No announcement yet.

Working with Lidar data in WW

  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    Yes, the app is contained in the branch.


    • #17
      Here are some recent updates and current directions. We have been working on segmentation of the unorganized point cloud data in an attempt to isolate structures like buildings from the lidar data and generate explicit boundary representations for them. This effort begins by defining filters that utilize the various per-point attributes in an attempt to extract just those points that are likely to be buildings. Our friends from the City of Springfield, Oregon have been developing some very nice filters for that part of the process. Given the (still spatially unorganized) points that pass through those filters, I send them through an implementation of the DBSCAN algorithm to collect individual objects together into so-called "clusters". Each cluster then corresponds to a building. The points in the building cluster are still spatially unorganized, so I create "Partitioned Neighborhoods" (PNs) for each point in a cluster. These PNs allow me to determine first what the boundary points of the cluster are, then I can also use the PNs to initiate a spatial ordering of the identified boundary points into a set of edges that completely bound the cluster/building. The boundaries tend to be a bit noisy and have excessive numbers of points, so I am currently experimenting with a set of postprocessing filters that use various strategies to thin out vertices, refine edge geometry, and sharpen corners. This work is ongoing, as I am still exploring how best to tune these individual postprocessing filters as well as determine the best way they can operate - in unison, one after the other, etc.


      • #18

        I work for the dod and have the same needs to work with Lidar within my ww application. Can you please provide a link to the respository location where the source code for ingesting Lidar data is stored. (Myspringfield)