1

I want to retile big TLS files into quite small chunks of maximum 20 m x 20 m size. I have to normalize the point cloud, thus I need to classify the points and create a DTM for each tile. For this process, especially smaller tiles seem to cause me trouble (I guess it's hard to fit a plane in only few points). But I noticed when splitting up the data into chunks, there are quite many chunks with only little information.

ctg <- readTLSLAScatalog(path_points)
opt_chunk_buffer(ctg) <- 0
opt_chunk_size(ctg) <- 20
plot(ctg, chunk=TRUE)

the chunk positions

Can I somehow reposition the tiles so the bottom left corner of the data is also the bottom left corner of the first chunk? Or (even better) a possibility to center the tiles and the chunks so all chunks on the borders have similar sizes? I saw there is an option opt_chunk_alignment(), but I don't understand how it works exactly. Also, I guess I could play around with the chunk size because it might help, but I want to automize the process for several LAS files.

Improve this question

1 Answer 1

Reset to default
0

I want to retile big TLS files into quite small chunks of maximum 20 m x 20 m size

I don't know what kind of problem you are trying to solve be it sounds like a bad idea. Moreover you can do it on-the-fly without physically splitting your data.

Can I somehow reposition the tiles so the bottom left corner of the data is also the bottom left corner of the first chunk? [...] I saw there is an option opt_chunk_alignment(), but I don't understand how it works exactly.

It does exactly what you want to do.

opt_chunk_alignment(ctg) <- c(min(ctg$Min.X), min(ctg$Min.Y))

Or (even better) a possibility to center the tiles and the chunks so all chunks on the borders have similar sizes?

There is no such option and I'm not sure to understand what you are expecting actually.

Improve this answer
6
  • My issue is that I have 30GB not normalized point clouds, but 16GB RAM and I want them normalized. Since I am pretty new to all this point cloud data stuff, I obviously have no better idea how to achieve this. I want to minimize the amount of chunks which only have little information within. In other words: I want to cut down the amount of chunks while maintaining chunk size, by aligning them nicely. I want to avoid chunks having only one stripe of 1 m x 20 m data.
    – Zoe
    May 19, 2021 at 11:00
  • 1
    I understand but you could split you data in 4 parts not 176.
    – JRR
    May 19, 2021 at 11:09
  • Well, yes. I just wondered if it was possible. Sorry for bothering. The thing is, that the data is quite unevenly distributed across the area, but I guess I will have to try bigger chunks then.
    – Zoe
    May 19, 2021 at 11:15
  • No bothering here. I'm just saying that having pixel-sized tiles is unlikely to solve efficiently your problem and I think it will ultimately generate more problems than solution.
    – JRR
    May 19, 2021 at 11:23
  • I will later process the data with a resolution of 1 cm = 1 pixel. So 20 m would be 200 pixels. I thought with such a high resolution (and about 25k points per squaremeter), this maybe could be reasonable. But as I said, this is all new to me.
    – Zoe
    May 19, 2021 at 11:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.