Skip to contents

Convert pixel-coordinate rectangle bounds to 0-1 image fractions

Source: R/shift_mask.R
region_bounds_from_pixels.Rd

Bridge between visual inspection of a base image (where people naturally think in pixel rows and columns) and the region_bounds argument of make_face_mask(), which expects 0-1 image fractions. Returns a length-4 numeric vector ready to pass straight to make_face_mask(region_bounds = ...).

Usage

region_bounds_from_pixels(row_min, row_max, col_min, col_max, img_dims)

Arguments

row_min, row_max, col_min, col_max

Integer pixel coordinates of the rectangle edges (1-indexed, with row 1 at the top of the image and col 1 at the left). Must satisfy row_min <= row_max and col_min <= col_max, and all four values must lie inside the image.

img_dims

Integer c(nrow, ncol) of the target image (or a single integer for a square image).

Value

Numeric length-4 vector c(row_min, row_max, col_min, col_max) in 0-1 image fractions.

See also

make_face_mask().

Examples

# On a 256x256 base, "rows 110-140, cols 60-115" becomes:
region_bounds_from_pixels(
  row_min = 110, row_max = 140,
  col_min = 60,  col_max = 115,
  img_dims = c(256L, 256L)
)
#> [1] 0.4274510 0.5450980 0.2313725 0.4470588

# Pass straight to make_face_mask():
bounds <- region_bounds_from_pixels(110, 140, 60, 115,
                                    img_dims = 256L)
m <- make_face_mask(c(256L, 256L), region = "left_eye",
                    region_bounds = bounds)
mean(m)
#> [1] 0.02648926