In UBC-MDS/pictureperrrfect: Pictureperrrfect

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

Welcome to Picture Perrrfect!

The pictureperrrfect package is made for anyone looking to make their pictures perrrfect! Users can easily gather statistical information about images, or just perform simple operations like image rotation, compression, and filtering. The pictureperrrfect package was built using functions from the imager package. package.

devtools::load_all()

# Load packages
library(pictureperrrfect)
library(imager)

# Load in the Image
vancouver <- load.image("../img/input_beautiful_Vancouver.jpg")

Demonstration

What does it mean to make your pictures perrrfect (TM)? To display how easy it is to manipulate and gather information on your images, I'll demonstrate each function on an image of the most beautiful city in the Western Hemisphere, Vancouver, Canada.

# Display image
plot(vancouver)

Rotation

The first operation is the most simple, a quick 90 degree rotation. The function rotation_perrrfect will rotate an image by some integer multiple of 90 degrees.

rotated_vancouver <- rotation_perrrfect(vancouver, 1)
plot(rotated_vancouver)

Compression

A common operation in Convolutional Neural Networks is image compression!

comp_vancouver <- compression_perrrfect(vancouver,
  kernel_size = 2,
  pooling_function = "max"
)
plot(comp_vancouver)

Filtering

Next we have a simple filtering operation. With this filter, we can blur or sharpen images using differing filter sizes for different effects.
It's grrreat!

build_filter("blur", 3)
filtered_vancouver <- filter_perrrfect(vancouver,
  filter_type = "blur",
  filter_size = 21L, custom_filter = NULL
)

plot(filtered_vancouver)

Image Propterties

Finally, lets get data on properties of our image, like width, height, pixel count, and RGB data.

get_property_perrrfect(vancouver)

UBC-MDS/pictureperrrfect documentation built on March 28, 2021, 1:28 p.m.