The following changes are done to make the CFA 16 bit compliant:
- Core filter data path is updated to 16 bits
- Gradient block
- The gradients are normalized by adding a right shift of 4, this
retains the range of gradients as the older implementation
- Gradients are compared with thresholds and wrt each other
- FLXD_Intensity
- Increated the size of the shift (after intensity calculation) by 1
bit (4 bits now)
- Refer to the description of SET0_INTENSITY0 (Intensity BitField),
(0x1d98)
- FLXD_ThrCalc()
- No update since intensity remains in the same range as previous
implementation