Skip to content

Getting started \ Preprocessing

Preprocessing

Preprocessing aka How to get ARD dataset ?

Many people use directly Analysis Ready Data (ARD), often referred as L2A images for optical context. But sometimes you have to preprocess data (raw satellite image or products, level 0/1) to work properly on it.

Several steps are required to get ready for analysis data. The objective is for example to correct atmospheric aberrations or project images on ground with precise localization. Regarding optical image preprocessing, OTB recipes gives an overview of main preprocessing steps.

For an introduction to SAR image preprocessing, you can browse Awesome SAR.

CESBIO Multitemp blog also presents quite clearly Product Level names

Generic processing

Homologous points extraction

Extraction of homologous points. Output = list of homologous points between several images

Pollux - CNES proprietary

otbcli_HomologousPointsExtraction

Refinement

LibGeo - CNES. Licence AD. Interacts with the image geometric model, and is used to build sampling grids. Tools to refine the geometrical models thanks to input homologous points. It contains also tools to manage sampling grids.

Euclidium - CNES proprietary. Interacts with the image geometric model, and is used to build sampling grids. Tools to refine the geometrical models thanks to input homologous points. It contains also tools to manage sampling grids.

Registration

Medicis - Computation of a disparity grid betwween two images. It may be used to precisely superimpose two images.

Sampling grid generation

LibGeo - CNES. Licence AD. Generates the sampling grid thanks to the geometrical model of the image.

Euclidium - CNES proprietary Generates the sampling grid thanks to the geometrical model of the image.

Resampling

Orion - CNES Proprietary. Used to sample an image with a sampling grid as input.

Mario - CNES Proprietary. Used to sample an image with a sampling grid as input.

otbSuperimpose - Combines methods in order to modify the polynomial geometrical model. This model is then used to reproject / resample the image in the chosen geometry. Output images should be perfectly superimposed.

Re-projection, ortho-rectification

otbOrthorectification

Optical images

Optical calibration

Optical calibration consists in converting raw satellite images in reflectance (top of atmosphere, bottom of atmosphere aka top of canopy). First step needs to know satellite orbit and altitude, sun elevation (to compute solar time) while second steps involves atmospheric models.

otbRadiometricCorrection

otbOpticalCalibration

SMAC (TBC)

Fusion, pan-sharpening

Pansharpening is a process of merging high-resolution panchromatic and lower resolution multispectral imagery to create a single high-resolution color image.

otbPanSharpening

otbBundleToPerfectsensor

Infrared (IR) images

Temperature/Emissivity maps Calculation (after atmospheric corrections) : DTN/TPI/MO Tools ?

SAR Preprocessing

/!\ For SAR data in NITF standard format, the use of specific tools is advised

SAR Deburst

SAR Amplitude

SNAP

otb

Time series

ARDEO