use L2-Norm or Euclidean Distance of image histograms to decide image similarity
from PIL import Image
from collections import Counter
import numpy as np
reference_image_1 = Image.open('Reference_image1.jpg')
reference_image_arr = np.asarray(reference_image_1)
flat_array_1 = array1.flatten()
print(np.shape(flat_array_1))
#generate count histogram
RH1 = Counter(flat_array_1)
H1 = []
for i in range(256):
if i in RH1.keys():
H1.append(RH1[i])
else:
H1.append(0)
# euclidean distance function
def L2Norm(H1,H2):
distance =0
for i in range(len(H1)):
distance += np.square(H1[i]-H2[i])
return np.sqrt(distance)
#evaluation
dist_test_ref_1 = L2Norm(H1,test_H)
print("The distance between Reference_Image_1 and Test Image is : {}".format(dist_test_ref_1))
dist_test_ref_2 = L2Norm(H2,test_H)
print("The distance between Reference_Image_2 and Test Image is : {}".format(dist_test_ref_2))
