Image hashing techniques

hash_engine.py

import os
import cv2
import hashlib
import numpy as np
import scipy.fftpack

class HashEngine:
    def __init__(self, htype="dhash", hash_size=8):
        self.hash_size = hash_size
        self.htype = htype

    def load_image(self, image):
        # either load image from disk or use array from buffer
        if isinstance(image, str):
            image = cv2.imread(image, 0)
        if len(image.shape) >= 3:
            image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
        return cv2.resize(image, (self.hash_size + 1, self.hash_size))
    
    @staticmethod
    def arr2hex(arr):
        h, s = 0, []
        for i, v in enumerate(arr.flatten()):
            h += 2**(i % 8) if v else 0
            if (i % 8) == 7:
                s.append(hex(h)[2:].rjust(2, '0')); h = 0
        return "".join(s)
    
    @staticmethod
    def hex2arr(hexstr):
        l = []
        if len(hexstr) != 16:
            raise ValueError('The hex string has the wrong length')
        for i in range(8):
            h = hexstr[i*2:i*2+2]
            v = int("0x" + h, 16)
            l.append([v & 2**i > 0 for i in range(8)])
        return np.array(l)
    
    @staticmethod
    def hex2int(hexstr):
        return int(hexstr, 16)
    
    @staticmethod
    def distance(a, b):
        try:
            # hamming distance between the integers
            return bin(int(a) ^ int(b)).count("1")
        except ValueError:
            # convert hex2int first
            return bin(int(a, 16) ^ int(b, 16)).count("1")
    
    def ahash(self, image):
        # average hash
        image = self.load_image(image)
        avg = image.mean()
        diff = image > avg
        return self.arr2hex(diff)[:16]
    
    def dhash(self, image):
        # difference hash
        image = self.load_image(image)
        diff = image[:, 1:] > image[:, :-1]
        return self.arr2hex(diff)
    
    def phash(self, image):
        # perceptual hash
        image = self.load_image(image)
        dct = scipy.fftpack.dct(scipy.fftpack.dct(image, axis=0), axis=1)
        dctlowfreq = dct[:self.hash_size, :self.hash_size]
        med = np.median(dctlowfreq)
        diff = dctlowfreq > med
        return self.arr2hex(diff)

    def hexdigest(self, image, htype=None):
        htype = htype or self.htype
        if htype == "ahash":
            return self.ahash(image)
        elif htype == "dhash":
            return self.dhash(image)
        elif htype == "phash":
            return self.phash(image)
        
        # all other inbuilt hashing algorithms
        algorithm = hashlib.new(htype)
        algorithm.update(self.load_image(image))
        return algorithm.hexdigest()

H = HashEngine("sha1")

hexd1 = H.hexdigest("sample3.jpg")
intd1 = H.hex2int(hexd1)

print("hexdigest1:", hexd1)
print("intdigest1:", intd1)

print("-"*10)

hexd2 = H.hexdigest("sample3.jpg")
intd2 = H.hex2int(hexd2)

print("hexdigest2:", hexd2)
print("intdigest2:", intd2)

print("-"*10)

hexdistance = H.distance(hexd1, hexd2)
print("hexdistance:", hexdistance)

intdistance = H.distance(intd1, intd2)
print("intdistance:", intdistance)

# hexdigest1: 576f176179c8e3fcb88e86b7f97d82e9b9e2b24f
# intdigest1: 499159616530556630584974124528933127117353431631
# ----------
# hexdigest2: 576f176179c8e3fcb88e86b7f97d82e9b9e2b24f
# intdigest2: 499159616530556630584974124528933127117353431631
# ----------
# hexdistance: 0
# intdistance: 0