R/interpolation.R

In MangoTheCat/visualTest: Perform Fuzzy Image Matching

## http://stackoverflow.com/questions/26142288

bilinearInterpolation <- function(image, out_dims = c(64, 64)) {

  in_rows <- nrow(image)
  in_cols <- ncol(image)
  out_rows <- out_dims[1]
  out_cols <- out_dims[2]

  S_R <- in_rows / out_rows
  S_C <- in_cols / out_cols

  cf <- outer(1:out_rows * 0, 1:out_cols, FUN = "+") * S_C
  rf <- outer(1:out_rows, 1:out_cols * 0, FUN = "+") * S_R

  r <- floor(rf)
  c <- floor(cf)

  r[r < 1] <- 1
  c[c < 1] <- 1
  r[r > in_rows - 1] <- in_rows - 1
  c[c > in_cols - 1] <- in_cols - 1

  delta_R <- rf - r
  delta_C <- cf - c

  in1_ind <- as.vector((c - 1) * in_rows + r)
  in2_ind <- as.vector((c - 1) * in_rows + r + 1)
  in3_ind <- as.vector(c * in_rows + r)
  in4_ind <- as.vector(c * in_rows + r + 1)

  out <- array(0, dim = c(out_rows, out_cols, dim(image)[3]))
  class(out) <- class(image)

  for (idx in 1:dim(image)[3]) {
    chan <- image[, , idx]
    out[, , idx] <-
      chan[in1_ind] * (1 - delta_R) * (1 - delta_C) +
      chan[in2_ind] * (    delta_R) * (1 - delta_C) +
      chan[in3_ind] * (1 - delta_R) * (    delta_C) +
      chan[in4_ind] * (    delta_R) * (    delta_C)
  }

  out
}