First-order texture

Revision as of 17:43, 18 November 2018

Local statistical moments (Mean, Variance, Skewness, Kurtosis) calculated on every pixel in the selected channel of the input image, over a specified neighborhood are called first-order textures.

• In the search engine of the Processing Toolbox, type Local Statistic and select LocalStatisticExtraction under Feature Extraction of OTB.
• Under the Parameters tab, select a single band or a multiband file as input layer.
• In case of a multiband file select the band number.
• Select 3 as Neighborhood radius in pixels.

The output image (Fig. B) is a multiband with 4 statistical moments per band in the order:

1. Mean
2. Variance
3. Skewness
4. Kurtosis

The standard deviation of gray levels in a region in the neighborhood of a pixel can be calculated by a GRASS GIS modul:

• In the search engine of the Processing Toolbox, type neighbors and select r.neighbors under Raster of GRASS GIS 7 commands.
• Under the Parameters tab, select a single band file as input layer.
• Select stddev from the drop-down list Neighborhood operation.
• Select an odd integer number as Neighborhood size in pixels.
• Tick Use circular neighborhood.

Figure A: Input image: Sentinel-2 synthesis image based on 4 10m bands, (University Göttingen Campus North)

Figure B: Output image: First-order textures of Sentinel-2 synthesis image RGB=Skewness, Variance, Mean (University Göttingen Campus North)

Figure C: Output image: Standard deviation texture of Sentinel-2 synthesis image (University Göttingen Campus North)