Messing with DHASH

README.md

Use DHASH to compare images

Based on this repo:

https://github.com/Jetsetter/dhash.git

I categorised a bunch of images and had a play with this PIP repo to see whether it could tell the difference between images that are quite similat

Conclusion

This could work but may need to be tuned a bit.

Even very similar photos can give a difference score of 45-50. Completely different images giving a difference of ~50.

dh.py

#! /usr/bin/env python
import glob
import dhash
from PIL import Image

testfiledir="/Users/jameshook/data/test_dhash/*.JPG"
files = glob.glob(testfiledir)


def calc_dhash( filename ):
        image = Image.open(filename)
        hash_d = dhash.dhash_int(image)
        return ( filename , hash_d  )


def calc_delta( h1 , h2 ):
        return dhash.get_num_bits_different( h1 , h2 )


def calc_deltas( files , file    ):
        return [ ( f1[0] , file[0] , calc_delta( f1[1] , file[1] ))  for f1 in files   ]


def display_grid( grid ):
        for f in grid:
                print "==============================="
                print f[0]
                print "======"
                sorted_hashes = sorted( f[1], key=lambda delta: delta[2] )
                for s in sorted_hashes:
                        print "%s %s" % ( s[0] , s[2]   )


hashes = map( calc_dhash , files)

print "HASHES"
print "======"
print hashes


print "DELTAS"
print "======"

grid = [ (  f1 , calc_deltas( hashes , f1 )   )  for f1 in hashes     ]
display_grid( grid )

output

HASHES
======
[('/Users/jameshook/data/test_dhash/class_2_1.JPG', 133444065540591572424685057146254969621L), ('/Users/jameshook/data/test_dhash/class_2_2.JPG', 288155882796435360591608754682148608983L), ('/Users/jameshook/data/test_dhash/class_4_1.JPG', 208172743778867845890888275352708787980L), ('/Users/jameshook/data/test_dhash/class_4_2.JPG', 41523451140898682701094046611791445839L), ('/Users/jameshook/data/test_dhash/class_1_1.JPG', 144119612009837309271364984800538487105L), ('/Users/jameshook/data/test_dhash/class_3_2.JPG', 150705215361708422803300784528586672896L), ('/Users/jameshook/data/test_dhash/class_1_2.JPG', 135810795512037234261205901712547742471L), ('/Users/jameshook/data/test_dhash/class_1_3.JPG', 144797399067380238615026752257795721473L), ('/Users/jameshook/data/test_dhash/class_3_1.JPG', 150788292111482414106044836724741491076L), ('/Users/jameshook/data/test_dhash/class_1_4.JPG', 285734173071906646471869495980529811472L), ('/Users/jameshook/data/test_dhash/class_5_1.JPG', 47874767624787918727612730191534344449L)]
DELTAS
======
===============================
('/Users/jameshook/data/test_dhash/class_2_1.JPG', 133444065540591572424685057146254969621L)
======
/Users/jameshook/data/test_dhash/class_2_1.JPG 0
/Users/jameshook/data/test_dhash/class_1_1.JPG 38
/Users/jameshook/data/test_dhash/class_1_2.JPG 42
/Users/jameshook/data/test_dhash/class_5_1.JPG 44
/Users/jameshook/data/test_dhash/class_1_3.JPG 47
/Users/jameshook/data/test_dhash/class_3_1.JPG 51
/Users/jameshook/data/test_dhash/class_1_4.JPG 51
/Users/jameshook/data/test_dhash/class_2_2.JPG 52
/Users/jameshook/data/test_dhash/class_3_2.JPG 57
/Users/jameshook/data/test_dhash/class_4_2.JPG 68
/Users/jameshook/data/test_dhash/class_4_1.JPG 70
===============================
('/Users/jameshook/data/test_dhash/class_2_2.JPG', 288155882796435360591608754682148608983L)
======
/Users/jameshook/data/test_dhash/class_2_2.JPG 0
/Users/jameshook/data/test_dhash/class_2_1.JPG 52
/Users/jameshook/data/test_dhash/class_1_1.JPG 52
/Users/jameshook/data/test_dhash/class_1_3.JPG 57
/Users/jameshook/data/test_dhash/class_3_2.JPG 61
/Users/jameshook/data/test_dhash/class_1_4.JPG 61
/Users/jameshook/data/test_dhash/class_3_1.JPG 63
/Users/jameshook/data/test_dhash/class_4_1.JPG 64
/Users/jameshook/data/test_dhash/class_5_1.JPG 66
/Users/jameshook/data/test_dhash/class_1_2.JPG 68
/Users/jameshook/data/test_dhash/class_4_2.JPG 70
===============================
('/Users/jameshook/data/test_dhash/class_4_1.JPG', 208172743778867845890888275352708787980L)
======
/Users/jameshook/data/test_dhash/class_4_1.JPG 0
/Users/jameshook/data/test_dhash/class_4_2.JPG 52
/Users/jameshook/data/test_dhash/class_3_2.JPG 59
/Users/jameshook/data/test_dhash/class_3_1.JPG 61
/Users/jameshook/data/test_dhash/class_5_1.JPG 62
/Users/jameshook/data/test_dhash/class_2_2.JPG 64
/Users/jameshook/data/test_dhash/class_1_1.JPG 64
/Users/jameshook/data/test_dhash/class_1_2.JPG 64
/Users/jameshook/data/test_dhash/class_1_3.JPG 65
/Users/jameshook/data/test_dhash/class_1_4.JPG 65
/Users/jameshook/data/test_dhash/class_2_1.JPG 70
===============================
('/Users/jameshook/data/test_dhash/class_4_2.JPG', 41523451140898682701094046611791445839L)
======
/Users/jameshook/data/test_dhash/class_4_2.JPG 0
/Users/jameshook/data/test_dhash/class_4_1.JPG 52
/Users/jameshook/data/test_dhash/class_1_1.JPG 62
/Users/jameshook/data/test_dhash/class_1_3.JPG 67
/Users/jameshook/data/test_dhash/class_2_1.JPG 68
/Users/jameshook/data/test_dhash/class_1_2.JPG 68
/Users/jameshook/data/test_dhash/class_2_2.JPG 70
/Users/jameshook/data/test_dhash/class_1_4.JPG 71
/Users/jameshook/data/test_dhash/class_5_1.JPG 74
/Users/jameshook/data/test_dhash/class_3_2.JPG 77
/Users/jameshook/data/test_dhash/class_3_1.JPG 77
===============================
('/Users/jameshook/data/test_dhash/class_1_1.JPG', 144119612009837309271364984800538487105L)
======
/Users/jameshook/data/test_dhash/class_1_1.JPG 0
/Users/jameshook/data/test_dhash/class_2_1.JPG 38
/Users/jameshook/data/test_dhash/class_1_2.JPG 40
/Users/jameshook/data/test_dhash/class_1_3.JPG 43
/Users/jameshook/data/test_dhash/class_5_1.JPG 48
/Users/jameshook/data/test_dhash/class_2_2.JPG 52
/Users/jameshook/data/test_dhash/class_3_2.JPG 57
/Users/jameshook/data/test_dhash/class_1_4.JPG 57
/Users/jameshook/data/test_dhash/class_3_1.JPG 61
/Users/jameshook/data/test_dhash/class_4_2.JPG 62
/Users/jameshook/data/test_dhash/class_4_1.JPG 64
===============================
('/Users/jameshook/data/test_dhash/class_3_2.JPG', 150705215361708422803300784528586672896L)
======
/Users/jameshook/data/test_dhash/class_3_2.JPG 0
/Users/jameshook/data/test_dhash/class_3_1.JPG 22
/Users/jameshook/data/test_dhash/class_5_1.JPG 39
/Users/jameshook/data/test_dhash/class_1_2.JPG 53
/Users/jameshook/data/test_dhash/class_1_3.JPG 54
/Users/jameshook/data/test_dhash/class_2_1.JPG 57
/Users/jameshook/data/test_dhash/class_1_1.JPG 57
/Users/jameshook/data/test_dhash/class_4_1.JPG 59
/Users/jameshook/data/test_dhash/class_1_4.JPG 60
/Users/jameshook/data/test_dhash/class_2_2.JPG 61
/Users/jameshook/data/test_dhash/class_4_2.JPG 77
===============================
('/Users/jameshook/data/test_dhash/class_1_2.JPG', 135810795512037234261205901712547742471L)
======
/Users/jameshook/data/test_dhash/class_1_2.JPG 0
/Users/jameshook/data/test_dhash/class_1_1.JPG 40
/Users/jameshook/data/test_dhash/class_2_1.JPG 42
/Users/jameshook/data/test_dhash/class_1_4.JPG 47
/Users/jameshook/data/test_dhash/class_1_3.JPG 49
/Users/jameshook/data/test_dhash/class_5_1.JPG 50
/Users/jameshook/data/test_dhash/class_3_2.JPG 53
/Users/jameshook/data/test_dhash/class_3_1.JPG 55
/Users/jameshook/data/test_dhash/class_4_1.JPG 64
/Users/jameshook/data/test_dhash/class_2_2.JPG 68
/Users/jameshook/data/test_dhash/class_4_2.JPG 68
===============================
('/Users/jameshook/data/test_dhash/class_1_3.JPG', 144797399067380238615026752257795721473L)
======
/Users/jameshook/data/test_dhash/class_1_3.JPG 0
/Users/jameshook/data/test_dhash/class_1_4.JPG 42
/Users/jameshook/data/test_dhash/class_1_1.JPG 43
/Users/jameshook/data/test_dhash/class_2_1.JPG 47
/Users/jameshook/data/test_dhash/class_1_2.JPG 49
/Users/jameshook/data/test_dhash/class_3_2.JPG 54
/Users/jameshook/data/test_dhash/class_2_2.JPG 57
/Users/jameshook/data/test_dhash/class_5_1.JPG 57
/Users/jameshook/data/test_dhash/class_3_1.JPG 60
/Users/jameshook/data/test_dhash/class_4_1.JPG 65
/Users/jameshook/data/test_dhash/class_4_2.JPG 67
===============================
('/Users/jameshook/data/test_dhash/class_3_1.JPG', 150788292111482414106044836724741491076L)
======
/Users/jameshook/data/test_dhash/class_3_1.JPG 0
/Users/jameshook/data/test_dhash/class_3_2.JPG 22
/Users/jameshook/data/test_dhash/class_5_1.JPG 41
/Users/jameshook/data/test_dhash/class_2_1.JPG 51
/Users/jameshook/data/test_dhash/class_1_2.JPG 55
/Users/jameshook/data/test_dhash/class_1_3.JPG 60
/Users/jameshook/data/test_dhash/class_1_4.JPG 60
/Users/jameshook/data/test_dhash/class_4_1.JPG 61
/Users/jameshook/data/test_dhash/class_1_1.JPG 61
/Users/jameshook/data/test_dhash/class_2_2.JPG 63
/Users/jameshook/data/test_dhash/class_4_2.JPG 77
===============================
('/Users/jameshook/data/test_dhash/class_1_4.JPG', 285734173071906646471869495980529811472L)
======
/Users/jameshook/data/test_dhash/class_1_4.JPG 0
/Users/jameshook/data/test_dhash/class_1_3.JPG 42
/Users/jameshook/data/test_dhash/class_1_2.JPG 47
/Users/jameshook/data/test_dhash/class_2_1.JPG 51
/Users/jameshook/data/test_dhash/class_1_1.JPG 57
/Users/jameshook/data/test_dhash/class_3_2.JPG 60
/Users/jameshook/data/test_dhash/class_3_1.JPG 60
/Users/jameshook/data/test_dhash/class_2_2.JPG 61
/Users/jameshook/data/test_dhash/class_5_1.JPG 63
/Users/jameshook/data/test_dhash/class_4_1.JPG 65
/Users/jameshook/data/test_dhash/class_4_2.JPG 71
===============================
('/Users/jameshook/data/test_dhash/class_5_1.JPG', 47874767624787918727612730191534344449L)
======
/Users/jameshook/data/test_dhash/class_5_1.JPG 0
/Users/jameshook/data/test_dhash/class_3_2.JPG 39
/Users/jameshook/data/test_dhash/class_3_1.JPG 41
/Users/jameshook/data/test_dhash/class_2_1.JPG 44
/Users/jameshook/data/test_dhash/class_1_1.JPG 48
/Users/jameshook/data/test_dhash/class_1_2.JPG 50
/Users/jameshook/data/test_dhash/class_1_3.JPG 57
/Users/jameshook/data/test_dhash/class_4_1.JPG 62
/Users/jameshook/data/test_dhash/class_1_4.JPG 63
/Users/jameshook/data/test_dhash/class_2_2.JPG 66
/Users/jameshook/data/test_dhash/class_4_2.JPG 74