imha

A more constrained and friendlier fork of ImageHash.

imha is constrained to not use numpy and keep dependencies to a minimum. This keeps the package simpler and also easier to be installed on limited environments. imha also offers more customization for parametrizing the hash computation and a better engineered hash representation that can be easily converted to different types and representations.

Usage

As a Python package:

>>> from PIL import Image
>>> import imha
>>> hash1 = imha.average_hash(Image.open("photo1.jpg"))
>>> hash1.bin()
'1111111111010111100100011000000110000001110010011111111111111111'
>>> hash1.hex()
'ffd7918181c9ffff'
>>> hash1.uint()
18435363585078722559
>>> bytes(hash1)
b'\xff\xd7\x91\x81\x81\xc9\xff\xff'
>>> int(hash1)
-11380488630829057
>>> len(hash1)  # hash length in bits
64
>>> hash2 = imha.average_hash(Image.open("photo2.jpg"))
>>> hash2.hex()
'9f172786e71f1e00'
>>> hash1 == hash2
False
>>> hash1 - hash2  # hamming distance between hashes
33

As a command-line tool:

$ imha --help
usage: imha [--size WIDTH HEIGHT] [--skip-corners] [--format FORMAT] ALGORITHM FILE [FILE ...]

A more constrained and friendlier fork of ImageHash.

positional arguments:
  ALGORITHM            one of: average_hash, dhash, dhash_vertical
  FILE                 input file(s)

options:
  -h, --help           show this help message and exit
  --version            show program's version number and exit
  --size WIDTH HEIGHT  dimensions to resize the image to (default: see API Reference)
  --skip-corners       ignore the four corners
  --format FORMAT      output format, one of: bin, hex, uint, int (default: hex)

$ imha average_hash *.jpg
ffd7918181c9ffff  photo1.jpg
9f172786e71f1e00  photo2.jpg

Goals

• Generate the same hash values as ImageHash.
• Require only the pillow package.
• Be compatible with currently supported Python versions.

API Reference

average_hash

def average_hash(
    image: Image.Image, size: tuple[int, int] = (8, 8), *, skip_corners: bool = False
) -> Hash

Compute Average Hash.

Computes hash with width*height bits. Enabling skip_corners reduces the hash length by 4 bits. This means a 64-bits hash can be generated with size=(8, 8) or a 16-bit hash be generated with either size=(4, 4) or size=(5, 4), skip_corners=True for example.

Arguments:

• image: Input image.
• size: Tuple with width and height to resize the image to. (default: (8, 8))
• skip_corners: Ignore the four corners. (default: False)

dhash

def dhash(
    image: Image.Image, size: tuple[int, int] = (9, 8), *, skip_corners: bool = False
) -> Hash

Compute Difference Hash by row.

Computes row hash with (width-1)*height bits. Enabling skip_corners reduces the hash length by 4 bits. This means a 64-bits hash can be generated with size=(9, 8) or a 16-bit hash be generated with either size=(5, 4) or size=(6, 4), skip_corners=True for example.

Arguments:

• image: Input image.
• size: Tuple with width and height to resize the image to. (default: (9, 8))
• skip_corners: Ignore the four corners. (default: False)

dhash_vertical

def dhash_vertical(
    image: Image.Image, size: tuple[int, int] = (8, 9), *, skip_corners: bool = False
) -> Hash

Compute Difference Hash by column.

Computes col hash with width*(height-1) bits. Enabling skip_corners reduces the hash length by 4 bits. This means a 64-bits hash can be generated with size=(8, 9) or a 16-bit hash be generated with either size=(4, 5) or size=(5, 5), skip_corners=True for example.

Arguments:

• image: Input image.
• size: Tuple with width and height to resize the image to. (default: (8, 9))
• skip_corners: Ignore the four corners. (default: False)

Hash Objects

class Hash()

Represents a hash object.

The hash value can be converted to various types and representations using the built-in functions and provided methods. The hamming distance between hashes can be computed by subtracting one hash from another.

__init__

def __init__(value: int, len_: int) -> None

Create hash object.

Arguments:

• value: The hash as an unsigned integer value.
• len_: The hash length in bits.

bin

def bin() -> str

Return the hash binary representation.

hex

def hex() -> str

Return the hash hexadecimal representation.

uint

def uint() -> int

Return the hash as an unsigned integer value.

__len__

def __len__() -> int

Return len(self), the hash length in bits.

__sub__

def __sub__(other: Self) -> int

Return self - other, the hamming distance between hashes.

__eq__

def __eq__(other: Any) -> bool

Return self == other.

__hash__

def __hash__() -> int

Return hash(self).

__bytes__

def __bytes__() -> bytes

Return bytes(self), the hash as bytes.

__int__

def __int__() -> int

Return int(self), the hash as a signed integer value.

__index__

def __index__() -> int

Return integer for bin(self), hex(self) and oct(self).

__repr__

def __repr__() -> str

Return repr(self).

__str__

def __str__() -> str

Return str(self).