Running CARS with sensor images as input

The standard configuration uses sensor images as inputs. Additional parameters can be used in inputs configuration :

Name	Description	Type	Default value	Required
sensors	Stereo sensor images	dict	No	Yes
pairing	Association of sensor image to create pairs	list of pairs of sensors	No	Yes (*)
initial_elevation	Low resolution DEM	string or dict	No	No
roi	Region Of Interest: Vector file path or GeoJson dictionary	string or dict	None	No
loaders	Input format for image and classification	dict	cf. Loaders tab	No
filling	Areas to fill in final result	dict	cf. Filling tab	No

(*) pairing is required if there are more than two sensors (see pairing section below)

SensorsPairingInitial ElevationROILoadersFilling

Basic configuration

For each sensor image, give a particular name (what you want), associated with the image path:

YAML JSON Copy

---
input:
  sensors:
    my_name_for_this_sensor: path_to_image.tif
    my_name_for_that_other_sensor: path_to_other_image.tif

{
    "input": {
        "sensors": {
            "my_name_for_this_sensor": "path_to_image.tif",
            "my_name_for_that_other_sensor": "path_to_other_image.tif"
        }
    }
}

Intermediate configuration

For each sensor, auxiliary files can be used : mask, classification, and geomodel if needed

Name	Description	Type	Default value	Required
image	Path to the image or dictionary readable by a sensor loader	string, dict		Yes
edges	Path to the edges mask or dictionary with edges_mask, normals, depth_map, tile_id	string, dict		No
geomodel	Path to the geomodel and plugin-specific attributes	string, dict		No
mask	Path to the binary mask	string, dict	None	No
classification	Path to the classification image or dictionary readable by a sensor loader	string, dict	None	No
regularization_mask	Path to the regularization mask, used in surface modeling	string, dict	None	No

In most cases, only the file path is required for each of these parameters and CARS will know how to read each file :

YAML JSON Copy

---
input:
  sensors:
    my_name_for_this_sensor: # at least two of these are required
      image: path_to_image.tif
      mask: path_to_mask.tif
      classification: path_to_classification.tif

{
    "input": {
        "sensors": {
            "my_name_for_this_sensor": {
                "image": "path_to_image.tif",
                "mask": "path_to_mask.tif",
                "classification": "path_to_classification.tif"
            }
        }
    }
}

However for each parameter it is possible to set a dictionary with additional parameters on how to read the file.

Advanced configuration

These parameters are described on the tabs below :

ImageEdgesGeomodelMaskClassificationRegularization mask

The standard method for passing sensor images as inputs is to put only the path of the image.

If the images are multi-band, CARS will automatically perform the matching steps on the first band (for example if the image is RGB, CARS will correlate on the red band).

It is possible to use a dictionary to define the no_data value of the image.

Name	Description	Type	Default value	Available values	Required
path	File path	str			Yes
no_data	No data value of file	int	0		No

An example is given below :

YAML JSON Copy

---
image:
  path: img1.tif
  no_data: -9999

{
    "image": {
        "path": "img1.tif",
        "no_data": -9999
    }
}