.array([[255, 255, 255], [127, 255, 127], [239, 255,  15], [127 ,  0,   0]])

plt.savefig('testtt.png', dpi=400)

which gives me this:
enter image description here

Does anyone know how to resolve this or what might be wrong? Or alternatively, if anyone knows a better way to display RGB pixels from an array, please let me know. Side note: I am trying to avoid OpenCV

Asked By: lordclyborne



So, you created an array of lists

img = np.array([[255, 255, 255],[127, 255, 127],[239, 255,  15],[127 ,  0,   0]])

But nothing tells python to treat this as and RGB image, so instead it created an array with four rows and three columns. Then when you plotted it here:


It simply applied colors from the viridis color pallet and displayed it as a visual array where each value got a color from that pallet.

As mentioned in the comments, you need to expressly tell python to treat this array as an image:

imshow(img, origin='upper', interpolation='nearest')

For a straight pixel-by-pixel translation

Answered By: sconfluentus

A 2D array will be interpreted as values to be color mapped. That’s why you see 4 rows of 3 colors.

To interpret the array as 4 rgb values, you can plot a 3D array:

import matplotlib.pyplot as plt
import numpy as np

img = np.array([[255, 255, 255], [127, 255, 127], [239, 255, 15], [127, 0, 0]])
plt.imshow([img]) # note the extra brackets

plt.imshow with rgb values

To illustrate what was happening with the original plot, seaborn’s heatmap can automatically add some annotations and a colorbar.

import seaborn as sns

sns.heatmap(img, annot=True, fmt='d', cmap='viridis')

sns.heatmap of intended rgb array

PS: To display RGB values as an NxM array of pixels, the values should form a 3D NxMx3 array:

img = np.array([[[255, 255, 255], [127, 255, 127]], [[239, 255, 15], [127, 0, 0]]])

a 2x2x3 array

Answered By: JohanC
Categories: questions Tags: , , , ,
Answers are sorted by their score. The answer accepted by the question owner as the best is marked with
at the top-right corner.