demcompare.classification_layer.classification_layer_template

Mainly contains the ClassificationLayer class. A classification_layer defines a way to classify the DEMs alti differences.

Module Contents

Classes

ClassificationLayerTemplate

ClassificationLayer class
class demcompare.classification_layer.classification_layer_template.ClassificationLayerTemplate(name: str, classification_layer_kind: str, cfg: Dict, dem: xarray.Dataset = None)[source]

ClassificationLayer class A classification_layer defines a way to classify the DEM for the stats computation.

fill_conf_and_schema(cfg: demcompare.internal_typing.ConfigType = None) → demcompare.internal_typing.ConfigType[source]

Add default values to the dictionary if there are missing elements and define the configuration schema

Parameters:

cfg (ConfigType) – coregistration configuration

Return cfg:

coregistration configuration updated

Return type:

ConfigType

check_conf(cfg: demcompare.internal_typing.ConfigType = None) → demcompare.internal_typing.ConfigType[source]

Check if the config is correct according to the class configuration schema

raises CheckerError if configuration invalid.

Parameters:

cfg (ConfigType) – coregistration configuration

Return cfg:

coregistration configuration updated

Return type:

ConfigType

compute_classif_stats(data: xarray.Dataset, stats_dataset: demcompare.stats_dataset.StatsDataset, metrics: List[dict | str] = None)[source]

Stats are computed based on the classification layers, which define classes of pixels that classify divide the input image. Stats are computed on each classes separately.

Input dems can be classified by two maps belonging to the same classification_layer. If both maps exist, then this method produces stats based on 3 modes:

standard mode, intersection mode: where only alti errors values associated with intersection classes between both classified images are used, exclusion mode: the intersection complementary one

Parameters:

  • data – array to compute stats from

  • stats_dataset (StatsDataset) – StatsDataset object

  • metrics (List[Union[dict, str]]) – metrics to be computed

Returns:

stats, masks, names per mode

Return type:

List, List List

create_metrics(input_metrics: List[dict | str] = None) → Tuple[Dict[str, demcompare.metric.Metric], List[bool]][source]

Create metric objects and remove_outliers_list

Parameters:

input_metrics (List of Dict and str) – list of input metrics

Returns:

Dict with metric objects and list with outlier handling per metric

Return type:

Tuple[Dict[str, Metric], List[bool]]

_get_outliers_free_mask(array: numpy.ndarray, nodata_value: int | None = None) → numpy.ndarray[source]

Get outliers free mask (array of True where value is no outlier) with values outside (mu + 3 sigma) and (mu - 3 sigma). Nan and nodata_value are not considered in mu and sigma computation.

Parameters:

  • array (np.ndarray) – input array to get the mask from

  • nodata_value (int or None) – no data value considered. Default: None

Returns:

outliers free mask (array of True where value is no outlier)

Return type:

np.ndarray

_create_mode_masks(alti_map: xarray.Dataset)[source]

Compute Masks for every required modes :

the ‘standard’ mode: nan free, nodata free mask

the ‘intersection’ mode: which is the ‘standard’ mode where only the pixels for which both sets (sec and ref) are intersection

the ‘exclusion’ mode: which is ‘intersection’ complementary

Note that ‘intersection’ and ‘exclusion’ mode masks can only be computed if len(_sets_masks)==2

Parameters:

alti_map – alti differences

Returns:

list of masks, associated modes, and error_img read as array

Return type:

List[np.ndarray]

static _get_nonan_mask(array: numpy.ndarray, nodata_value: int | None = None) → numpy.ndarray[source]

Get no data and nan mask value

Parameters:

  • array (np.ndarray) – input array to get the mask from

  • nodata_value (int or None) – no data value considered. Default: None

Returns:

nan and nodata_value if exists mask on array.

Return type:

np.ndarray

_compute_mode_stats(dz_values: numpy.ndarray, stats_dataset: demcompare.stats_dataset.StatsDataset, mode_mask: numpy.ndarray = None, mode_name: str = None, metrics: List[dict | str] = None) → demcompare.stats_dataset.StatsDataset[source]

Get stats for a specific mode

Parameters:

  • dz_values (np.ndarray) – alti map

  • stats_dataset (StatsDataset) – StatsDataset object

  • mode_mask (List[bool]) – boolean mask with True values for pixels to use

  • mode_name (str) – mode name

  • metrics (List[Union[dict, str]]) – metrics to be computed

Returns:

StatsDataset with computed metrics (set_name, nbpts, %(out_of_all_pts), max, min, mean, std, rmse, …)

Return type:

StatsDataset

_create_class_masks()[source]

Returns a list of masks, by class. Each masks indicates which pixels belong to the class.

Returns:

None

abstract _create_labelled_map()[source]

Creates labelled map :return: None

stats_computation(data: numpy.ndarray, outliers_free_data: numpy.ndarray, input_metrics: List[str | Dict] = None) → Dict[source]

Compute stats for a specific array

Parameters:

  • data (np.ndarray) – 2D input data

  • outliers_free_data (np.ndarray) – input outliers_free_data

  • input_metrics (List[Union[str, Dict]]) – input metrics to use

Returns:

dict with computed metric values

Return type:

Dict

save_map_img(map_img: numpy.ndarray, map_support: str)[source]

Save the classification layer map to file

Parameters:

  • map_img (np.ndarray) – input data

  • map_support (str) – map support “ref” or “sec

Returns:

None