In villegar/IPA: Image Processing and Analysis Pipeline

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%",
  dpi = 400
)

IPA: Image Processing & Analysis pipeline

r badger::badge_devel("villegar/MetaPipe", "blue") r badger::badge_codecov("villegar/IPA")

Overview

The goal of IPA is to provide a set of functions for image processing and analysis.

Installation

You can install the development version from GitHub with:

install.packages("remotes") # if not installed
remotes::install_github("villegar/IPA")

Example

You should start by loading IPA on your session.

library(IPA)

Remove background (rm_background)

This function removes the background from an image, based on a threshold value (bkg_thr). This can be found by creating a histogram of the image.

Note: currently, it only removes dark backgrounds; a future version will have an option for light backgrounds as well.

AB_001_B <- system.file("extdata", "AB_001_B.jp2", package = "IPA")
IPA::rm_background(AB_001_B, bkg_thr = 0.4)
AB_001_B_img <- imager::load.image(AB_001_B)

1. Start by loading the image to your workspace

AB_001_B <- system.file("extdata", "AB_001_B.jp2", package = "IPA")
AB_001_B_img <- imager::load.image(AB_001_B)

1. (Optional) Plot the raw image

plot(AB_001_B_img)

1. (Optional) Generate a "pixel histogram"

hist(AB_001_B_img, main = "Pixel histogram for AB_001_B")
abline(v = 0.4, col = "red") # Mark bkg_thr 

Based on the histogram, we should look for pixel concentration (background), for this example, the background is dark, so the threshold should be close to zero.

1. Call the rm_background function with the corresponding background threshold (bkg_thr = 0.4 for the example image).

IPA::rm_background(image_path = AB_001_B, bkg_thr = 0.4)

1. Load the newly created image without background. By default, this new image will be saved under the same path as the original one, with the same name, the suffix _wb (without background, transparent), and extension .png.

AB_001_B_wb <- system.file("extdata", "AB_001_B_wb.png", package = "IPA")
AB_001_B_wb_img <- imager::load.image(AB_001_B_wb)
plot(AB_001_B_wb_img)

Note: the function plot displays transparent pixels as 0 (black) values, that is why the previous plot shows a black background (to see the transparent image, look up for the generated *_wb.png file).

AB_001_B <- system.file("extdata", "AB_001_B.jp", package = "IPA")
IPA::rm_background(AB_001_B, bkg_thr = 0.4)
AB_001_B_wb <- system.file("extdata", "AB_001_B_wb.png", package = "IPA")
plot(imager::load.image(AB_001_B))
plot(imager::load.image(AB_001_B_wb))

1. (Optional) Detect unwanted elements and remove them, this can be done by extracting the alpha channel (transparency layer) and using the function find_area.

alpha <- imager::channel(AB_001_B_wb_img, 4) # the alpha channel is the fourth one
blobs <- IPA::find_area(alpha, start = c(2000, 5750), px_tol = 500)

if (!is.null(blobs)) {
  row.names(blobs) <- 1:nrow(blobs)
  knitr::kable(blobs)
}

Next, remove the previously detected objects, this can be done using the function add_alpha.

for (a in seq_len(nrow(blobs))) {
  area <- as.numeric(blobs[a, ])
  AB_001_B_wb_img <- IPA::add_alpha(AB_001_B_wb_img, area = area)
}
plot(AB_001_B_wb_img)
abline(h = 5750, v = 2000, col = "red") # Mark the trimmed area

Detect unwanted elements (find_area)

TODO

Remove unwanted elements (add_alpha)

TODO

RGB decomposition (rgb_decomposition)

This function extracts each layer from an image as a matrix, for further processing.

1. Start by creating an example image, in this case a simple barplot

test_data <- data.frame(name = c("R", "G", "B"), values = c(2, 2, 2))
RGB <- c("red", "green", "blue")
png("inst/figures/test_plot.png")
  barplot(height = test_data$values, names = test_data$name, col = RGB)
dev.off()

This code generates the following barplot (inst/figures/test_plot.png)

Which we want to decompose into 3 images:

Red layer	Green layer	Blue layer

For this purpose we can use the function rgb_decomposition, which can be called as follows

rgb_decomposition(subdirectory, 
                  # optional
                  extension = "jpg", 
                  RData = TRUE, 
                  recursive = TRUE)

where subdirectory is the name of a directory where to search for the images. The other arguments are optional; extension is the file format of the images, RData is a boolean flag to indicate whether or not the layers should be stored as RData format or CSV, the latter requires more disk space. Finally, recursive is a boolean flag on whether or not explore the subdirectory recursively for more images.

1. Call the rgb_decomposition function to extract the layers of the example image previously created:

rgb_decomposition("inst/figures/", "png", recursive = FALSE)

After running this, three new files (per image) will be on disk, called IMAGE-NAME-red.RData, IMAGE-NAME-green.RData, and IMAGE-NAME-blue.RData.

villegar/IPA documentation built on Aug. 11, 2021, 12:50 a.m.