Python-Remote-Sensing-Scripts

A set of python scripts for remote sensing processing. All functions were created using Python 3.6

As an example, I will use a UAV-based hyperspectral image of a peatland in central-south Chile. The image have 41 bands (10 nm width) ranging form 480-880 nm and a pixel size of 10 cm. The green dots correspond to plots where measurements of biomass, species composition and carbon stock information were taken:

With our scripts you can first extract the spectra in the plots location in a few seconds

python ExtractValues.py -r peatland.tif -s plots.shp -i ID

Check out the output

	N	B1	B2	B3	B4	B5	B6	B7	B8	B9	...	B32	B33	B34	B35	B36	B37	B38	B39	B40	B41
0	1.0	1509.813451	2291.564899	3109.637130	3962.194325	4674.092289	5177.553225	5526.183572	5698.716701	5730.444486	...	15122.021006	15696.353105	15488.906034	15259.373692	15370.818661	15988.557692	16801.212295	16728.998597	16846.117603	17324.593108
1	2.0	1708.011608	2617.267914	3579.341584	4624.986406	5542.596761	6221.865500	6708.542956	6970.212103	7036.433086	...	20393.572983	20991.396063	20751.881674	20494.922936	20547.978802	21333.448745	21717.872021	21661.447804	21420.735874	21518.602349
2	3.0	12.875854	211.439792	1201.264700	2444.936965	3603.781002	4468.655196	5054.639060	5373.096807	5479.100261	...	40287.680262	40283.248111	39078.153197	37895.060076	38651.953385	39584.352791	40869.073804	39726.095768	39378.668239	37695.083542
3	4.0	214.950019	852.952500	1827.763722	2946.394303	3962.429581	4700.627478	5189.396823	5428.162923	5479.461922	...	16865.061517	17382.712572	17022.812516	16595.132046	16618.242789	17052.205065	17653.929336	17440.521948	17410.552106	17704.214614
4	5.0	2704.663992	3474.524823	4291.852040	5139.117787	5834.414719	6324.879534	6673.001772	6841.418896	6859.041616	...	19038.206233	19714.628088	19361.587963	19008.547838	19256.278283	20024.457928	20623.883552	20655.435636	20865.324231	20861.972693

You can also perform a MNF transformation of the data. This function have several options, like applying Savitzky Golay filtering and brightness normalization of the spectra. The basic function is like (image resample to 2m in the example):

python MNF.py -i peatland.tif

Get the Gray-Level Co-Occurrence Matrix (GLCM) textures from an image. Here, we used the first MNF component as raster imput with a moving window of 5 X 5 pixels (default):

python GLCM.py -i peatland_MNF.tif

Finally, we can also obtain texture information from point clouds (in this case based in the UAV photogrametric point cloud) based on the Canupo algorithm proposed by This paper, which is also implemented in the CloudCompare LiDAR software. Nevertheless, both the paper and the software implement the transformation to generate poin-based classification while this python script produces texture rasters to be use in any application:

python canupo.py -i lidar.txt -s 1 5 1 -r 1
# scales: 1,2,3,4,5 m; output resolution 1 m