Remote Sensing
The Remote Sensing toolbox works on imagery: satellite scenes, aerial photographs, drone orthophotos, radar acquisitions. Its 152 tools cover the whole chain, from correcting the raw radiometry to producing a land cover map and comparing two dates.
It is the largest toolbox in the panel, and the one where knowing which sub toolbox you need saves the most time. Everything runs in your browser, on rasters loaded through Add Data.
The sub toolboxes
Section titled “The sub toolboxes”| Sub toolbox | What it is for | Tools |
|---|---|---|
| Filters | Smooth, sharpen, denoise, measure texture. The general purpose image operations. | 46 |
| OBIA | Object based analysis: segment the image into regions, describe them, classify them. | 33 |
| Enhancement & Contrast | Make an image readable: stretches, composites, mosaics, pan sharpening. | 21 |
| Classification | Turn pixels into classes, supervised or unsupervised. | 17 |
| SAR | Radar imagery: speckle filtering and polarimetric decompositions. | 10 |
| Edge & Feature Detection | Find boundaries and linear features. | 7 |
| Change Detection | Compare two dates and map what changed. | 5 |
| Radiometric Correction | Turn sensor values into physical quantities, remove haze and terrain effects. | 5 |
| Spectral Analytics | Work with spectra: unmixing, spectral angles, library matching. | 5 |
| Thermal & Emissivity | Retrieve land surface temperature from thermal bands. | 3 |
The usual order of work
Section titled “The usual order of work”Not every study needs every step, but the order rarely changes.
- Correct. Convert digital numbers to reflectance, remove atmospheric haze, and compensate for terrain if your area is mountainous. See Radiometric Correction. Skip this only when your product is already delivered as surface reflectance.
- Prepare. Mosaic the tiles, sharpen with the panchromatic band, denoise. See Enhancement & Contrast and Filters.
- Analyse. Either pixel by pixel, in Classification, or object by object, in OBIA.
- Compare. With two dates processed the same way, map the change. See Change Detection.
Pixels or objects
Section titled “Pixels or objects”The one structural choice of this toolbox.
Pixel based classification assigns a class to every pixel independently, from its spectral values. Fast, well understood, and the right approach for medium resolution imagery, where a pixel covers 10 or 30 metres and averages a whole scene element.
Object based analysis, OBIA, first groups pixels into homogeneous regions, then classifies those regions using their spectra, their shape, their texture and their neighbours. It is the right approach for high resolution imagery, where a single roof spans hundreds of pixels and a pixel based map comes out speckled.
The rule of thumb: if your objects of interest are much larger than your pixels, use OBIA.
Bands and stacks. Most tools here expect a multiband raster, one band per spectral channel. Load the scene as a single multiband file rather than as separate images whenever you can.
Size. The computation happens in your browser’s memory, so clip a full satellite scene to your area of interest before running a classification over it.
Credits
Section titled “Credits”These tools come from WhiteboxTools by Prof. John Lindsay, University of Guelph, and from its Next Generation port.