Overview
One of the most common issues
with e-mail today is that even with most stringent e-mail
policies in place, it is very difficult to effectively control
the content of image and movie attachments: most messages
with silly jokes, offensive or pornographic attachments
do not include any keywords in the message body that would
allow easy identification by keyword scan. Surely, the network
administrator can quarantine all movie clips and review
their content on a regular basis before allowing them through.
But it is a considerably time consuming and manual process
that only partially solves the problem.
What about JPEG and GIF and other image attachments? With
so many legitimate messages today containing company logos,
signatures or other legitimate business graphics, control
of visual e-mail content proves to be an unwelcome chore
for many IT managers.
Omniquad ImageTrap, a complete e-mail visual content management
system will address all these issues and save yourself a
great deal of time. This server-based plug-in is integrated
into Omniquad Mailwall e-mail content security framework
and is compatible with most mail
servers; it looks for image and movie attachments in inbound/outbound
mail and intercepts them, allowing messages through or stops
them depending on the content of attached images / movie
clips.
ImageTrap classifies image/movie attachments into:
- neutral content (company logos, technical drawings, family
photos, landscapes
etc)
- jokes (anything that makes you laugh but is unlikely to
cause offence)
- offensive/shocking (pictures sent to shock, provoke or
offend the viewer for example
images depicting women in derogatory manner etc. )
- partial nudity (partially exposed body parts, bikinis
etc)
- pornography
A unique fingerprint is generated for each image/movie clip
and the master Image Bank is queried to get the content
classification for the attachment itself. If the image content
is not known, a thumbnail is uploaded to the Image Bank
server for categorisation where an automated intelligent
process links it with a similar series already classified
or a human researcher in the operations centre will enter
correct classification. Only thumbnails are sent without
any sender/recipient information. Once your local ImageTrap
gets assigned a category for an image, it is stored in a
local cache and does not query the ImageBank again when
the same image is intercepted the next time.