CODEWITHSUNDEEP
rmatlabimage-processing
Omar Wagih
Omar Wagih

Reputation: 8732

Slow image loading R using EBImage

Why is it much slower to load in a large image (~4MB) in R using the EBImage package, compared to matlab? is there anyway around this?

Note: I'm pretty sure EBImage is in some way a wrapper for imagemagick in R

in R:

system.time(im <- readImage("~/Desktop/image.jpg"))
Elapsed time is 10.935 seconds.

in Matlab:

tic; 
im = imread('~/Desktop/image.jpg');
toc;
Elapsed time is 0.555381 seconds.

using the raster package

system.time(im <- brick('image.jpg'))

takes 0.264 seconds

When trying to get the values

system.time(vals <- getValues(im))

takes 8.617 seconds so I'm back to square one since I need to extract a channel

Edit 2

I ended up using the package jpeg which has function readJPEG and performs better than anything i've tried out there. 

 system.time(x<-readJPEG('~/Desktop/image.jpg'))

Takes about 1.431 seconds for a 4mb image and returns a n x m x 3 matrix with each layer of the matrix being a color channel

Upvotes: 3

Views: 1225

Answers (1)

Simon O&#39;Hanlon
Simon O&#39;Hanlon

Reputation: 59960

I don't know why it's slow, but try using raster instead. You can plot the results of reading it in with image

require( raster )
im <- brick("~/Desktop/image.jpg")
image( im , y = 1 )

# Or for a rgb coloured image...
plotRGB( im , r=1 , g = 2 , b = 3 )

# To extract values of each layer
vals <- getValues( im )

# A quicker way to get the values would be to use the as.matrix method for rasters
vals <- as.matrix( im )

Raster might give you a warning about georeferencing, but it should work and it should be pretty darn quick.

Upvotes: 2

Related Questions