I'm trying to use lidar point X, Y, and Z coordinates (from .las files) in calculations. I'm using Python 2.7.13 (x64), with laspy 1.5.0.
Reading attributes one at a time with the following is amazingly fast:
inFile = laspy.file.File(las_file, mode="r")
x = inFile.X
But things slow down considerably when concatenating different single-attribute ndarrays into one array with the following (for example, when getting nx3 array of positions):
coords = np.array((inFile.X, inFile.Y, inFile.Z)).transpose()
I see in Inserting LiDAR points (from laspy) in GeoDataFrame without using a numpy array? that getting inFile.X, inFile.Y, and inFile.Z is as fast as it is because they're numpy-wrapped memory views with little copying going on.
Is there a way to create numpy-wrapped memory views containing multiple attributes, so time doesn't need to be spent copying multiple single attributes to one array, on top of doing something with that array?
