1
0
2
3
1
0
2
3
5
4
1
0
2
3
1
0
2
3
1
0
2
3
• • •
• • •
• • •
Input pixels
Output pixels
1
0
2
3
5
4
X–1
X–2
X–3
X–4
Y–1
camisp-116
Tap
c
oefficients
from a single
phase
Public Version
Camera ISP Functional Description
www.ti.com
Figure 6-92. Camera ISP VPBE Resizer Alignment of Input Pixels to Tap Coefficients
also shows how the first output is computed when all taps are aligned with input pixels. To
compute the last several pixels in each line/column, the filter requires more input pixels than the following
equation calculates:
input size = output size*RSZ/256
In the former example, where the input size is X and the output size is Y, three extra input pixels are
required to generate the correct number of output pixels:
X = Y*256/256 + 3 (to account for the extra pixels required for filtering)
The input size calculation depends on the starting phase and rounding issues in the algorithm.
lists the input size calculations derived from the algorithm description in
. The input width
and height parameters must be programmed strictly according to these equations; otherwise, incorrect
hardware operation may occur.
Table 6-43. Camera ISP VPBE Resizer Input Size Calculations
8-Phase, 4-Tap Mode
4-Phase, 7-Tap Mode
[12:0] HORZ
(32*sph + (ow - 1)*hrsz + 16) >> 8 + 7
(64*sph + (ow - 1)*hrsz + 32) >> 8 + 7
[28:16] VERT
(32*spv + (oh - 1)*vrsz + 16) >> 8 + 4
(64*spv + (oh - 1)*vrsz + 32) >> 8 + 7
Where:
sph = Start phase horizontal (
[22:20] HSTPH)
spv = Start phase vertical (
[25:23] VSTPH)
ow = Output width (
[11:0] HORZ + extra)
oh = Output height (
[27:16] VERT)
hrsz = Horizontal resize value (
[9:0] HRSZ + 1)
vrsz = Vertical resize value (
[19:10] VRSZ + 1)
extra = 0 when
[17:16] ALGO = 0 (edge enhancement disabled)
extra = 4 when
[17:16] ALGO != 0 (edge enhancement enabled)
1218
Camera Image Signal Processor
SWPU177N – December 2009 – Revised November 2010
Copyright © 2009–2010, Texas Instruments Incorporated
