kevinmowers@yahoo.com wrote:

Does anyone have a good explanation to what u and v are and what x and y are and why we have to have 8x8 blocks?

I'm still working on this too so don't take this as fact, but from what I've figured out, x and y are your coordinates in colorSpace(rgb/YUV) and after the DCT -> u, v. so For F(u,v), u, v are fixed, but x and y are being indexed in the double sum. for 8x8 its just part of the algorithm

Do you accumulate all of the Fuv values of each pixel?

 

Yes, but you do it in 8x8 block, then quantize and sequence in the same blocks. and put it all together in the sequencing(i guess, not there yet).