Digital cameras store small image called "thumbnail" inside big image content. Usually this image gets lost in later image processing, but sometimes thumbnail contain sensitive information cropped off from big image.
Automated script downloads JPEG files from web and anayses them for thumbnail and image differences and and puts possible candidates in MySQL database. Result are visible on this page.
If you think some of images crawler found is very good, mail me tonu@@@@@jes.ee. Fix e-mail address to make it work :)
Source code of current page is available here and crawler is mnogosearch 3.2.37 patched to parse exif information. Patch is here
Idea itself came from outside. There was one famous case with Cath Schwartz and some blog entries but not real code or original images with thumbnails. So I decided to investigate problem myself.
Original picture:
EXIF Thumbnail:
Images possibly exposing extra information (dynamically updating):
'.htmlentities($url)."";
$query="SELECT count(*) AS count FROM exif_votes WHERE vote=3 AND url_id=".$id;
$res2=mysql_query($query) or die (mysql_error());
if($r=mysql_result($res2,0,"count")) {
for($j=0; $j<$r;$j++)
echo '
';
}
if(mysql_result($result,$i,"count")=='0')
echo " Not voted yet!";
echo "
\n";
$i++;
}
mysql_close();
?>