topic Re: Deinterlacing a video frame in Intel® Integrated Performance Primitives

Deinterlacing a video frame

adi_shavit — Thu, 15 Jan 2004 01:55:04 GMT

Hi,

I want to use IPPi for some video analysis and I was wondering what's the best way to deinterlace a full-size video frame into 2 separate field frames.

In analog video (NTSC/PAL) a full size frame usually consists of two parts/fields. The even rows are the even rows of one captured frames and the odd rows are the odd rows of the following frame. This means that a 24 frame per sec. sequence is really 48 inteleaved frames, with each consequtive pair combined into one frame.

Is there a way to use IPP (4.0) to split each frame into 2 half height frames? A way other than just looping on all the rows and copying them using ippiCopy_<mod>() or ippsCopy_<mod>().

Is there an eqivalent to the stride concept in IPPS or IPPI?

Thanks,

Adi

Re: Deinterlacing a video frame

Intel_C_Intel — Wed, 28 Jan 2004 15:21:40 GMT

There is step parameter in functions. For ippIP functions it means the number of bytes between adjucent lines. Please, look on ippi manual, chapter 2, IPP Concepts for the details.

So you need to call ippiCopy twice, one time for the first field and the second time for the other field.

Regards,
Vladimir

Re: Deinterlacing a video frame

adi_shavit — Sun, 01 Feb 2004 17:03:46 GMT

Right! I don't know why I didn't think of that.

Thanks,

Adi