ImageSimilarity.zip_image similarity_imagesimilarity_图像相似_图片相似_图

from PIL import Image import numpy as np import math import os import matplotlib.pyplot as plt import colorsys # Compute low order moments(1,2,3) def color_moments(filename): img = Image.open(filename) if img is None: return #9 vectors color_feature = [] #get r,g,b r,g,b = img.split() # The first central moment - average r_mean = np.mean(r) # np.sum(h)/float(N) g_mean = np.mean(g) # np.sum(s)/float(N) b_mean = np.mean(b) # np.sum(v)/float(N) color_feature.extend([r_mean, g_mean, b_mean]) # The second central moment - standard deviation r_std = np.std(r) # np.sqrt(np.mean(abs(h - h.mean())**2)) g_std = np.std(g) # np.sqrt(np.mean(abs(s - s.mean())**2)) b_std = np.std(b) # np.sqrt(np.mean(abs(v - v.mean())**2)) color_feature.extend([r_std, g_std, b_std]) # The third central moment - the third root of the skewness r_skewness = np.mean(abs(r - r_mean)**3) g_skewness = np.mean(abs(g - g_mean)**3) b_skewness = np.mean(abs(b - b_mean)**3) r_thirdMoment = r_skewness**(1./3) g_thirdMoment = g_skewness**(1./3) b_thirdMoment = b_skewness**(1./3) color_feature.extend([r_thirdMoment, g_thirdMoment, b_thirdMoment]) return color_feature def calculate_distance(feature_1, feature_2): distance = 0 for i in range(0,9): distance += (abs(feature_1[i]-feature_2[i]))**2 distance = math.sqrt(distance) return distance def similarity(doc,rootdir): feature_1 = color_moments(doc) print(doc) min = 10000 location = -1 list = os.listdir(rootdir) distance = [0]*len(list) for i in range(0,len(list)): path = os.path.join(rootdir,list[i]) if os.path.isfile(path): if os.path.samefile(path, doc): continue else: feature_2 = color_moments(path) distance[i] = calculate_distance(feature_1, feature_2) if min>distance[i]: min = distance[i] location = i print(os.path.join(rootdir,list[location])) print(min) plt.figure("result") plt.subplot(2,2,1),plt.title("Input") img = Image.open(doc) plt.imshow(img) plt.axis('off') plt.subplot(2,2,2),plt.title("most similarity\ndistance:"+str(round(distance[location],2))) img = Image.open(os.path.join(rootdir,list[location])) plt.axis('off') plt.imshow(img) count = 0 for i in range (0,len(list)): if os.path.samefile(os.path.join(rootdir,list[i]), doc): continue if i==location: continue count+=1 if count==9: break plt.subplot(4,4,count+8),plt.title("distance:"+str(round(distance[i],2))) img = Image.open(os.path.join(rootdir,list[i])) plt.imshow(img) plt.axis('off') plt.show() def picture_similarity(rootdir): mainlist = os.listdir(rootdir) for i in range(0,len(mainlist)): path = os.path.join(rootdir,mainlist[i]) if os.path.isfile(path): similarity(path,rootdir) rootdir = ".\img" picture_similarity(rootdir)