ImageDupLib is a .NET standard library offering several different perceptual hashing algorithms for detecting similar or duplicate images.

DupImageLib is available as a Nuget package. You can laso donwload the Nuget packages from the releases page.

DupImageLib implements the following perceptual hash algorithms:

• Average

  • Calculates the hash based on the average value of scaled down image. More in-depth explanation of the algorithm can be found here. Extremely fast algorithm, but generates a lot of false positives. Can generate 64 or 256 bit hashes.

• Median

  • Similar to average, but instead of average pixel value, median value is used. This should make the algorithm bit more resistant to non-linear changes in image. Slightly slower than average hash. Can generate 64 or 256 bit hashes

• Difference

  • Constructs the hash by comparing the gradients of pixel values on each row of the scaled image. In-depth explanation found here. Performs faster than the average hash and provides better results. Produces 64 or 256 bit hashes.

• DCT (pHash)

  • Discrete cosine transform based algorithm. Similar to the one implemented in pHash library. Detailed explanation of the algorithm can be found here. Slower than any of the previous algorithms, but has better tolerance to image modifications. Produces 64 bit hashes.

All algorithms accepts the image input as a stream or as a path to filesystem location.

DupImageLib also provides a function to compare hashes to return the similarity of the hashes.

DupImageLib uses Magick.NET for image processing needs. Users of DupImageLib can use their own image processing library by providing an implementation of IImageTransformer and passing that to the ImageHashes constructor.

// Create new ImageHashes instance using ImageMackick as image manipulation library
var imageHasher = new ImageHashes(new ImageMagickTransformer());
// Calculate 64 bit hashes for the images using difference algorithm
var hash1 = imageHasher.CalculateDifferenceHash64(@"testimage1.png");
var hash2 = imageHasher.CalculateDifferenceHash64(@"testimage2.png");
// Calculate similarity between the hashes. Score of 1.0 indicates identical images.
var similarity = ImageHashes.CompareHashes(hash1, hash2);

This software is licensed under Apache 2.0 license. See LICENSE file for more information.