I am not too familiar with DBSCAN images. Is there a reason the segmentation had to be done with R? Would you be able to post an example image?

Cheers,

John

https://www.r-bloggers.com/color-quantization-in-r/

The example uses k-means clustering, should probably also work using DBSCAN? There is a R-package 'dbscan' available on CRAN.

Tobias Egli thanks for your reply,

I will try to adapt this example to my segmentation problem, but I still have some unclear points like how to plot the segmented image using the clusters labels obtained using 'DBSCAN'.

For example : If I do like this seg=dbscan(img1,eps,minpts)

how I will plot the result? does it works like that : plot(img1,col=seg) or using function image(seg)?

John Sosa thanks for your reply,

I want to use the implemented version of DBSCAN in R to detect skin lesion. DBSCAN algorithm is used because it is able to detect arbitrary shaped clusters. 

You could do something like this:

pic

Tobias Egli thanks for your help.