Enhancement & Contrast

Enhancement & Contrast covers everything that makes an image look right and fit together: stretching the contrast so the detail becomes visible, building colour composites from separate bands, merging tiles into a seamless mosaic, and sharpening with a panchromatic band.

These operations change how the image reads, and some of them change its values, so run analysis on the original and enhancement on a copy.

Start here

The contrast stretches

Satellite imagery almost never uses its full value range. A raw scene displayed as is looks flat and dark. Stretching redistributes the values so the detail becomes visible.

Tool	When to use it
Percentage Contrast Stretch	The everyday choice. Clips a small percentage at each end, typically 2 percent, and stretches the rest. Handles the few extreme pixels that would otherwise flatten everything.
Min Max Contrast Stretch	Stretches between bounds you set. Use it when you need two images to be comparable, since the same bounds give the same rendering.
Standard Deviation Contrast Stretch	Stretches between the mean plus or minus a number of standard deviations. Well behaved on normally distributed data.
Histogram Equalization	Redistributes values so every level is equally represented. Reveals the most detail and distorts the relationship between brightness and value the most.
Gaussian Contrast Stretch	Matches the histogram to a Gaussian shape.
Sigmoidal Contrast Stretch	An S shaped curve, gaining contrast in the middle tones and compressing the extremes.
Piecewise Contrast Stretch	A curve through breakpoints you define, for full control.
Gamma Correction	Power law adjustment of the brightness. The gentlest correction for an image that is simply too dark or too bright.

Create Colour Composite

Combines three single band rasters into one RGB image. This is how you build a true colour view from the red, green and blue bands, or a false colour infrared view by putting the near infrared band in the red channel, which is the classic way of reading vegetation.

Split Colour Composite does the reverse, which you need when you receive an RGB image and want to work band by band.

Panchromatic Sharpening

Fuses a high resolution panchromatic band with lower resolution colour bands, giving an image with the geometry of the first and the colours of the second. This is what turns a 10 metre multispectral scene into a 2.5 metre colour image on satellites that carry both.

Mosaic and Mosaic With Feathering

Mosaic assembles several overlapping rasters into one. Mosaic With Feathering blends the overlaps with distance weights, so the seams disappear instead of showing as straight lines across the image. Prefer the second whenever the scenes differ slightly in radiometry, which is almost always.

Normalized Difference Index

Computes the generic normalised difference between two bands of a multiband raster. This one tool gives you the whole family of spectral indices, depending on which bands you point it at.

Index	Bands	What it shows
NDVI	near infrared and red	Vegetation vigour. The most used index in remote sensing.
NDWI	green and near infrared	Open water.
NDBI	shortwave infrared and near infrared	Built up areas.
NBR	near infrared and shortwave infrared	Burnt areas.

Threshold the result with Otsu Thresholding, in Classification, to turn an index into a mask.

All the tools

Tool	What it does
Percentage Contrast Stretch	Linear stretch after clipping a percentage at each end.
Min Max Contrast Stretch	Linear stretch between bounds you set.
Standard Deviation Contrast Stretch	Linear stretch around the mean, in standard deviations.
Gaussian Contrast Stretch	Matches the histogram to a Gaussian distribution.
Sigmoidal Contrast Stretch	S shaped stretch controlled by gain and cutoff.
Piecewise Contrast Stretch	Stretch through breakpoints you define.
Histogram Equalization	Equalises the histogram, maximising visible detail.
Histogram Matching	Matches the histogram to a reference distribution.
Histogram Matching Two Images	Matches one image’s histogram to another’s. Essential before comparing two dates.
Gamma Correction	Power law brightness adjustment.
Balance Contrast Enhancement	Removes colour bias in an RGB composite with per channel stretches.
Direct Decorrelation Stretch	Increases colour saturation by reducing the achromatic component.
Create Colour Composite	Builds an RGB composite from three bands, with optional opacity.
Split Colour Composite	Splits an RGB composite back into three single band rasters.
Rgb To Ihs	Converts red, green, blue into intensity, hue, saturation.
Ihs To Rgb	The inverse conversion.
Panchromatic Sharpening	Fuses a panchromatic band with multispectral bands.
Mosaic	Assembles several rasters into one.
Mosaic With Feathering	The same, blending the overlaps so seams disappear.
Resample	Changes the pixel size, with a choice of interpolation method.
Normalized Difference Index	The normalised difference between two bands, giving NDVI and its family.

Notes

Stretch for display, analyse the original. A stretch changes pixel values. Any index, classification or change detection must run on the original data, or on a radiometrically corrected version of it, never on a stretched one.

Before comparing two dates, match their histograms. Two scenes acquired months apart differ in illumination and atmosphere, and a change detection run on unmatched images mostly maps the weather.

Resampling method matters. Nearest neighbour keeps the values intact, which is mandatory for a classified raster. Bilinear and cubic give a smoother result on continuous imagery and invent intermediate values.