How to check if an RGB image contains only one color?

Question

I'm using Python and PIL.

I have images in RGB and I would like to know those who contain only one color (say #FF0000 for example) or a few very close colors (#FF0000 and #FF0001).

I was thinking about using the histogram but it is very hard to figure out something with the 3 color bands, so I'm looking for a more clever algorithm.

Any ideas?

ImageStat module is THE answer! Thanks Aaron. I use ImageStat.var to get the variance and it works perfectly.

Here is my piece of code:

from PIL import Image, ImageStat

MONOCHROMATIC_MAX_VARIANCE = 0.005

def is_monochromatic_image(src):
    v = ImageStat.Stat(Image.open(src)).var
    return reduce(lambda x, y: x and y < MONOCHROMATIC_MAX_VARIANCE, v, True)

score 6 · Accepted Answer · answered Mar 24 '09 at 14:03

6

Try the ImageStat module. If the values returned by extrema are the same, you have only a single color in the image.

answered Mar 24 '09 at 14:03

Aaron Digulla

321,842
108
597
820

score 0 · Answer 2 · answered Mar 24 '09 at 13:41

0

First, you should define a distance between two colors. Then you just have to verify for each pixel that it's distance to your color is small enough.

answered Mar 24 '09 at 13:41

SmuK

1
1

Computing the Euclidian distance between each pixel in the 3 dimensions RGB space is theoreticaly fine, but I think it won't work with multi million pixels images... (at least in Python) – Grégoire Cachet Mar 24 '09 at 13:47
Why Euclidian distance? You can imagine something else depending on what you want. – SmuK Mar 24 '09 at 13:51
The problem is not in the type of distance, but in the fact that the algorithm is O(N^2), N being the number of pixels in the image (I have cases with N=10,000,000) – Grégoire Cachet Mar 24 '09 at 13:59
Perhaps i misunderstood your question. If you have to check that each pixel is close enough to a given color, it's linear time. – SmuK Mar 24 '09 at 14:06
@Smuk: I don't think it's a given color -- the color they are all closest to could be anything, specific to each image, right? – Doug Kavendek Mar 24 '09 at 14:09

score 0 · Answer 3 · answered Mar 24 '09 at 14:09

Here's a little snippet you could make use of :


import Image

im = Image.open("path_to_image")
width,height = im.size

for w in range(0,width):
  for h in range(0,height):
     # this will hold the value of all the channels
     color_tuple = im.getpixel((w,h))
     # do something with the colors here

Maybe use a hash and store the tuples as the key and it's number of appearances as value?

How to check if an RGB image contains only one color?

3 Answers3

Linked