topic RGBToHLS in Intel® Integrated Performance Primitives

RGBToHLS

Adriaan_van_Os — Sat, 24 Sep 2016 09:43:42 GMT

In the headers and documentation for ippiRGBToHLS (both for ipp v9 and v2017) the following pseudo-code is given

if R = M2 then H = Cb - Cg

if G = M2 then H = 2 + Cr - Cb

if B = M2 then H = 4 + Cg - Cr

referring to

David F.Rogers

Procedural Elements for Computer Graphics

1985.pp.(403-406)

However, in that book (2nd ed. 1998) on page 629 I read

if R = M1 then H = Cb - Cg

if G = M1 then H = 2 + Cr - Cb

if B = M1 then H = 4 + Cg - Cr

and this does give different results. I haven't checked what ipp actually does, but at least there is a discrepancy in the documentation and headers.

Regards,

Adriaan van Os

Zhen_Z_Intel — Mon, 26 Sep 2016 07:33:10 GMT

Hi Adriaan:

It should be the mistake of document. I tested RGBToHLS function with simply code. The result is correct.

For instance, I tested with value of RGB is {56,45,36}, the output of MKL ippiRGBToHLS_8u_C3R function is like below:

And you could calculate the get value of H,L,S:

H=19/255*360=26.8 same as 60*(G - B)/(Max-Min)

L=46/255*100%=18

S=55/255*100%=21.6

Thank you for posting the mistake of developer guide document. We will fix it.

Best regards,
Fiona

Zhen_Z_Intel — Sat, 01 Apr 2017 02:24:58 GMT

Hi Adriaan,

The problem has been fixed in document, please refer to have a check.

Best regards,
Fiona