The algorithm developped in Simulink is based on the same method used in Avalon Edge detector example from Altera. This example uses Avalon-MM.
But I don't have so much experience with Avalon and I don't see clear what will be the best for me Avalon-ST or Avalon-MM.
All incoming images will be 320x240, 76800 pixels * 16 bits = 153.6Kbytes.
I'm using Cyclone III NIOSII eval kit
Bandwith will be 5-10 frame/s
I can have 2 posibilities ,
1) After receiving an image by UDP protocol (Nios+niche stack), I think I should write the image to memory (SRAM or DDR) in this case, Should I use Avalon-MM?
2) In the case that I would receive images from hardware by BT.656 parallel interface (no NIOS intervention), I think I wouldn't have to write data in external memory. Should I use Avalon-ST in this case?