topic Re: Problems with using ippiDilateBorder_1u_C1R in Intel® Integrated Performance Primitives

Problems with using ippiDilateBorder_1u_C1R

XDu — Mon, 13 Sep 2021 11:23:27 GMT

Hi，

I am a college student and took the course of basic image processing. So I tested some morphological functions of the ipp library.such as the question I submitted last time about ippiMorphOpenBorder and got a satisfactory reply, I am very grateful.

When I was testing ippiDilateBorder_1u_C1R this time, I noticed that it had two extra parameters: bitOffset. According to the description of the document, I understand it as the offset of src and dst.My tests are as follows:

IppiSize roiSize = { 16, 3 };
Ipp8u pSrc[80] = {
10, 126, 83, 244, 162, 22, 183, 137, 224, 163, 114, 112, 42, 13, 160, 225,
39, 106, 206, 197, 105, 21, 189, 110, 79, 0, 99, 134, 41, 75, 51, 69,
158, 21, 158, 231, 75, 146, 218, 94, 173, 93, 226, 68, 65, 16, 190, 19,
233, 19, 4, 152, 146, 177, 159, 165, 20, 113, 207, 7, 89, 163, 113, 183,
194, 57, 129, 85, 198, 216, 221, 209, 221, 47, 75, 108, 156, 136, 190, 223
};
Ipp8u dst[80] = { 0 };
Ipp8u mask[3 * 3] = { 1,1, 1, 1,1, 1, 1, 1,1 };
IppiSize maskSize = { 3, 3 };
IppiMorphState *spec;
Ipp8u *buffer;
Ipp32s bufferSize, specSize;
ippiMorphologyBorderGetSize_1u_C1R(roiSize, maskSize, &specSize, &bufferSize);
spec = (IppiMorphState*)ippsMalloc_8u(specSize);
buffer = (Ipp8u*)ippsMalloc_8u(bufferSize);
ippiMorphologyBorderInit_1u_C1R(roiSize, mask, maskSize, spec, buffer);

IppStatus sts = ippiDilateBorder_1u_C1R(pSrc, 16, 0,dst, 16,0, roiSize, (IppiBorderType)1, 0, spec, buffer);

for (int i = 0; i < 5; i++){
for (int j = 0; j < 16; j++){
printf("%d ", dst[i * 16 + j]);
}
printf("\n");
}
printf("\n");
printf("\n");
printf("%d\n\n\n\n\n", sts);
ippsFree(buffer);
ippsFree(spec);
system("pause");

When the dstBitOffset is 0, the result is as follows：

127 255 0 0 0 0 0 0 0 0 0 0 0 0 0 0
255 255 0 0 0 0 0 0 0 0 0 0 0 0 0 0
255 255 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

and when the dstBitOffset is 7, the result is as follows：

128 191 127 0 0 0 0 0 0 0 0 0 0 0 0 0
128 255 127 0 0 0 0 0 0 0 0 0 0 0 0 0
128 255 127 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

It's same as when the dstBitOffset is 0,7bits is offset only when storing.

But when the dstBitOffset is 8, the result is as follows:

0 127 255 0 0 0 0 0 0 0 0 0 0 0 0 0
0 254 255 0 0 0 0 0 0 0 0 0 0 0 0 0
0 255 255 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Instead of offsetting a byte in storage as I thought, the result has changed. I don't understand this phenomenon, so I'd like to ask what this BitOffset parameter stands for.
Looking forward to your reply,
Thanks

Re: Problems with using ippiDilateBorder_1u_C1R

VidyalathaB_Intel — Tue, 14 Sep 2021 06:59:14 GMT

Hi,

Thanks for reaching out to us.

The issue raised by you is reproducible.

We are looking into it. We will get back to you soon.

Regards,

Vidya.

Re: Problems with using ippiDilateBorder_1u_C1R

Andrey_B_Intel — Mon, 20 Sep 2021 09:04:05 GMT

Hi, XDu.

I confirm this issue. dstBitOffset equal 8 means 1 byte of destination offset. As workaround you can add next "if" before calling "ippiDilateBorder_1u_C1R":

if (dstBitOffset >= {
dst = dst + (dstBitOffset >> 3);
dstBitOffset &= 7;
}

We will fix this issue in one of the next IPP releases.

Andrey

Re: Problems with using ippiDilateBorder_1u_C1R

XDu — Wed, 22 Sep 2021 02:24:20 GMT

thanks for your reply，but I don't quite understand the meaning of the following code.

if (dstBitOffset >= {
dst = dst + (dstBitOffset >> 3);
dstBitOffset &= 7;
}

Do you mean to offset dst dstBitOffset/8 bytes ahead of time when dstBitOffset > = 8?

But isn't the role of dsiBitOffset to offset every line of dst from the same bit?And whether there is a same question with srcBitOffset?

Thanks again for the reply,and because I don't know much about the written politeness of English, please forgive me if there is any offense.

XDu

Re:Problems with using ippiDilateBorder_1u_C1R

Gennady_F_Intel — Mon, 10 Jan 2022 14:50:37 GMT

XDu,

please check the latest IPP 2021 u5 and let us know if the problem you reported still exists.

thanks,

Gennady

Re:Problems with using ippiDilateBorder_1u_C1R

Gennady_F_Intel — Thu, 20 Jan 2022 06:14:31 GMT

This issue has been resolved and we will no longer respond to this thread. If you require additional assistance from Intel, please start a new thread. Any further interaction in this thread will be considered community only.