Implementation of DBSCAN using R Trees

I am trying to implement DBSCAN using R开发者_如何学Go tree.We can store data in the form of R trees.So my question is how can i store real time data in R trees and how should i implement region query to find neighborhood of a point with it?

---

Implement R-Trees first, then DBSCAN.

As for realtime, you probably want to look into specialized clustering algorithms for data streams.

---

I am not sure what you mean by real-time data. In case you refer to changing data or stream data, you can just as well delete stuff that is outdated from the R-tree. If you mean that your data also has a time-dimension than you can simply extend the number of coefficients that the R-tree manages to (x,y,t) -- I assume you use the simple bi-variate version.

If you want to implement DBSCAN you will need to perform range queries in order to compute densities of spherical areas around points. Therefore, your region queries should be able to handle spherical query regions (for that matter I would also suggest you check out the SR-Tree by Shinichi Satoh, might be helpful). Again, if by real time data you mean that your data has a time-feature, you might want to consider using query regions that are ellipsoidal (allow for a separate scaling of the spatial and the temporal features).

Hope that helps :]