histogram of gray scale values in numpy image

Question

I loaded an image into a numpy array and want to plot its color values in a histogram.

import numpy as np

from skimage import io
from skimage import color

img = io.imread('img.jpg')
img = color.rgb2gray(img)

unq = np.unique(img)
unq = np.sort(unq)

When we inspect the value of unq we will see something like

array([  5.65490196e-04,   8.33333333e-04,   1.13098039e-03, ...,
         7.07550980e-01,   7.09225490e-01,   7.10073725e-01])

which has still too much values for matplotlib so my idea was to loop over unq and remove every value which deviates only x from it's predecessor.

dels = []

for i in range(1, len(unq)):
    if abs(unq[i]-unq[i-1]) < 0.0003:
        dels.append(i)

unq = np.delete(unq, dels)

Though this method works it is very inefficient as it does not uses numpy's optimized implementations.

Is there a numpy feature would could do this for me?

Just noticed that my algorithm looses information about how often a color occurs. Let me try to fix this.

Answer 1

If you just want to compute the histogram, you can use np.histogram:

bin_counts, bin_edges = np.histogram(img, bins, ...)

Here, bins could either be the number of bins, or a vector specifying the upper and lower bin edges.

If you want to plot the histogram, the easiest way would be to use plt.hist:

bin_counts, bin_edges, patches = plt.hist(img.ravel(), bins, ...)

Note that I used img.ravel() to flatten out the image array before computing the histogram. If you pass a 2D array to plt.hist(), it will treat each row as a separate data series, which is not what you want here.