i2_vectorization

builders

Name	Default Value	Description	Type	Mandatory	Name
builders_class_name	[‘i2_classification’]	The name of the class defining the builder	list	False	builders_class_name
builders_paths	/path/to/iota2/sources	The path to user builders	list	False	builders_paths

Notes

builders_class_name 

Available builders are : ‘i2_classification’, ‘i2_features_map’, ‘i2_obia’ and ‘i2_vectorization’

builders_paths 

If not indicated, the iota2 source directory is used: */iota2/sequence_builders/

chain

Name	Default Value	Description	Type	Mandatory	Name
output_path	None	Absolute path to the output directory.	str	True	output_path
proj	EPSG:2154	The projection wanted. Format EPSG:XXXX is mandatory	str	False	proj

Notes

output_path 

Absolute path to the output directory.It is recommended to have one directory per run of the chain

simplification

Name	Default Value	Description	Type	Mandatory	Name
angle	True	If True, smoothing corners of pixels (45°)	bool	False	angle
bingdal	None	path to GDAL binaries	str	False	bingdal
blocksize	2000	block size to split raster to prevent Numpy memory error	int	False	blocksize
chunk	10	Number of chunks for statistics computing	int	False	chunk
classification	None	Input raster of classification	str	True	classification
clipfield	None	field to identify distinct areas	str	False	clipfield
clipfile	None	vector-based file to clip output vector-based classification	str	False	clipfile
clipvalue	None	value of field which identify distinct areas	int	False	clipvalue
confidence	None	Input raster of confidence	str	True	confidence
douglas	10	Douglas-Peucker tolerance for vector-based generalization	int	False	douglas
dozip	True	Zip output vector-based classification (OSO-like production)	bool	False	dozip
grasslib	None	path to grasslib	str	False	grasslib
gridsize	None	Number of lines and columns of serialization process	int	False	gridsize
hermite	10	Hermite Interpolation threshold for vector-based smoothing	int	False	hermite
higher_stats	False	If True, compute more complexe statistics (Shanon, majority order and difference, etc.)	bool	False	higher_stats
inland	None	Inland water limit shapefile	str	False	inland
lcfield	Class	Name of the field to store landcover class in vector-based classification	str	False	lcfield
lib64bit	None	Path of BandMath and Concatenate OTB executables returning 64-bits float pixel values	str	False	lib64bit
mmu	1000	MMU of output vector-based classification (projection unit),(Default : 0.1 ha)	int	False	mmu
nomenclature	None	configuration file which describe nomenclature	str	False	nomenclature
outprefix	dept	Prefix to use for naming of vector-based classifications	str	False	outprefix
prod	None	OSO-like output vector (aliases) is produced. Other possible value : carhab	str	False	prod
rssize	20	Resampling size of input classification raster (projection unit)	int	False	rssize
seed	1	Seed of input raster classification	int	False	seed
statslist	{1: ‘rate’, 2: ‘statsmaj’, 3: ‘statsmaj’}	dictionnary of requested landcover statistics	dict	False	statslist
systemcall	False	If True, use yours gdal lib (cf. bingdal)	bool	False	systemcall
umc1	None	MMU for first regularization	int	False	umc1
umc2	None	MMU for second regularization	int	False	umc2
validity	None	Input raster of validity	str	True	validity
vectorize_fusion_of_classifications	False	flag to inform iota2 to vectorize the fusion of classifications	bool	False	vectorize_fusion_of_classifications
zonal_vector	None	vector file to compute zonal statistics of classification	str	False	zonal_vector

Notes

bingdal 

Some GDAL lib versions (automatically set up with iota2) are not efficient to handle topology errors, use yours !

blocksize 

Numpy memory error may occur for large areas during serialization process. Split in sub-rasters prevents memory error

chunk 

Number of chunks (groups of vector-based features) for parallel computing landcover statistics

classification 

This parameter is automatically set if the configuration file use the classification and vectorization builders

clipfield 

field to identify distinct geographical/administrativeareas (cf. “clipfile” parameter)

clipfile 

vector-based file can contain more than one feature (geographical/administrative areas). An output vector-based classification is produced for each feature (cf. ‘clipfield’ parameter).

clipvalue 

output vector-based classification is only produced on the specific area (clipfield=clipvalue in clipfile) (cf. ‘clipfield’ parameter). If None, all areas are produced.

confidence 

This parameter is automatically set if the configuration file use the classification and vectorization builders

grasslib 

Some functions of GRASS GIS software are used to vectorize, simplify and smooth vector layer. This path corresponds to GRASS install folder

This parameter is useful only for large areas for which vectorization process can not be executed (memory limitation). By ‘serialization’, we mean parallel vectorization processes. If not None, regularized classification raster is splitted in gridsize x gridsize rasters

inland 

to vectorize only inland waters, and not unnecessary sea water areas

lib64bit 

Band math and concatenate OTB executables with 64 bits capabilities (only for large areas where clumps number > 2²³ bits for mantisse)

nomenclature 

This configuration file includes code, color, description and vector field alias of each class

Classes:
{
        Level1:
        {
                "Urbain":
                {
                code:100
                alias:"Urbain"
                color:"#b106b1"
                }
                ...
        }
        Level2:
        {
               "Urbain dense":
               {
               code:1
               alias:"UrbainDens"
               color:"#ff00ff"
               parent:100
               }
               ...
        }
}

outprefix 

Naming of vector-based classifications is as following : prefix_clipvalue

rssize 

OSO-like vectorization requires a resampling step in order to regularize and decrease raster polygons number, If None, classification is not resampled

seed 

This parameter is usefull to vectorize one specific output classification seed

statslist 

Different landcover statistics can be computed for vector-based classification file. A python-like dictionnary must be provided. This is the OSO-like statistics :{1: “rate”, 2: “statsmaj”, 3: “statsmaj”}

Where:

{1: “rate”} : rates of classification classes are computed for each polygon
{2: “statsmaj”} : descriptive stats of classifier confidence are computed for each polygon by using only majority class pixels
{3: “statsmaj”} : descriptive stats of sensor validity are computed for each polygon by using only majority class pixels

list of available statistics :

stats : mean_b, std_b, max_b, min_b
statsmaj : meanmaj, stdmaj, maxmaj, minmaj of maj. class
rate : rate of each pixel value (classe names)
stats_cl : mean_cl, std_cl, max_cl, min_cl of one class

umc1 

It is an interface of parameter ‘-st’ of gdal_sieve.py function. If None, classification is not regularized

umc2 

OSO-like vectorization process requires 2 successive regularization, if you need a single regularization, let this parameter to None

validity 

This parameter is automatically set if the configuration file use the classification and vectorization builders

vectorize_fusion_of_classifications 

This flag is only useful if the vectorization is chained with classification workflow

zonal_vector 

Compute the zonal statistics on the geometries ofa user-provided vector file. Zonal statistics arecomputed on classification, confidence and rasterinput rasters

task_retry_limits

Name	Default Value	Description	Type	Mandatory	Name
allowed_retry	0	Allow dask to retry a failed job N times.	int	False	allowed_retry
maximum_cpu	4	The maximum number of CPU available	int	False	maximum_cpu
maximum_ram	16	The maximum amount of RAM available. (gB)	int	False	maximum_ram

Notes

maximum_cpu 

the amount of cpu will be doubled if the task is killed due to ram overconsumption until maximum_cpu or allowed_retry are reach

maximum_ram 

the amount of RAM will be doubled if the task is killed due to ram overconsumption until maximum_ram or allowed_retry are reach