Download the PHP package jenssegers/imagehash without Composer

ImageHash

A perceptual hash is a fingerprint of a multimedia file derived from various features from its content. Unlike cryptographic hash functions which rely on the avalanche effect of small changes in input leading to drastic changes in the output, perceptual hashes are "close" to one another if the features are similar.

Perceptual hashes are a different concept compared to cryptographic hash functions like MD5 and SHA1. With cryptographic hashes, the hash values are random. The data used to generate the hash acts like a random seed, so the same data will generate the same result, but different data will create different results. Comparing two SHA1 hash values really only tells you two things. If the hashes are different, then the data is different. And if the hashes are the same, then the data is likely the same. In contrast, perceptual hashes can be compared -- giving you a sense of similarity between the two data sets.

This code was inspired/based on:

• https://github.com/kennethrapp/phasher
• http://www.phash.org
• http://blockhash.io
• http://www.hackerfactor.com/blog/?/archives/529-Kind-of-Like-That.html
• http://www.hackerfactor.com/blog/?/archives/432-Looks-Like-It.html
• http://blog.iconfinder.com/detecting-duplicate-images-using-python

Requirements

• PHP 8.1 or higher
• The gd or imagick extension
• Optionally, install the GMP extension for faster fingerprint comparisons

Installation

This package has not reached a stable version yet, backwards compatibility may be broken between 0.x releases. Make sure to lock your version if you intend to use this in production!

Install using composer:

composer require jenssegers/imagehash

Usage

The library comes with 4 built-in hashing implementations:

• Jenssegers\ImageHash\Implementations\AverageHash - Hash based the average image color
• Jenssegers\ImageHash\Implementations\DifferenceHash - Hash based on the previous pixel
• Jenssegers\ImageHash\Implementations\BlockHash - Hash based on blockhash.io Still under development
• Jenssegers\ImageHash\Implementations\PerceptualHash - The original pHash Still under development

Choose one of these implementations. If you don't know which one to use, try the DifferenceHash implementation. Some implementations allow some configuration, be sure to check the constructor.

The resulting Hash object, is a hexadecimal image fingerprint that can be stored in your database once calculated. The hamming distance is used to compare two image fingerprints for similarities. Low distance values will indicate that the images are similar or the same, high distance values indicate that the images are different. Use the following method to detect if images are similar or not:

Equal images will not always have a distance of 0, so you will need to decide at which distance you will evaluate images as equal. For the image set that I tested, a max distance of 5 was acceptable. But this will depend on the implementation, the images and the number of images. For example; when comparing a small set of images, a lower maximum distances should be acceptable as the chances of false positives are quite low. If however you are comparing a large amount of images, 5 might already be too much.

The Hash object can return the internal binary hash in a couple of different format:

Choose your preference for storing your hashes in your database. If you want to reconstruct a Hash object from a previous calculated value, use:

Demo

These images are similar:

Image 1 hash: 3c3e0e1a3a1e1e1e (0011110000111110000011100001101000111010000111100001111000011110)
Image 2 hash: 3c3e0e3e3e1e1e1e (0011110000111110000011100011111000111110000111100001111000011110)
Hamming distance: 3

These images are different:

Image 1 hash: 69684858535b7575 (0010100010101000101010001010100010101011001010110101011100110111)
Image 2 hash: e1e1e2a7bbaf6faf (0111000011110000111100101101001101011011011101010011010101001111)
Hamming distance: 32

Security contact information

To report a security vulnerability, follow these steps.