Intel® Integrated Performance Primitives
Community support and discussions relating to developing high-performance vision, signal, security, and storage applications.
Announcements
This community is designed for sharing of public information. Please do not share Intel or third-party confidential information here.

Resizing with subpixel precision - Ipp32f?

John_M_1
Beginner
172 Views

If I want to work with Subpixel precision, I read on another post that someone is using Ipp32f types for that purpose.

But I fail to understand how that can help me.

If I have a Grayscale 16 image, where 1 pixel == 2 bytes, how would I represent the byte array in Ipp32f? And the function ippiResizeLinear_32f_C1R, for example, still takes IppiPoint and IppiSize which contain height and width in integers...

What am I missing here?

0 Kudos
3 Replies
Valentin_K_Intel
Employee
172 Views

Hi John,

If you need subpixel precision, you can use the function ippiResizeSqrPixel_16u_C1R. It works with real scale factors and shifts, but this function is deprecated and can be removed in next releases. Alternatively, ippiWarpAffine_16u_C1R can be used for resize processing, because resize operation is the special case of an affine transformation. ippiWarpAffine is slow compared with ippiResizeSqrPixel, we are working on warp functions to reach better performance.

Regards, Valentin

John_M_1
Beginner
172 Views

Thank you Valentin,

So, eventually when ippiResizeSqlPixel is dropped, ippiWarpAffine will be the ideal substitute for it?

Thanks

Valentin_K_Intel
Employee
172 Views

I consider that warp affine functions will be the most suitable for a resize transformation with subpixel precision, if ippiResizeSqlPixel is dropped.

Regards,
Valentin

Reply