Hello,
I have two datasets;
1. Point cloud created from RGB photography aquired from a Hasselblad A6D and processed in Photoscan
2. Point cloud sourced from a commercial LiDAR provider with ground classified.
The area covered by both data sets is approximately 7km x 23km
I am attempting to adjust the z value of my Photoscan sourced point cloud to match a LiDAR sourced DEM.
I am making the assumption that the LiDAR sourced DEM is positioned correctly.
The two datasets match very well in x and y only the z requires adjustment
The following workflow throws an error if I have the -step less than 100 for the LAStools function lasthin.
REM Use LiDAR Class 2 points to create control point file
lasthin -i INPUT_FOLDER\*.las -odir OUTPUT_FOLDER -keep_class 2 -step 75 -central
REM Merge thinned files
lasmerge -v -i INPUT_FOLDER\*.las -o OUTPUT_FOLDER\merge_thin.las
REM Convert merged control point las to csv file
las2txt -i INPUT_FOLDER\merge_thin.las -o OUTPUT_FOLDER\cp.csv -parse xyz -sep comma
REM Apply control point file to photoscan sourced point cloud
lascontrol -i INPUT_FOLDER\*.las -cp INPUT_FOLDER_CONTROL_POINT\cp.csv -adjust_z -odir OUTPUT_FOLDER
The following error was thrown for -step 75
U:\>lascontrol -i INPUT_FOLDER\*.las -cp INPUT_FOLDER_CONTROL_POINT\cp.csv -adjust_z -odir OUTPUT_FOLDER
WARNING: files have different point types: 2 vs 1
WARNING: files have different point sizes: 26 vs 28
ERROR (TINclean): failed malloc for 54611274 TINtriangles.
ERROR: cannot alloc enough TIN triangles to triangulate 27305637 points.
The following error was thrown for -step 50
U:\>lascontrol -i INPUT_FOLDER\*.las -cp INPUT_FOLDER_CONTROL_POINT\cp.csv -adjust_z -odir OUTPUT_FOLDER
WARNING: files have different point types: 2 vs 1
WARNING: files have different point sizes: 26 vs 28
ERROR: cannot realloc point_buffer for 40000000 points.
The following error was thrown for -step 90
U:\>lascontrol -i INPUT_FOLDER\*.las -cp INPUT_FOLDER_CONTROL_POINT\cp.csv -adjust_z -odir OUTPUT_FOLDER
WARNING: files have different point types: 2 vs 1
WARNING: files have different point sizes: 26 vs 28
ERROR (TINclean): failed malloc for 38750484 TINtriangles.
ERROR: cannot alloc enough TIN triangles to triangulate 19375242 points.
When I set lasthin -step to greater than 100 lascontrol processes the files successfully (I assume).
This is the last couple of lines from a successful lascontrol using lasthin -i INPUT_FOLDER\*.las -odir OUTPUT_FOLDER -keep_class 2 -step 100 -central
2.83026,143.15,436243.89,6194047.01,140.32
WARNING: there were 359 control points without sufficient LIDAR coverage.
sampled TIN at 19671 of 20030 control points.
avgabs/rms/stddev/avg of elevation errors are 11.5168/15.4981/10.9885/10.9293 meter. skew is 0.695508.
Is there a way to process the lidar tiles without merging so that I can create an increased control point resolution?
Is 100m x 100m control point resolution more than enough?
---------------------------------------------------------
This question is a continuation of a question I asked Martin Isenburg.
Hello
Martin,
We
are using a Hasselblad A6D in a Cessna 172 and processing the images and
outputting a non-classified dense point cloud using Agisoft Photoscan
(Classified using LASTools).
We
have recently acquired ground classified LiDAR data from Fugro in Western
Australia.
What
I am attempting to do.
I
want to calculate the height of the trees using the surface model from the
photogrammetric point cloud sourced from photoscan and use the ground classified
points from LiDAR.
The
problem that I have is there is around 5m difference in Z between the two data
sets.
I
am currently using the las2las -reoffset 0 0 -5 function but have noted that
the lascontrol tool maybe a better solution.
Is
there a way to create a control file from the LiDAR sourced DEM that I could
then use to vertically reposition my photoscan pointcloud in Z.
And
is this a good way to approach the problem or is there other solutions.
And this is Martins reply
don't use reoffset for small translations as this changes
the offsets in the LAS header without moving the data but it can have bad
side-effect to do this (i.e. integer overflow). use '-translate_z 5.0' or
'-translate_z -5.0'.
Maybe you
can use lasthin to create a control point file. What spacing do you want? But
you should only pick control points in areas where the photoscan poincloud
points are *all* on the ground.
lasthin -i
lidar.laz -keep_class 2 -step 10 -central -o controlpoints.laz
Maybe we can discuss this (with links to example data) in
the LAStools user forum. There is only a vertical error? No horizontal one?
Regards
Ben
