Finding buildings height using LIDAR data

1
votes

I am trying to evaluate buildings height using LIDAR data, for simplicity let's say that the highest point will do.

My problem is to filter the high peaks and the noise in data. I tried to take the polygon area of the building, and the area around it, bucket the data into buckets of 1 feet and filter the small buckets, then subtract the building height from the ground height in order to get the real height but it's not working so well.

I also thought about machine-learning with regression algorithm.

My questions:

  1. Are there tools and ready made functions that can help me with the building height evaluation? What are the best tools to filter LIDAR data?

  2. I read about rapidlasso ground classification --> building heights, do you think it cab help me evaluate the height more accurately?

  3. Are there sources for some real (ground-truth) building heights in the us, so I can train and evaluate my model?

Any other input in the subject will be appreciated. Thanks.

1
lidarlidar-data

1 Answers

2
votes

As you are already looking at rapidlasso, and obviously LAStools, I would say you need to look no further as LAStools has all the 'tools' you require.

There are several posts on the rapidlasso site regarding how to approach this and you can join the email group and post questions out to the many users. I dont know what system you are using, but LAStools will also integrate with ArcMap - so perhaps this is another software you are familiar with?

I don't classify buildings, but I do classify heights of other things eg trees/vegetation and yes, lasground is very useful for this.

Sorry, but I don't know of any US ground truth sources. Can you not train your model with local data ie buildings that you have measured with a clinometer or similar?