Image transfer is very common. But images are defined by a considerable quantity of data, so that transfer can put strains on the communications lines supporting transfer. We are working on various products for progressive transmission. Here, we see the tremedous gains which can be made through image compression. Shown is a mammogram image. This was compressed to less than 3% of the original image's size. Then the image was uncompressed, and is the second image shown below.
The original image was 1024x1024 and just over 4 MB in size. The compressed image was just 126 kB. The images shown have been downsized to 512x512 for convenience. Visual fidelity was excellent, and the difference map between input and output contained small noise values.
mr_comp -R 20 im1s.fits im1scomp.fits
mr_decomp -i im1scomp.fits.MRC -o im1scomp.fits
Then in IDL, write_gif,'im1s512comp.gif',bytscl(rebin(b,512,512)) following false-colouring.
Image and Data Analysis: The Multiscale Approach, J-L Starck, F Murtagh and A Bijaoui, Cambridge University Press, 1998.