Use CARS with Pleiades images

Note

CARS is used in the same way with Pléiades and Pléiades Neo data.

With raw data

If you want to generate a 3D model with the following pair:

IMG_PHR1B_MS_003
IMG_PHR1B_MS_004
IMG_PHR1B_P_001
IMG_PHR1B_P_002

You should find in each folder the following data:

...
DIM_PHR1B_***.XML
IMG_PHR1B_***.TIF
RPC_PHR1B_***.XML

For each product, the user must provide the path to the pancromatic data (P.TIF) with its geomodel, all contained in the DIMAP file (DIMAP*P*.XML):

YAML JSON Copy

---
input:
  sensors:
    one:
      image: IMG_PHR1B_P_001/DIM_PHR1B_***.XML
    two:
      image: IMG_PHR1B_P_002/DIM_PHR1B_***.XML
  pairing:
  - - one
    - two

{
    "input": {
        "sensors": {
            "one": {
                "image": "IMG_PHR1B_P_001/DIM_PHR1B_***.XML"
            },
            "two": {
                "image": "IMG_PHR1B_P_002/DIM_PHR1B_***.XML"
            }
        },
        "pairing": [
            [
                "one",
                "two"
            ]
        ]
    }
}

If you want to add the colors, a P+XS fusion must be done, to specify a color.tif with the same shape and resolution than the Pancromatic data. It can be performed with otbcli_BundleToPerfectSensor as explained in Make a simple pan sharpening.

YAML JSON Copy

---
input:
  sensors:
    one:
      image: color_one.tif
    two:
      image: color_two.tif
  pairing:
  - - one
    - two

{
    "input": {
        "sensors": {
            "one": {
                "image": "color_one.tif"
            },
            "two": {
                "image": "color_two.tif"
            }
        },
        "pairing": [
            [
                "one",
                "two"
            ]
        ]
    }
}

The stereo matching will be performed using the first band. If you want to change that, you can use sensor loaders for inputs (cf. Input).

With a region of interest

There are two different ways to use a ROI in CARS:

• Crop input images: the whole pipeline will be done with cropped images

• Use input roi parameter: the whole images will be used to compute grid correction and terrain + epipolar a priori. Then the rest of the pipeline will use the given roi. This allows a better correction of epipolar rectification grids.

If you want to work with cropped image by using a region of interest for the whole pipeline, use cars-extractroi:

cars-extractroi -il DIM_PHR1B_***.XML -out ext_dir -bbx -58.5896 -34.4872 -58.5818 -34.4943

It generates a .tif and .geom to be used as:

YAML JSON Copy

---
input:
  sensors:
    one: ext_dir/ext_000.tif
    two : ext_dir/ext_001.tif

{
    "input": {
        "sensors": {
            "one": "ext_dir/ext_000.tif",
            "two": "ext_dir/ext_001.tif"
        }
    }
}

Then use the generated data as you would with raw data.

If you want to compute the grid correction and compute the epipolar/terrain a priori on the whole image, keep the same input images but specify the terrain ROI to use:

YAML JSON Copy

---
input:
  roi:
    type: FeatureCollection
    features:
    - type: Feature
      properties: {}
      geometry:
        coordinates:
        - - - 5.194
            - 44.2064
          - - 5.194
            - 44.2059
          - - 5.195
            - 44.2059
          - - 5.195
            - 44.2064
          - - 5.194
            - 44.2064
        type: Polygon

{
    "input": {
        "roi": {
            "type": "FeatureCollection",
            "features": [
                {
                    "type": "Feature",
                    "properties": {},
                    "geometry": {
                        "coordinates": [
                            [
                                [
                                    5.194,
                                    44.2064
                                ],
                                [
                                    5.194,
                                    44.2059
                                ],
                                [
                                    5.195,
                                    44.2059
                                ],
                                [
                                    5.195,
                                    44.2064
                                ],
                                [
                                    5.194,
                                    44.2064
                                ]
                            ]
                        ],
                        "type": "Polygon"
                    }
                }
            ]
        }
    }
}

See Usage Sensors Images Inputs configuration for more information.

Note

CARS also works with other types of data: SPOT 6-7, WorldView, etc.