nppiWarpAffine_32f_C1R problem

Accelerated Computing CUDA CUDA Programming and Performance
1

Hello,

I am trying to apply an image warping operator to an image using the nppiWarpAffine_32f_C1R function. I am looking for a function to have the same behavior as this fproj function: http://megawave.cmla.ens-cachan.fr/stuff/guid3/node130.html.

My code looks like:

width_t = 799;

        width_r = 799;

        height_t = 599;

        height_r = 423;

NppiSize srcSize = {width_r, height_r};

        NppiRect srcROI;

        srcROI.x = 0;

        srcROI.y = 0;

        srcROI.height = height_r;

        srcROI.width = width_r;

	NppiRect dstROI;

		

	dstROI.x = 0;

        dstROI.y = 0;

        dstROI.height = height_t;

        dstROI.width = width_t;

	double coeffs[2][3];

	coeffs[0][0] = (double)width_r/(double)width_t;

	coeffs[0][1] = 0;

	coeffs[0][2] = 0;

	coeffs[1][0] = 0;

	coeffs[1][1] = (double)height_r/(double)height_t;

	coeffs[1][2] = 0;

	NppStatus stat = nppiWarpAffine_32f_C1R(dev_image_t, srcSize , width_r * 4, srcROI, dev_image_after_proj, width_t * 4, dstROI, coeffs, NPPI_INTER_CUBIC);

However, the only behavior that I see is that the output image is the same as the input image after the kernel call.

I use CUDA 5, because for CUDA 4.2 the function outputted an error status.

Do you have any suggestions? Maybe I can use a different function?

Thank you,

Valeriu

2

What do you expect your code should be doing?