Coversion of a digits dataset

Vote count:

0

I'm trying to understand how was the digits dataset coverted from pendigits to 8x8 optdigits.

The dataset is located here: http://ift.tt/13uJjD3

If you download the pendigits-orig.tra.Z archive and uncompress it, you will see files pendigits.tra and optdigits.tra and others, I have read about the dataset and also about how it was collected but I can't understand how was the pendigits coverted into a optdigits which is a 8x8 reprezentation of a written digits.

Could someone explain it to me, or point me to an algoritm or better a script which can covert it?

Many thanks for any help

asked 54 secs ago

Jan

153

Coversion of a digits dataset