msssim: Multiscale structural similarity index
In colinr23/spatialcompare: Spatial Pattern Comparison Metrics

Description Usage Arguments Details Value See Also

This function implements the multiscale structural similarity index of Wang et al. (2003) formulated for comparison of raster spatial objects at multiple scales.

1
2
3

msssim(img1, img2, w, gFIL = TRUE, edge = FALSE, ks = c(0.01, 0.03),
  level = 5, weight = c(0.0448, 0.2856, 0.3001, 0.2363, 0.1333),
  method = "product")

`img1`	is a `raster` object to compare
`img2`	is a `raster` object to compare
`w`	is the width of the neighbourhood in number of pixels out from centre cell
`edge`	is a binary flag for whether a torroidal edge correction should be applied
`ks`	is a vector of length 2 which contains values for constants in the SSIM formula. If ignored default values will be used.
`level`	is the number of scales to evaulate statistic at, default is 5
`weight`	is the weights for scales in `level`, and should be a vector of length `level`. Defaults are optimal values determined empirically in Wang et al. (2003)
`method`	is the method for combining SSIM components, currently the only option
`GFil`	is a binary flag for whether a Gaussian filter should be applied as a smoothing function

The msssim function can be used to generate the global multiscale ssim. Local values are not supported. The weights used in the default are those given in Wang et al. (2003) and were determined empirically for a five-level application. The weights determine how cross-scale values of the SSIM are weighted.

msssim returns a vector with the value of the multiscale structural similarity index

ssim

colinr23/spatialcompare documentation built on May 13, 2019, 9:54 p.m.