From cloud to image

We have to visually inspect every LiDAR cloud file we receive, in a way to find some problems that are not (yet) possible to find by programming. This is a very boring and error-prone process, even in a fast computer. It takes time to open LiDAR cloud files for visualization, as the clouds have around 50 million points each. In the first bach files we received, it took us around two weeks to inspect 149 clouds.

Therefore, we decided to generate a TIFF for each cloud, and use an image visualization software to inspect the flight results. Then, we created a small Python script to do that. In the first images we create we “grayscaled” the points height, and produce images like that:

Nevertheless, what about to use colors? In addition, what if we find a color palette that will make a forest-based cloud points looks like forest? Some Googles after, we could find some interesting “Earth tones” palette that made a nice job:

Each TIFF color image is around 100 Mbytes sized, comparing to 3 Gbytes corresponding LAS file. It took us far less time to check all clouds and we find more points of interest than when we use LAS file itself. The script we used was https://github.com/assismauro/pyLiDARForest/blob/master/stuff/las2img.py.

It takes as params:

inputfilename: LAS file to be processed

destinationpath: output path. File name will be the original LAS file with TIF extension.

-s, --scale: image scale, that is, the ration between image size and LAS file size. Default value is 0.0018, and you can try with this.

-p, --pallete: color palette to be used,options: 0: grayscale, 1: earthtones pallete, 2. Opencv green to yellow.

-f, --imagefformat, image file format, options: 0: tiff, 1: jpeg, 2. png

-v, --verbose: show progress messages.

That´s it!

#Python #LiDAR #Amazon