Hi,
I'm trying some experiments with the AutoFocus module that comes packaged with the DM36x IPNC (I'm using IPNC v2.5).
1. The AF engine runs off a 6x16 sized grid and the FVs here are all weighted and then averaged to arrive at a FV for the whole frame. But in the alg_afocus.c file, the weighting + accumulation is performed over a 4x14 grid. If I dump values for the whole 6x16, then the bottom 4 rows are all 0s. I'm not able to understand why this happens. So out of 96 (6x16) blocks, only 56 (4x14) are considered when 72 (6x12) have valid data. Can someone explain the logic behind doing things this way?
2. I can't assume that the FV is being calculated for the inner blocks leaving the border paxels out because the AF stats pointer is incremented without skipping at the beginning and end of each row; nor is the top row skipped.
Because of these issues, I am not really able to map the FV data for each individual block with the focus observed in the frame.
Any help would be appreciated.
Thanks,
Akshay