MATLAB Code for K-Harmonic Means clustering

Hi , how I can write Matlab code for K-Harmonic Means clustering to extract feature from image background.I respectfully request to reply my question as soon as possible.

[Image Analyst]

I haven't heard of that one. Where did you hear of it? There is just the regular k-means in the Stats toolbox. Would that work for you? Post your image if you want advice on how to extract features from the background. Usually though, people want to ignore the background and extract features from the foreground. But actually you can get information from both the foreground and background if you want.

[D.Regan]

hi..! This is the paper in which the implementation of K-Harmonic Means available,"Spatial kernel K-harmonic means clustering for multi-spectral image segmentation".

Answer 1

[Bernhard Suhm]

This algorithm is not supported in the Statistics and Machine Learning toolbox, only standard k-means with various distance functions. However you could try to use regular K-Means to do this. You will have to carefully select the feature space to cluster in as 3x image dimensions are likely too big and noisy.

Our usual recommendation is to use the Image Segmented app from the Image Processing toolbox and try a few segmentation techniques that use more sophisticated clustering under the hood. If that doesn’t work you could try using extractFeatures from the Computer Vision toolbox.