[Image-SIG] Embedding Data in Images

Jason Cunliffe Jason Cunliffe" <jasonic@nomadics.org
Sat, 17 Nov 2001 17:38:17 -0500


I am designing net applications which take advantage of embedding text in
images. I don't mean visible rendered typographics. I mean ways to insert
and extract text in data blocks within the image format.
It is rich topic full of exciting potential..

PNG is a good candidate. The PNG spec defines tEXt chunks for doing this and
they work very nicely :-) I am using PNG therefore as my default format for
the application.

Problem now is I am concerned about completeness of PNG support, coupled
with larger file sizes than JEPG images. So wondering about adding JPEG

Does anyone have and knowledge/experience/examples/ideas/opinions about

A. Embedding and extracting text data in JPEG images ?

B. Using Python/PIL for this ?

I am aware that where there is not a defined matadata provision as PNG has,
that one might use rather more clandestine techniques, such as adding a
graphic 'edge' strip using it as super bar-code etc. The fact that JEPG is
'lossy' concerns me, especially as I don't _really_ understand what that
means at a low level.