I wrote earlier about using a custom
urchin.js. This is, of course, way more preferable, because it saves quite some bandwidth, and you won’t have to update every time Google updates it.
What you have to do is this:
custom_se.js adds the extra image search engines amongst other things, so you can see stuff is coming from image search. Otherwise the hits would be counted as “plain” Google. The
track-imagesearch.js rewrites the
_uInfo function, and makes sure Google “get’s” it’s own image search parameters.
You can download the two zipped together: Track Image-search Scripts.
This post will tell you how to do this with the new ga.js script.