• 2019 Update 4
  • 03/20/2019
  • Public Content
Contents

Use Signed Integer Data Types

Many image-processing kernels operate on
uchar
input. To avoid overflows, you can convert 8-bit input values and process as
16
- or
32
-bit
integer
values. Use signed data types (
shorts
and
ints
) in both cases, if you need to convert to floating point and back.

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.