How can I apply agglomerative hierarhical clustering based on the distance between regions?

Hello,

I am trying to apply hierarchical clustering based on the distance between regions. I was able to able apply clustering based on the centres of the white regions but could not find a way to cluster them based on the distance between areas (not between the centres). It is important to apply it between areas but not the centres because when I apply clustering on the centres, I could miss two big white regions. Attach you can find a representative image that I am dealing with.

Answer 1

[Akshat Dalal]

Hi Demir,

I understand that you want to do cluster analysis of irregularly shaped data but do not want to use centroid-based algorithms. Based on your requirements and the image you attached, I would suggest you to use density-based clustering algorithms instead of the widely used distance-based clustering algorithms such as 'k-means' clustering. To read more about density-based clustering and how it is different from distance-based clustering, you could refer the following documentation: https://www.geeksforgeeks.org/difference-between-k-means-and-dbscan-clustering/

DBScan is one such density-based clustering algorithm. The key fact of this algorithm is that the neighbourhood of each point in a cluster which is within a given radius (R) must have a minimum number of points (M). This algorithm has proved extremely efficient in detecting outliers and handling noise. It is availabe in MATLAB and you can use the 'dbscan' function to implement it. To read more about it, you could refer the following documentation: https://www.mathworks.com/help/releases/R2020a/stats/dbscan.html

Hope this helps!

Answer 2

[Image Analyst]

Not sure why I didn't see your question years ago (since I already have a demo), but I agree that dbscan seems to be the algorithm you're asking for. For more info see https://en.wikipedia.org/wiki/DBSCAN

Attached is my dbscan demo. Adapt as needed.