segmentRaster: segmentRaster

In RRemelgado/CAWaR: CAWa Project Tools

Description

Segmentation of a raster object using k-means clustering and connected component labeling.

Usage

segmentRaster(x, change.threshold = 0, n.size = NULL)

Arguments

x	Object of class RasterLayer.
change.threshold	Numeric element (0 - 100).
n.size	Numeric element.

Details

The function clusters a raster object with k-means through unsuperClass and segments the output using ccLabel. The number of samples used to determine the number of clusters can be defined through n.size to reduce the required computational time. The optimal number of clusters is determined with the elbow method. he elbow method looks at the percentage of variance explained by the number of clusters. If adding a new cluster, referred here as k, leads to no improvement, the function will use k-1 to derive. This breakpoint is determined by change.threshold which controls the percent change between the amount of variance explained by two consequent k values. the output is a list consisting of:

• class - Cluster image.

• regions - Segmented region image.

Value

A list.

RRemelgado/CAWaR documentation built on June 7, 2020, 10:27 p.m.