FileIO.jl integration for image files
Format | Extensions | Provider | Implementation | Comment |
---|---|---|---|---|
JPEG | .jpg , .jpeg |
JpegTurbo.jl | Julia wrapper of libjpeg-turbo | Benchmark results against other backends |
OpenEXR | .exr |
OpenEXR.jl | Julia wrapper of OpenEXR | |
Portable Bitmap formats | .pbm , .pgm , .ppm |
Netpbm.jl | pure Julia | |
PNG (Portable Network Graphics) | .png |
PNGFiles.jl | Julia wrapper of libpng | Benchmark vs. ImageMagick & QuartzImageIO |
QOI (Quite Okay Image) format | .qoi |
QOI.jl | pure Julia | |
DEC SIXEL (six-pixels) graphics | .six , .sixel |
Sixel.jl | Julia wrapper of libsixel | |
TIFF (Tag Image File Format) | .tiff , .tif |
TiffImages.jl | pure Julia | check OMETIFF.jl for OMETIFF support |
Install with Pkg:
pkg> add ImageIO # Press ']' to enter te Pkg REPL mode
using FileIO
save("test.png", rand(Gray, 100, 100))
load("test.png")
save("test.ppm", rand(RGB, 100, 100))
load("test.ppm")
save("test.tiff", rand(RGB, 100, 100))
load("test.tiff")
Some image loaders may return custom AbstractArray types. By default, this package "canonicalizes" the returned type to be either Array
or IndirectArray
.
An exception is for calls like load(filename; mmap=true)
where the image data will be "lazily" loaded using memory-mapped IO, in which case the default is to allow the lower-level I/O package to return whatever AbstractArray type it chooses.
You can manually control canonicalization with load(filename; canonicalize=tf)
where tf
is true
or false
.
If you're using old Julia versions (VERSION < v"1.3"
), a dummy ImageIO version v0.0.1 with no real function will be installed.
In this case, you still need to install ImageMagick.jl to make FileIO.save
/FileIO.load
work.