Intel® Integrated Performance Primitives
Deliberate problems developing high-performance vision, signal, security, and storage applications.

IPP support for 16bit

debitsch
Beginner
369 Views
I'm working with 16bit images. When going from IPL to IPP the 16bit support for some functions is dropped. These are
- correlation functions and
- warp functions.
Are the any plans to add 16bit support for this functions in ipp? Especially for unsigned 16bit data.
0 Kudos
3 Replies
Intel_C_Intel
Employee
369 Views

Hi,

unfortunately, we have no support for 16s data in correlation function. If you find this functionality useful, please submit your request through the Intel Premier Support service, so we willconsider itat the planning stage for the next IPP versions.

Note, you can convert your Ipp16s data to Ipp32fto use existing IPP functions.
Regards,
Vladimir
0 Kudos
kirilf
Beginner
369 Views
Hi
I am using CrossCorrValid_NormLevel for 8u data for the integer output. It works fine. How do I use this function for the different types of data (16, 32)?
Should I ippiConvert all other types into 32f or ippiScale them into 8u?
Regards
Kiril

0 Kudos
Vladimir_Dudnik
Employee
369 Views
Right, you can convertother data types
Regards
Vladimir
0 Kudos
Reply