Next: Full Text Index Contents: Conference
Page
Multiresolution Image Representation Through Wavelet
Compression for Speeding Up Navigation in the Visible Human Data Set Archive
|
Pietro Cerveri, M.S.
National Ph.D. Program in Biomedical Engineering
Bioengineering Department – Politecnico of Milano, Italy
cerveri@biomed.polimi.it
|
|
Francesco Pinciroli, Professor
Coordinator of the "Visible Human Dataset – Milano Mirror Site
" at CILEA
Chair of Medical Informatics – Bioengineering Department – Politecnico
of Milano, Italy
pinciroli@biomed.polimi.it
|
Abstract
The great drawback while navigating the Visible
Human Dataset repository is the huge dimension of each single fullcolor
image (2048x1216x24bit) although zip compressed. Dealing with a particular
set of fullcolor images, without being sure of their content, can require
to access many times to the VHD archive. Then in order to obtain the desired
samples the user is forced to download a great number of the files leading
to waste much time in waiting for. While this can not be a relevant problem
for batch visualization and processing, real-time consultations suffer
for that shortcoming. A suitable solution is to give the user a low resolution
image (lossy compressed) corresponding to the selected image. Once downloaded
in reduced time, decompressed and displayed, he/she can determine if download
the full resolution image. VHD image compression can become basic for many
applications requiring speedy image transmission across the Internet. Simple
tests show that compression ratios of 150 (7.5MB to 50KB), however maintaining
good visual properties, allow to download an image in about a few seconds.
While this approach requires the duplication of the whole archive however
this can be a very useful service for the users. In this ambit VISIBLE
HUMAN DATASET-MILANO MIRROR SITE can play a relevant role as a prototype.
Following developments involve remote browing, i.e. the on-line consultation
of the archive VHD-MMS through graphics interface acting on low resolution
version of the images.
Keywords: wavelet, lossy compression, multiresolution, progressive
transmission, embedded zero tree.
Next: Full Text Index Contents: Conference
Page
