JImageHash is a performant perceptual image fingerprinting library entirely written in Java. The library returns a similarity score aiming to identify entities which are likely modifications of the original source while being robust various attack vectors ie. color, rotation and scale transformation.

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.

This library was inspired by Dr. Neal Krawetz blog post "kind of like that" and incorporates several improvements. A comprehensive overview of perceptual image hashing can be found in this paper by Christoph Zauner.

The project is hosted on maven central

<dependency>
	<groupId>dev.brachtendorf</groupId>
	<artifactId>JImageHash</artifactId>
	<version>1.0.0</version>
</dependency>

<!-- If you want to use the database image matcher you need to add h2 as well -->
<dependency>
	<groupId>com.h2database</groupId>
	<artifactId>h2</artifactId>
	<version>1.4.200</version>
</dependency>

Please be aware that migrating from one major version to another usually invalidates created hashes in order to retain validity when persistently storing the hashes. The algorithm id of hashes is adjusted in order for the jvm to throw an error if the possibility exist that hashes generated for the same input image are not consistent throughout the compared versions.

Hashes generated with the following 2 algorithm have to be regenerated:

• RotPAverage hash was fixed to correctly return hashes when the algorithm is used multiple times.
• KernelAverageHash algorithm id changed due to JVMs internal hashcode calculation and the package name update. Hashes generated with this algorithm have to be regenerated.

The package is now published to maven central under a new group id. The internal package structure has been adjusted from com.github.kilianB to dev.brachtendorf.jimagehash. Adjust your imports accordingly.

File img0 = new File("path/to/file.png");
File img1 = new File("path/to/secondFile.jpg");

HashingAlgorithm hasher = new PerceptiveHash(32);

Hash hash0 = hasher.hash(img0);
Hash hash1 = hasher.hash(img1);

double similarityScore = hash0.normalizedHammingDistance(hash1);

if(similarityScore < .2) {
    //Considered a duplicate in this particular case
}

//Chaining multiple matcher for single image comparison

SingleImageMatcher matcher = new SingleImageMatcher();
matcher.addHashingAlgorithm(new AverageHash(64),.3);
matcher.addHashingAlgorithm(new PerceptiveHash(32),.2);

if(matcher.checkSimilarity(img0,img1)) {
    //Considered a duplicate in this particular case
}

Examples and convenience methods can be found in the examples repository

Support for transparent images has to be enabled specifically due to backwards compatibility and force users of the libraries to understand the implication of this setting.

The setOpaqueHandling(Color? replacementColor, int alphaThreshold) will replace transparent pixels with the specified color before calculating the hash.

• the replacement color must be consistent throughout hash calculation for the entire sample space to ensure robustness against color transformations of the images.
• the replacement color should be a color that does not appear often in the input space to avoid masking out available information.
• when not specified Orange will be used as replacement. This choice was arbitrary and ideally, a default color should be chosen which results in 0 and 1 bits being computed in 50% of the time in respect to all other pixels and hashing algorithms.
• supplying a replacement value of null will attempt to either use black or white as a replacement color conflicting with the advice given above. Computing the contrast color will fail if the transparent area of an image covers a large space and comes with a steep performance penalty.

HashingAlgorithm hasher = new PerceptiveHash(32);

//Replace all pixels with alpha values smaller than 0-255. The alpha value cutoff is taken into account after down scaling the image, therefore choose a reasonable value.  
int alphaThreshold = 253;
hasher.setOpaqueHandling(alphaThreshold)

The persistent package allows hashes and matchers to be saved to disk. In turn the images are not kept in memory and are only referenced by file path allowing to handle a great deal of images at the same time. The cached version keeps the BufferedImage image objects in memory allowing to change hashing algorithms on the fly and a direct retrieval of the buffered image objects of matching images. The categorize package contains image clustering matchers. KMeans and Categorical as well as weighted matchers. The exotic package features BloomFilter, and the SingleImageMatcher used to match 2 images without any fancy additions.

Image		High	Low	Copyright	Thumbnail	Ballon
High Quality
Low Quality
Altered Copyright
Thumbnail
Ballon

Image matchers can be configured using different algorithm. Each comes with individual properties

Image clustering with fuzzy hashes allowing to represent hashes with probability bits instead of simple 0's and 1's

See the wiki page on how to test different hashing algorithms with your set of images Code available at the example repo: https://github.com/KilianB/JImageHash-Examples/tree/main/src/main/java/com/github/kilianB/benchmark