areal_edge_detection: Edge detection for areal data
In jlegewie/BoundaryDetection: Areal boundary detection in R

Description Usage Arguments Value Examples

areal_edge_detection ...

areal_edge_detection(shp, vars, scale_smoothing = 0, scale_sobel = 0,
  domain = "neighbors", natural_boundaries = NULL, N = NULL, C = NULL,
  method = "integration", smoothing = "gaussian", combine = NULL,
  res = 50, .progress = "none")

`shp`	`SpatialPolygonsDataFrame` object (sf objects as converted).
`vars`	Vector of variable names.
`scale_smoothing`	Scale parameter for gaussian smoothing to pre-process data. `scale_smoothing` corresponds to the standard deviation of the normal distribution. Note that the scale depends on the projection of the shapefile. 0 (the default) uses no smoothing.
`scale_sobel`	Scale parameter for the edge detection algorithm. The parameter corresponds to the standard deviation of the first derivative of the multivariate normal distribution (the edge detection filter). Note that the scale depends on the projection of the shapefile. 0 (the default) automatically adapts to cover the adjacent areas.
`domain`	Domain for edge detection algorithms: `neighbors` or `unbounded`. `neighbors` just uses neighboring areas. `unbounded` uses all areas within 3 times the standard deviation. In theory, the normal distribution is non-zero everywhere but in practice it is close to zero more than three standard deviations from the mean.
`natural_boundaries`	natural_boundaries
`N`	Adjacency matrix as neighbours list based on `poly2nb` (optional, saves time for multiple calls).
`C`	SpatialPoints object with coordinates of cendroids (optional, saves time for multiple calls).
`method`	Centroid or numerical integration.
`smoothing`	Smoothing method. 'gaussian' and 'bilateral' are supported. 'bilateral' requires two scale parameter.
`combine`	Function to combine edges based on different variables. `NULL` (the default) returns an array with the edges for each variables.
`res`	Resolution for the numerical integration. The number of integration points is `res * res`, which is directly related to speed and precision.
`.progress`	Show process bar.

Numerical vector or array with edge intensity variable.

1
2
3

areal_edge_detection(...)
areal_edge_detection(..., combine = function(x) max(x, na.rm = TRUE))
areal_edge_detection(..., combine = function(x) prod(tail(sort(x), 2)))