Comparing the Fitness of Two Images having the Same Size

This post provides a function which can compare two images, pixel by pixel, of the same size and return a measure between 0 and 1 indicating the fitness of the images.

1. Python Image Library (PIL) will be called. This can be installed by:
   # apt-get install python-pil
2. The two images for comparison should have the same dimensions.
3. Pixel by pixel comparison is performed. For each pair of corresponding pixels from the two images, if their colours are different, even it is a small change, they are considered to be different.

Source Code

---

from PIL import ImageChops # measure the fitness (0-1) between # two images pImgOrg and pImgNew # Notes: # 1. The two images should have # the same dimensions # 2. This function will perform a # pixel by pixel comparison # # return # 0 - totally different # 1 - match exactly # 0 < x < 1 - fitness def ImgFitness(pImgOrg, pImgNew): mf = 0.0 if not pImgOrg.size == pImgNew.size: return mf mdf = ImageChops.difference(pImgOrg, pImgNew) mb = mdf.tobytes() mf = float(mb.count('\x00'))/len(mb) return mf from PIL import Image im1 = Image.open('p01.png') im2 = Image.open('p01a.png') print ImgFitness(im1,im2)

Test Images

p01.png	p01a.png	p02.png	p03.png
The original image	Slightly different from p01.png	The colour is different	The colour and shape are different
Fitness: 1.0	Fitness: 0.993580246914	Fitness: 0.431604938272	Fitness: 0.31341563786