I'm trying to identify snags using LiDAR stored in las format. The Wing 2015 algorithm used in segment_snags requires the values be in the range from 0 to 255. Is it possible to edit and scale intensity values (not by adding a new attribute)? Alternatively, is it possible to edit the segment_snags tool in lidR?
Add a comment
|
1 Answer
If your intensities are recorded on 16 bits you can use the data.table syntax
las@data[, Intensity := as.integer(Intensity/(2^16-1)*255L)]
If your intensities are somehow arbitrary recorded
las@data[, Intensity := as.integer(Intensity/max(Intensity)*255L)]
With a more regular R syntax
las$Intensity = as.integer(las$Intensity/max(las$Intensity)*255L)
I don't remember how Wing's algo work but first read the original paper to ensure that rescaling like that has a meaning for the method.
Alternatively, is it possible to edit the
segment_snagstool in lidR?
This might be the object of a feature request.
-
After testing, rescaling does indeed work. However, some snags that fall within the right intensity range do not end up being classified as snags. My guess is that the BBPR thresholds have to be manipulated but this would be a result of the nature of the lidar itself and not the algorithm or its implementation.– Ray JCommented Jul 10, 2020 at 15:53
-
It should be noted that arbitrary scaled intensities are non-conformant to the las specification and anyone that supplies such data should be kicked. Incidentally don't expect much from intensity. practically it is a very poor metric. The author only claimed 56% detection rate. Commented Mar 31, 2023 at 0:14