图片相似度识别：aHash算法

$ /usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"$ brew install opencv

from PIL import Image #用PIL处理图像import osimport numpy as np#import cv2 ——如果要用opencv时需导入
#均值哈希算法def aHash(image):    image_new=image    #计算均值    avreage = np.mean(image_new)     hash = []     for i in range(image.shape[0]):         for j in range(image.shape[1]):             if image[i,j] > avreage:                 hash.append(1)             else:                 hash.append(0)     return hash
#计算汉明距离def Hamming_distance(hash1,hash2):     num = 0    for index in range(len(hash1)):         if hash1[index] != hash2[index]:             num += 1    return num
if __name__ == "__main__":    #PIL    image1 = Image.open('image1.png')    image2 = Image.open('image2.png')    #缩小尺寸并灰度化    image1=np.array(image1.resize((8, 8), Image.ANTIALIAS).convert('L'), 'f')    image2=np.array(image2.resize((8, 8), Image.ANTIALIAS).convert('L'), 'f')        #opencv    #img1 = cv2.imread('image1')    #img2 = cv2.imread('image2')     #缩小尺寸并灰度化    #image1=cv2.cvtColor(cv2.resize(img1,(8,8),interpolation=cv2.INTER_CUBIC),cv2.COLOR_BGR2GRAY)    #image2=cv2.cvtColor(cv2.resize(img2,(8,8),interpolation=cv2.INTER_CUBIC),cv2.COLOR_BGR2GRAY)    hash1 = aHash(image1)    hash2 = aHash(image2)    dist = Hamming_distance(hash1, hash2)    #将距离转化为相似度    similarity = 1 - dist * 1.0 / 64     print('dist is '+'%d' % dist)    print('similarity is ' +'%d' % similarity)