This tutorial will guide you through the processing of radar images available on HEDAVI, covering tools dedicated to beginners or others suited for expert users.

The tutorial follows the two previous ones dedicated to finding products and organising a layerstack in the HEritage DAta VIsualisation (HEDAVI) Explorer, and it is conceived as an extension of the more extensive tutorial focused on the processing of optical images.

Similarly to optical images, when you search for and prepare a radar product in the HEDAVI Explorer it will be Processed On-the-Fly (POF). During preparation, products are subdivided in a pyramid of tiles to ensure fast on-demand processing and display. While the processing options are largely similar to those of optical images, this tutorial will highlight a few specificities of radar images.

1 Default style

When you select a radar product in the list of results, it is shown in the default style associated to the dataset.

2 Predefined styles

You can apply other predefined styles to the selected product in the same way you would do for an optical product. The predefined styles shown depend on the bands available. In addition to a different selection of bands, they may also include variations in calibration settings.

3 Custom colour composite

You can select a single channel (in grey levels or using a look-up table) or three channels (Red Green Blue) rendering by selecting any polarisation of your product or building a Normalised Difference Index (NDI) that uses these polarisations.

4 Advanced styles

Beyond colour composites, you can create your own custom style as described previously for optical sensors.

4.1 Pre-processing

Unlike optical sensors, you can customise pre-processing parameters as well. They include the choice of the Digital Elevation Model (DEM) used for the orthorectification, the type of calibration, the possibility to use a linear or logarithmic (decibel) scale if backscatter coefficients are used, options to clip the image etc. If you use a single channel rendering (grey levels or using a look-up table), a single button allows specifying the pre-processing parameters. In the case of a RGB rendering with three input channels, three buttons are available but they will use the same pre-processing parameters by default. Untick the “Same pre-processing” box if you prefer to select the parameters for each band individually.

4.1.1 Orthorectification

While optical images are subject to the parallax effect, radar images require correcting the range inversion that makes high ground seem closer to the sensor than it actually is (in contrast to further away with the parallax effect). In the pre-processing window, you can select which source of topography to use for the orthorectification.

In most cases, the default setting (i.e. orthorectification from COP-DEM GLO-30) will provide the best results.

4.1.2 Thermal denoising

The thermal noise correction removes the thermal noise that is particularly visible at near range. It is applied by default but can be removed by clicking the “No” button.

4.1.3 Calibration

Backscatter coefficient

Instead of using the native Digital Number of a product, it is possible to use a backscatter coefficient. The sigma-nought (σ⁰), gamma-nought (γ⁰) and beta-nought (β⁰) coefficients are available. You can also apply the gamma-nought terrain-corrected coefficient that flattens the relief and makes the result more even and mostly dependent on the land cover.

Conversion to decibel and stretching

If you have selected to use a backscatter coefficient, you can then choose to convert the result in decibels (dB). Using the linear scale usually results in a darker image with very bright spots while the dB scale provides less skewed and more even images, with a more compressed dynamic.

In order to use a backscatter coefficient, you should apply a “Linear stretching” in the histogram transform step. Then select the intervals, using not the Digital Numbers but a [0, 1] ratio or its decimal logarithm in the case of dB.

Radar Cross Section Equalisation

Finally, the “Radar Cross Section Equalisation” (RCSE) process homogenises the swath and sub-swath on seas and oceans. Dedicated to marine and coastal users, it sets the mean and standard deviation on water according to the specified parameters. The default parameters highlight the contrast on the usually dark water areas.

4.1.4 Borders clipping

Some images, in particular older ones, can include some background or some “scale effect” at the edges of a segment acquisition. Using the “Clip borders” process, you can remove such areas located at near/far range and/or at the start/end of a segment acquisition.

5 Expert style edition - Processing On-the-Fly Macro-Language

Finally, it is also possible to create your own Processing On-the-Fly without the complete user interface but using the POF-ML instead. More resources are referenced in the previous tutorial.

Below are a few examples:

Conclusion

HEDAVI Explorer is even more powerful to process radar images than it is with optical images. Following our tutorials, with a few clicks you will be able to display an image in a set of predefined styles or to master a customised processing chain to suit your need. With HEDAVI you can select and process your layers, organise them in a layerstack, save and share your work with hyperlooks, build a sequence of hyperlooks to provide context and develop a narrative in the form of stories. A future tutorial will deal with this last step.