Home /Processing/Processes

Pansharpening

A fusion algorithm that upsamples the image to the resolution of the panchromatic band.

Overview

The process fuses the higher-resolution panchromatic band with lower-resolution multispectral bands of a data item from an optical collection, using the Brovey method. The result is a GeoTIFF file with an upsampled spatial resolution.

Most often, optical collections produce the following bands:

  • A panchromatic band with a high spatial resolution but a low spectral resolution
  • Several multispectral bands with a low spatial resolution but a high spectral resolution

The fusion that pansharpening provides allows you to benefit from the complementary qualities of panchromatic and multispectral bands. The result is a pansharpened reflectance product.

An input Pléiades image of Berlin (Germany)

A pansharpened output image

Download sample data

Requirements for input imagery

The data item must be CNAM-compatible. CNAM-compatible data meets both criteria:

The data item must come from an optical collection.

The data item must have panchromatic and multispectral bands. To check whether the chosen collection provides the necessary bands, see Spectral bands.

Input parameters

You must specify the data items you want to apply the process to.

You must specify the title of the output data item.

You can specify the weights for multispectral bands or use automatically optimized weights by not defining them. Depending on the input data item, the algorithm either uses sensor-optimized weights or, if this data isn’t available, generates new optimal weights.

The weighted Brovey method, applied for this algorithm, uses the following values:

  • The pseudo panchromatic intensity. It’s derived from a weighted average of the multispectral bands. Each multispectral band may contribute differently to this average, and the weights reflect their importance in approximating the panchromatic data.
  • The real panchromatic intensity. It’s the actual panchromatic intensity data that serves as a reference value.

The output value of a multispectral band is computed as follows:

Band output value=Band input value×Real panchromatic intensityPseudo panchromatic intensity\text{Band output value} = \text{Band input value} \times \frac{\text{Real panchromatic intensity}}{\text{Pseudo panchromatic intensity}}

API input

Use the pansharpening name ID for the processing API.

JSON
{
  "inputs": {
    "title": "Pansharpened SPOT imagery over Germany",
    "item": "https://api.up42.com/v2/assets/stac/collections/21c0b14e-3434-4675-98d1-f225507ded99/items/23e4567-e89b-12d3-a456-426614174000",
    "greyWeights": [
      {
        "band": "blue",
        "weight": 0.2
      },
      {
        "band": "green",
        "weight": 0.34
      },
      {
        "band": "red",
        "weight": 0.23
      }
    ]
  }
}
ParameterOverview
inputs.title

object | required

The title of the output data item.
inputs.item

object | required

A link to the data item in the following format: https://api.up42.com/v2/assets/stac/collections/{collection-id}/items/{item-id}.
inputs.greyWeights

array of objects

The weight factors by which spatial details of multispectral bands are scaled.

  • If not specified, either sensor-optimized weights or optimal generated weights are used.
  • If you specify the weights yourself, you must define a minimum of 3 bands.
inputs.greyWeights.band

string | required if

Required if greyWeights is used.

The name of the band from the STAC asset with the ["data", "multispectral"] roles.
inputs.greyWeights.weight

float | required if

Required if greyWeights is used.

The multiplication value that lets you modulate the influence of multispectral bands on the final image. The range of allowed values spans from -1 to 1.

Learn more

Last updated: