Re: DM: Sensor question

[Mike Muratet]

We did some work for NASA a few years back analyzing synthethic 
aperture
radar images for general aviation crash sites. The sites were under 
the
forest canopy, necessiting the use of a low frequency radar, which 
meant
that the sites occupied less than a pixel. We were given four or five
images in a blind test, and told to identify the pixel containing the
crash. We found the sites by identifying outliers, those pixels that
didn't match the pattern of their surrounding pixels. We found all of
the crash sites, and had four (give or take) false alarms. There's a
paper, but I'll have to go dig up the reference. Hope this helps.

Regards,

Mike

Carol Romanowski wrote:

> Hi there -
>
> Is anyone doing work on mining sensor data?  Are there some good
> references/websites/etc around?
>
> Carol