colocr (version 0.1.1)

roi_select: Select regions of interest

Description

Select regions of interest in an image using different morphological operations

Usage

roi_select(
  img,
  threshold,
  shrink = 5,
  grow = 5,
  fill = 5,
  clean = 5,
  tolerance = 0.1,
  n = 1
)

Arguments

img

An object of class cimg or a list of multiple cimg items

threshold

A numeric to be passed to threshold or a vector of values for each image in img

shrink

A numeric to be passed to shrink or a vector of values for each image in img

grow

A numeric to be passed to grow or a vector of values for each image in img

fill

A numeric to be passed to fill or a vector of values for each image in img

clean

A numeric to be passed to clean or a vector of values for each image in img

tolerance

A numeric to be passed to label or a vector of values for each image in img

n

A numeric of the number of regions of interest or a vector of values for each image in img

Value

A cimg. The original input img with an additional attribute label. label is a vector of integers. The labels for the selected regions of interests starts from 1 and 0 is ignored. When img is a list, a list is returned.

Details

The function applies several imager morphological manipulations to select the regions of interest. These include threshold which sets all values below certain cut to 0; shrink/grow for pixel set dilation and erosion; fill/clean for removing isolated regions and holes. When n is provided, the individual regions (connected components) are selected where tolerance is used to determine if two pixels belong to the same region.

Examples

# NOT RUN {
# load images
fl <- system.file('extdata', 'Image0001_.jpg', package = 'colocr')
img <- image_load(fl)

# choose ROI
newimg <- roi_select(img, threshold = 90)

# check the ROI labels
unique(attr(newimg, 'label'))

# }