The perceptual hashes are calculated from reduced 8x8 grayscale images:
Algorithm | Description |
---|---|
Average | Use the pixel's value and the average color |
Block | Divide image in blocks and use their mean value |
Difference | Use the difference between adjacent pixels |
Perceptive | Use a discrete cosine transform (DCT) and compare frequencies rather than color values |
Find similar images can be built from source with the 'SimilarImages.sln' Visual Studio 2013 solution.
Install the NuGet package 'OpenCvSharp3-AnyCPU' https://github.com/shimat/opencvsharp
PM> Install-Package OpenCvSharp3-AnyCPU
David Laperriere [email protected]