fix: correct import

feat: add input

scripts/image.R

@@ -6,7 +6,10 @@
 import::here("dplyr", c("mutate", "sample_n"), .character_only = TRUE)
 import::here(
   "imager",
-  c("grayscale", "HSVtoRGB", "imappend", "imsplit", "map_il", "RGBtoHSV"),
+  c(
+    "as.cimg", "grayscale", "HSVtoRGB", "imappend", "imsplit",
+    "map_il", "RGBtoHSV"
+  ),
   .character_only = TRUE
 )
 import::here("magrittr", "%>%", .character_only = TRUE)
@@ -126,10 +129,11 @@ invert_grayscale <- function(img) {
 #'
 #' @param img an imager::cimg
 #' @param quantile a real, the quantile level used for thresholding.
+#' @param G an integer, the number of mixture components.
 #' @param ... mclust::densityMclust parameters
 #'
 #' @returns an imager::cimg
-binarize <- function(img, quantile = 0.95, ...) {
+binarize <- function(img, quantile = 0.95, G = 1L, ...) {
   # convert to grayscale if needed and invert
   if (rev(dim(img))[1L] > 1L) {
     stop("A grayscale image is expected.")
@@ -137,7 +141,7 @@ binarize <- function(img, quantile = 0.95, ...) {
   # sample 
   sample <- sample_histogram(img)
   # fit Gaussian mixture
-  gm <- densityMclust(sample, G = 1L, plot = FALSE, ...)
+  gm <- densityMclust(sample, G = G, plot = FALSE, ...)
   # threshold based on 95% quantile
   threshold <- qnorm(
     quantile, gm$parameters$mean[1L], sqrt(gm$parameters$variance$sigmasq[1L])