imghash
is a crate that allows you to generate different hashes for images. The following hashes can be generated using this crate:
There are multiple ways how to utilize imghash
depending on your use case.
The easy way to use imghash
is by using the provided utility functions which assume reasonable defaults.
use imghash::{average_hash, difference_hash, perceptual_hash};
let path = Path::new("path/to/my/image");
let average = average_hash(path);
let difference = difference_hash(path);
let perceptual = perceptual_hash(path);
Each of these functions return a Result<ImageHash, String>
-type. The ImageHash
object is essentially a container for the encoded bit matrix of the image (learn more here). The ImageHash
can be encoded into hexadecimal string by calling the encode
-method:
let res: String = hash.encode();
Hashes can be encoded into hexadecimal string by using the encode()
-method:
let res: String = hash.encode();
A hexadecimal string can also be decoded back into an ImageHash
:
let res: Result<ImageHash, String> = hash.decode("24f0", 4, 4);
The first argument of the hash is the string, the second and third are the width and height of the underlying matrix. This is required as each string can be encoded into different sizes matricies. If you want to understand more about the underlying bit matrix read the documentation about encoding.
The hamming distance is the distance of two hashes defined by the number of bits that differ between them. This distance can be easily computed:
let distance: Result<usize, String> = hash.distance(other_hash);
This can produce an error if the hashes are not of the same size.
If you need more flexibility, for example computing a larger bit matrix than the default, you can use a custom Hasher
.
For each hash type the crate provides a custom hasher, for the example here we will use the AverageHasher
:
use imghash::{average::AverageHasher};
let path = Path::new("path/to/my/image");
let hasher = AverageHasher {
width: 10,
height: 10,
};
let hash = hasher.hash_from_path(path);
Hasher
-instances also allow you to create hashes for already loaded images:
let img = ImageReader::open(...);
let hasher = AverageHasher { ..Default::default() };
let hash = hasher.hash_from_img(&img);
Each hasher also implements the Default
-trait, allowing you to create them with their default values:
let hasher = AverageHasher { ..Default::default() };
One of the major factors that drove development of this crate was the need to have a hasher implementation that matches the imagehash
-package for Python.
A wrapper with Python-bindings is now available here.
As of Version 1.2.0 all hashes generated by this crate should match hashes generated by imagehash
- however it is not guaranteed for any other package or crate. Previous version of this crate (<1.2.0) were not generated the same hashes. To make sure you are generating the same hashes, set the color space of the hasher to REC601
, as this will make sure the same grayscaling as Pillow is used - this is configured as the default.