LiDARHD to raster image

Use the sampling to create raster image with classification property

To raster the LiDAR point cloud, to save the memory, we can only save one label for each pixel, because building is the important part in the project (GENESE). So if in the pixel, there is a building point, we set this pixel to building label. For the other class, we save the last label.

The out put file is a Tiff file with tfw, so that this file can have the geo-information, and this can be open with QGIS, so that the output can be overlap with the other data.

After discussing with Bruno, we think it's better to use a priority and z value to select the class, so the priority should be set, the class should set a positive value for the priority. (This first version is equal to building is 1 and the other classes are 0).

Introduction

The class in LiDARHD, detail can find in the document :

Valeur	Description
1	Non classé
2	Sol
3	Végétation basse (0-50 cm)
4	Végétation moyenne (50 cm-1,50 m)
5	Végétation haute (+1,50 m)
6	Bâtiment
9	Eau
17	Tablier de pont
64	Sursol pérenne
65	Artefacts
66	Points virtuels
67	Divers - bâtis

~~For class 66, "Seuls les « nuages de points optimisés » contiennent des points classés avec cette valeur.", so at present, this class is not shown in the image.~~

For the class 64-67, the color bar depends on the QGIS label, so the color probably not same with QGIS.

Dependency

The code depends on my code projects :

OpenImageoiioLib

Because the code depends on OpenImageIO , because we only write Tiff format, so it is easy to install.

# libtiff
sudo apt-get update && apt-get install -y \
		    build-essential \
		    make \ 
            git \
			automake \
			libtool \
			zlib1g-dev \
			libboost-all-dev \
			libtiff-dev \
			libpng-dev \
			libjpeg-dev \
			libeigen3-dev

# code from OpenImageIO
git clone --recurse https://github.com/OpenImageIO/oiio.git && \
    cd oiio && \
    mkdir build && \
    cd build && \
    cmake -DOIIO_BUILD_TESTS=0 -DUSE_OPENGL=0 -DUSE_QT=0 -DUSE_PYTHON=0 -DUSE_FIELD3D=0 -DUSE_FFMPEG=0 -DUSE_OPENJPEG=1 -DUSE_OPENCV=0 -DUSE_OPENSSL=0 -DUSE_PTEX=0 -DUSE_NUKE=0 -DUSE_DICOM=0 ../ && \
    make -j16 && 
    sudo make install

WuLasLib

This part is already in the code repository, so it doesn't matter.

Example

It's easy to use the code, the comande line is :

EXE LAS GSD TIF (WEIGHT)

If there is no WEIGHT file, so the max Z is used in the class priority.

There is an example using the LiDARHD :

 cd LiDAR_Raster_Class
 mkdir build
 cd build
 cmake ..
 make -j4
 
 # EXE LAS GSD TIF
 ./ConvertLasImage '/home/whuwuteng/AI4GEO-Toulouse-IMAGE/ALS/LHD_FXX_0574_6277_PTS_C_LAMB93_IGN69.copc.laz' 0.15 '/home/whuwuteng/AI4GEO-Toulouse-IMAGE/ALS/raster2.tif' ../weight.txt

The example is shown (convert to png) :


Raster image

Update

In order to analysis the building footprint in 2D, so only building class is considered in the output, so there are two versions of code, only produce the color of building :

./ConvertLasImageBat '/home/TWu-Admin/Lyon/all.copc.laz' 0.15 '/home/TWu-Admin/Lyon/all.copc2.tif'

The origin all class image is shown (convert to png) :


Raster image Lyon

Only building class image is shown (convert to png) :


Raster image Lyon building

The second version is similar, but the height of the building is an important information in 3D, so the height of the building is saved in an image, the command line is :

./ConvertLasImageHeigth '/home/TWu-Admin/Lyon/all.copc.laz'  0.15 '/home/TWu-Admin/Lyon/all.copc3.tif'

Large LAS file

In order to process Large LAS file, a new code is available :

./ConvertBigLasImageBat '/home/TWu-Admin/loos/all.copc.laz'  0.15 '/home/TWu-Admin/loos/all.copc.png'