You would need to do two steps, both of which are notoriously unreliable.
the first step would be to use OCR (optical character recognition) software that supports the Greek alphabet to convert the written document into proper text the computer recognises. OCR software is temperamental at the best of times (even using printed text) so if they are hand written letters there is little chance this will work well.
Once you have it in a format the computer recognised as text rather than a JPEG (which is essentially just a photo of the letter) you would have to run it through a translation site such a babelfish (
http://babelfish.yahoo.com/ ) to convert it from Greek to English.
the problem is with any language conversion is that every language uses different sentence structures, so the results from any computer translation will read like pigeon English at best and be indecipherable at worst.
To give an example, I will translate this sentence to Greek and then back to English using babelfish.
In Greek it translates to (and I bet this is dreadful Greek.. (rojash??)
Για να δώσω ένα παράδειγμα, θα μεταφράσω αυτήν την πρόταση στα ελληνικά και έπειτα πίσω στα αγγλικά χρησιμοποιώντας babelfish.
and then translated straight back to English comes out as
In order to I give a example, I will then translate this prop