Mistake in Document for Cross Correlation

[kierangu]

Seems there is a mistake in IPP document about its cross correlation. It says pSrc2 is shifted for calculation. But my test indicates pSrc1 is shifted. The comments in header file seems comfirming my observation.

Anyone have the same conclusion?

[Chao_Y_Intel]

Hi, Kierangu

Which cross correlation function areyou using? We will check the document.

Regards,

Chao

[kierangu]

The function is ippsCrossCorr_***.

I realized that the documentation was right. However, the shifting is against common sense. Usually, the signal is shifted from left to right. The IPP function shifts from right to left. That's equivelant to shifting the other signal from left right. It caused me to think of mistaken document.

Anyway, calling

ippsCrossCorr_64f(x1,len,x2,len,y,(2*len-1),-(len-1))

equals matlab

y = xcorr(x2,x1);

[Chao_Y_Intel]

Thanks for letting us know this.