R/rect2sqMatrix.R

In shoebodh/radialpsd: Calculates Radially Averaged Power Spectrum (periodograms) of a 2D matrix

#' Convert a rectangular matrix object to a square on by addign rows/columns
#'
#' @param x input matrix
#' @param pad.values values to be applied to the new rows/columns. devault = NaN
#'
#' @export
#' @examples
#'
#' x = matrix(0, 6, 3)
#' make.square(x)

rect2sqMatrix = function(x, pad = NULL, trim = NULL, pad.values = NaN, ...) {

 stopifnot(class(x) =="matrix")

 N = nrow(x)
 M = ncol(x)

 pad.matrix = function(x, pad.values = pad.values, ...){
       N = nrow(x)
       M = ncol(x)

       dim_diff = abs(N - M)

       if(N > M) {
             if(dim_diff %% 2 == 0) {
                   pad_matrix = matrix(pad.values, nrow = N, ncol = dim_diff/2)
                   pad_x= cbind(pad_matrix, x, pad_matrix)
             } else {

                   pad_matrix1 = matrix(pad.values, nrow = N, ncol = floor(dim_diff/2))
                   pad_matrix2 = matrix(pad.values, nrow = N, ncol = floor(dim_diff/2)+1)
                   pad_x = cbind(pad_matrix1, x, pad_matrix2)
             }

       } else if(N < M) {
             if(dim_diff %% 2 == 0) {
                   pad_matrix = matrix(pad.values, nrow = dim_diff/2, ncol = M)
                   pad_x = rbind(pad_matrix,x, pad_matrix)
             } else {
                   pad_matrix1 = matrix(pad.values, nrow = floor(dim_diff/2), ncol = M)
                   pad_matrix2 = matrix(pad.values, nrow = floor(dim_diff/2 + 1), ncol = M)
                   pad_x = rbind(pad_matrix1, x, pad_matrix2)

             }

       }

   return(pad_x)

 }

 trim.matrix = function(x) {

       N = nrow(x)
       M = ncol(x)

       dim_diff = abs(N - M)

       if(N > M) {
             if(dim_diff %% 2 == 0) {

                   trup = 1: (dim_diff/2)
                   trdn = (N-dim_diff/2 + 1):N

                    } else {
                          trup = 1:floor(dim_diff/2)
                          trdn = (N-floor(dim_diff/2)):N

                    }
             trimmed_x = x[-c(trup, trdn), ]

       } else if(N < M) {
             if(dim_diff %% 2 == 0) {
                  trlt = 1:(dim_diff/2)
                  trrt = (M-dim_diff/2 + 1):M

                                } else {
                                      trlt = 1:(floor(dim_diff/2))
                                      trrt = (M-floor(dim_diff/2)):M

             }
            trimmed_x = x[, -c(trlt, trrt)]
       }
   return(trimmed_x)
 }

 if(is.null(trim)){
  mat_out = pad.matrix(x)

 } else {
      mat_out =  trim.matrix(x)
 }

return(mat_out)
}