Hi Stefan. pixel-wise sum of differences? that sounds cool, I never thought of that! that would make it really damn fast. if that wasnt too convoluted that would be an amazing way to do it!
I was just going to store the "samplet" itself as the class centres. but i guess if you sum it up all into 1 number, that would be even better. speed, simplicity and performance is everything, especially if your on a little microcontroller on a robot. Getting all overlaps, Yeh, maybe every single sample is a bit too much, but that would be ideal. A gpu however would do it easily, if you did it parallel chunks. I cant guess those sounds dude... could be anything. :P ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T89dc799e249e2d0b-M95bcfcaf02b59c03ef58ce77 Delivery options: https://agi.topicbox.com/groups/agi/subscription