Package 'GCSM'

Title: Implements Generic Composite Similarity Measure
Description: Provides implementation of the generic composite similarity measure (GCSM) described in Liu et al. (2020) <doi:10.1016/j.ecoinf.2020.101169>. The implementation is in C++ and uses 'RcppArmadillo'. Additionally, implementations of the structural similarity (SSIM) and the composite similarity measure based on means, standard deviations, and correlation coefficient (CMSC), are included.
Authors: Yadong Liu [aut, cre]
Maintainer: Yadong Liu <[email protected]>
License: MIT + file LICENSE
Version: 0.1.1.9000
Built: 2024-11-21 06:21:38 UTC
Source: https://github.com/liuyadong/gcsm

Help Index

Composite similarity between vectors

Description

Compute composite measures, GCSM or CMSC, between two vectors.

Usage

cmsc(
  x,
  y,
  rescale = FALSE,
  xmin = NA_real_,
  xmax = NA_real_,
  ymin = NA_real_,
  ymax = NA_real_,
  comp = "si"
)

cmsc_e1(
  x,
  y,
  rescale = FALSE,
  xmin = NA_real_,
  xmax = NA_real_,
  ymin = NA_real_,
  ymax = NA_real_,
  comp = "si"
)

cmsc_e2(
  x,
  y,
  rescale = FALSE,
  xmin = NA_real_,
  xmax = NA_real_,
  ymin = NA_real_,
  ymax = NA_real_,
  comp = "si"
)

gcsm(
  x,
  y,
  rescale = FALSE,
  xmin = NA_real_,
  xmax = NA_real_,
  ymin = NA_real_,
  ymax = NA_real_,
  comp = "si"
)

Arguments

x

A vector.
y

The other vector.
rescale

Rescale or not before computation.
xmin, xmax, ymin, ymax

Normalization parameters. If NA, are calculated from the ranges of x and y, respectively. See Details.
comp

Variable to return. If "si", the composite measure, if "s1","s2" or "s3", the corresponding component.

Details

These functions compute composite measures between vectors. Missing values are omitted. Normalization parameters are used to rescale x and y, and determine the global minimum (min) and maximum (max). If rescale is TRUE, x and y are rescaled to (x-xmin)/(xmax-xmin) and (y-ymin)/(ymax-ymin); and set min=0, max=1. If FALSE, min=min(xmin,ymin), max=max(xmax,ymax).

Value

A number.

Examples

x = runif(9)
gcsm(x, x)
cmsc(x, x)
# mean shift
gcsm(x, x - 0.2, xmin = 0, xmax = 1, ymin = 0, ymax = 1)
cmsc(x, x - 0.2, xmin = 0, xmax = 1, ymin = 0, ymax = 1)
gcsm(x, x + 0.2, xmin = 0, xmax = 1, ymin = 0, ymax = 1)
cmsc(x, x + 0.2, xmin = 0, xmax = 1, ymin = 0, ymax = 1)
## dissimilarity
y = 1 - x # y is the perfect antianalog of x
gcsm(y, x)
gcsm(y, x - 0.2, xmin = 0, xmax = 1, ymin = 0, ymax = 1)
gcsm(y, x + 0.2, xmin = 0, xmax = 1, ymin = 0, ymax = 1)

# random noise
noise = rnorm(9, mean = 0, sd = 0.2)
gcsm(x, x + noise, xmin = 0, xmax = 1, ymin = 0, ymax = 1)
cmsc(x, x + noise, xmin = 0, xmax = 1, ymin = 0, ymax = 1)
## dissimilarity
gcsm(y, x + noise, xmin = 0, xmax = 1, ymin = 0, ymax = 1)

Composite similarity on spatial windows

Description

Compute composite measures, GCSM, CMSC or SSIM, on spatial windows.

Usage

cmsc_sw(
  x,
  y,
  rescale = FALSE,
  xmin = NA_real_,
  xmax = NA_real_,
  ymin = NA_real_,
  ymax = NA_real_,
  ksize = 9,
  globe = FALSE,
  comp = "si"
)

cmsc_e1_sw(
  x,
  y,
  rescale = FALSE,
  xmin = NA_real_,
  xmax = NA_real_,
  ymin = NA_real_,
  ymax = NA_real_,
  ksize = 9,
  globe = FALSE,
  comp = "si"
)

cmsc_e2_sw(
  x,
  y,
  rescale = FALSE,
  xmin = NA_real_,
  xmax = NA_real_,
  ymin = NA_real_,
  ymax = NA_real_,
  ksize = 9,
  globe = FALSE,
  comp = "si"
)

gcsm_sw(
  x,
  y,
  rescale = FALSE,
  xmin = NA_real_,
  xmax = NA_real_,
  ymin = NA_real_,
  ymax = NA_real_,
  ksize = 9,
  globe = FALSE,
  comp = "si"
)

ssim_sw(
  x,
  y,
  rescale = FALSE,
  xmin = NA_real_,
  xmax = NA_real_,
  ymin = NA_real_,
  ymax = NA_real_,
  ksize = 11,
  sigma = 1.5,
  globe = FALSE,
  comp = "si"
)

Arguments

x

A matrix.
y

The other matrix.
rescale

Rescale or not before computation.
xmin, xmax, ymin, ymax

Normalization parameters. If NA, are calculated from the ranges of x and y, respectively. See Details.
ksize

Side length of spatial windows.
globe

Are data at the global scale? If TRUE, two vertical borders will be padded before computation.
comp

Variable to return. If "si", the composite measure, if "s1","s2" or "s3", the corresponding component.
sigma

Standard deviation of Gaussian weighting function depending on the distance between the cell and kernel center.

Details

These functions slide the spatial window over space. Missing values are omitted. Normalization parameters are used to rescale x and y, and determine the global minimum (min) and maximum (max). If rescale is TRUE, x and y are rescaled to (x-xmin)/(xmax-xmin) and (y-ymin)/(ymax-ymin); and set min=0, max=1. If FALSE, min=min(xmin,ymin), max=max(xmax,ymax). OpenMP is used for parallel computing.

Value

A matrix.

Examples

x = matrix(runif(36), nrow = 6, ncol = 6)

gcsm_sw(x, x + 0.2, xmin = 0, xmax = 1, ymin = 0, ymax = 1, ksize = 3)
cmsc_sw(x, x + 0.2, xmin = 0, xmax = 1, ymin = 0, ymax = 1, ksize = 3)
ssim_sw(x, x + 0.2, xmin = 0, xmax = 1, ymin = 0, ymax = 1, ksize = 3)

Composite similarity on temporal windows

Description

Compute composite measures, GCSM or CMSC, on temporal windows.

Usage

cmsc_tw(
  xxx,
  yyy,
  rescale = FALSE,
  xmin = NA_real_,
  xmax = NA_real_,
  ymin = NA_real_,
  ymax = NA_real_,
  comp = "si"
)

cmsc_e1_tw(
  xxx,
  yyy,
  rescale = FALSE,
  xmin = NA_real_,
  xmax = NA_real_,
  ymin = NA_real_,
  ymax = NA_real_,
  comp = "si"
)

cmsc_e2_tw(
  xxx,
  yyy,
  rescale = FALSE,
  xmin = NA_real_,
  xmax = NA_real_,
  ymin = NA_real_,
  ymax = NA_real_,
  comp = "si"
)

gcsm_tw(
  xxx,
  yyy,
  rescale = FALSE,
  xmin = NA_real_,
  xmax = NA_real_,
  ymin = NA_real_,
  ymax = NA_real_,
  comp = "si"
)

Arguments

xxx

A 3-d array with the 3rd dimension representing time.
yyy

The other 3-d array.
rescale

Rescale or not before computation.
xmin, xmax, ymin, ymax

Normalization parameters. If NA, are calculated from the ranges of xxx and yyy, respectively. See Details.
comp

Variable to return. If "si", the composite measure, if "s1","s2" or "s3", the corresponding component.

Details

These functions slide the temporal window over space. Missing values are omitted. Normalization parameters are used to rescale xxx and yyy, and determine the global minimum (min) and maximum (max). If rescale is TRUE, xxx and yyy are rescaled to (xxx-xmin)/(xmax-xmin) and (yyy-ymin)/(ymax-ymin); and set min=0, max=1. If FALSE, min=min(xmin,ymin), max=max(xmax,ymax). OpenMP is used for parallel computing.

Value

A matrix.

Examples

x = array(runif(81), dim = c(3, 3, 9))

gcsm_tw(x, x + 0.2, xmin = 0, xmax = 1, ymin = 0, ymax = 1)
cmsc_tw(x, x + 0.2, xmin = 0, xmax = 1, ymin = 0, ymax = 1)