R/Conversion.R

In ClairBee/IO.Pixels: Functions for analysis of distribution of dead pixels (part of Inside Out project).

#' Convert points to raster
#' 
#' Support function: convert matrix of coordinates into a raster, so that convolution, clumping etc can be applied
#' @param px data.frame or matrix, the first two columns of which contain the X and Y coordinates of points to be marked on the raster
#' @param im.dim Vector of dimensions for the final raster. Default is c(2048, 2048).
#' @return Raster object with the coordinates marked as 1, everything else as NA.
#' @export
#' 
px2r <- function(px, im.dim = c(2048, 2048)) {
    
    mat <- array(0, dim = im.dim)
    mat[as.matrix(px[,1:2])] <- 1
    
    # rotate matrix by reversing and transposing, then convert to raster
    mat <- t(mat)[im.dim[2]:1,]
    raster(mat, xmn = 0.5, xmx = im.dim[1] + .5, ymn = 0.5, ymx = im.dim[2] + .5)
}


#' Convert matrix to raster
#' 
#' Convert an image array to a raster file, maintaining the correct orientation for plotting
#' @param im Single-layer image array to be converted to a raster
#' @return RasterLayer object containing the image values
#' @export
#' 
m2r <- function(im) {
    # rotate matrix by reversing and transposing, then convert to raster
    mat <- t(im)[dim(im)[2]:1,]
    raster(mat, xmn = 0.5, xmx = dim(im)[1] + .5, ymn = 0.5, ymx = dim(im)[2] + .5)
}


#' Convert raster to matrix
#' 
#' Convert a raster to an image array, maintaining the correct orientation for plotting
#' @param rast Single-layer RasterLayer object to be converted to a 2d array
#' @return Single-layer image array
#' @export
#' 
r2m <- function(rast) {
    mat <- matrix(getValues(rast), ncol = dim(rast)[2], byrow = T)
    t(mat[dim(mat)[1]:1,,drop = F])
}


#' Convert pixel coordinates to PPP object
#' 
#' Convert coordinates from bad pixel map to a PPP object for spatial analysis. The resulting image can be further cropped if necessary.
#' @param px data.frame or matrix, the first two columns of which contain the X and Y coordinates of points to be marked on the raster
#' @param im.dim Vector of dimensions for the final raster. Default is c(2048, 2048).
#' @param crop List of two vectors of crop boundaries. Default is list(c(1, im.dim[1]), c(1, im.dim[2])), which crops to original image area.
#' @return ppp object of the required dimensions
#' @export
#' 
px2ppp <- function(px, im.dim = c(2048, 2048), crop = list(c(1, im.dim[1]), c(1, im.dim[2]))) {
    
    # crop image if necessary
    tmp <- array(0, dim = im.dim)
    tmp[as.matrix(px[,1:2])] <- 1
    tmp[c(1:crop[[1]][1], crop[[1]][2]:im.dim[1]),] <- 0    # crop left & right edges
    tmp[,c(1:crop[[2]][1], crop[[2]][2]:im.dim[2])] <- 0    # crop top & bottom edges
    px <- which(tmp == 1, arr.ind = T)
    
    ppp(px[,1], px[,2], crop[[1]], crop[[2]])
}