@author Kilian

An abstract implementation of the categorical image matcher supporting classes to keep track of the categories an image was matched to. <p> It is up t

Settings used while computing if an algorithms consideres two images to be a close match @author Kilian

Calculate a hash value based on the average rgb color in an image. @author Kilian @since 2.0.0 similar to ahash from version 1.0.0

@author Kilian

Calculate a hash value based on the average luminosity in an image. @author Kilian @since 1.0.0 @since 2.0.0 use luminosity instead of average pixel

Calculate a hash value based on the average luminosity in an image. The kernel hash variant applies additional kernels to the already rescaled version

@author Kilian

A not thread safe binary tree implementation used to quickly compute the <a href="https://en.wikipedia.org/wiki/Hamming_distance">hamming distance</a>

A bloom filter is an approximative data structure with constant space and time guarantee to check if an image is likely in the set or definitely not p

Image matcher allowing to group images into similar categories. <p> Usually cluster results are requested immediately and images may get added over ti

Cluster images into common categories. This matcher clusters images by computing the distance to the closest cluster and adds an image if it is within

@author Kilian

A categorization result describes the membership to a cluster. It contains the identifier of the category this image was matched to as well as a quali

@author Kilian

@author Kilian @since 3.0.0

@author Kilian

@author Kilian

Convenience class allowing to chain multiple hashing algorithms to find similar images. The ConsecutiveMatcher keeps the hashes and buffered images in

Convenience class allowing to chain multiple hashing algorithms to find similar images. The ConsecutiveMatcher keeps the hashes and buffered images in

Convenience class allowing to chain multiple hashing algorithms to find similar images. The CumulativeMatcher keeps the hashes and buffered images in

Instead of early aborting if one algorithm fails like the {@link dev.brachtendorf.jimagehash.matcher.cached.ConsecutiveMatcher}, this class looks at t

@author Kilian

An extended hash class allowing dhashes to be visually represented. @author Kilian @since 3.0.0

A naive database based image matcher implementation. Images indexed by this matcher will be added to the database and retrieved if an image match is q

Calculates a hash based on gradient tracking. This hash is cheap to compute and provides a high degree of accuracy. Robust to a huge range of color tr

@author Kilian

Currently only tests index out of boundes exceptions @author Kilian

The edge handling strategy defines the behaviour when a kernel reaches the edge of an image and does not have enough information to compute it's value

A filter modifies the content of an image while leaving the source image intact. @author Kilian @since 2.0.0

A binary tree implementation allowing to quickly lookup be aware that fuzzy trees currently are of one time use and are rendered invalid as soon as o

@author Kilian

A fuzzy hash is an aggregation of multiple hashes mapped to a single mean hash representing the average hash while keeping track of fractional (probab

@author Kilian

A tree node saving references to it's children. @author Kilian @since 3.0.0

Kernels whose filter class work with grayscale instead of seperate color channels @author Kilian

A naive database based image matcher implementation. Images indexed by this matcher will be added to the database and retrieved if an image match is q

@author Kilian

Hashes are bit encoded encoded values (0101011101) created from images using a hashing algorithm. Hashes enable a quick approximate similarity compari

Helper class to quickly create a bitwise byte array representation which can be converted to a big integer object. <p> To maintain the capability to

Base test class applicable to all hashing algorithms @author Kilian

@author Kilian

Base class for hashing algorithms returning perceptual hashes for supplied images reducing the number of bits needed to represent said image. <p> Opp

Image Hash on HOG feature descriptor. Not ready yet. Most likely use a very high bit resolution similar to how hog feature descriptors are actually us

@author Kilian @deprecated not ready to use yet @since 2.0.0

@author Kilian

Image Hash on HOG feature descriptor. Not ready yet. Most likely use a very high bit resolution similar to how hog feature descriptors are actually us

@author Kilian

@author Kilian

@author Kilian

The kMeans approach requires to know the number of clusters beforehand. @author Kilian @since 3.0.0

@author Kilian

Kmeans plus plus implementation. Opposed to Kmeans this algorithm strategically chooses it's starting clusters to decrease iteration time at the later

@author Kilian