topic Logarithm base 10 in Intel® Integrated Performance Primitives

Logarithm base 10

Andreoli__Carlo — Wed, 27 Jan 2010 14:17:52 GMT

In the software i'm actually developing i have to apply a log₁₀(x) to a 16 bit images...from what i have seen the only similar function are:

- ipps10Log10_32s_Sfs ... but it works only on 32bit of information

- ippiLn_ this sound good(natural log) and i can apply log₁₀(x)=log_e(x)/log_e(10) ...but still i'm not totally stasfied

It's a common use on image processing to do something like c*log_b(X) where c and b are variables.
I have miss some function to implement this function?
Can i expect a future implementation on this?

Sorry for my bad english,

Mr. Andreoli Carlo

Logarithm base 10

Artem_V_Intel — Fri, 29 Jan 2010 07:43:06 GMT

Hello Mr. Carlo,

Could you please specify how many color channels your image have? Is it a right understanding that you have one color channel images with 16-bit unsigned pixel values?

In this case you may use for instance ippiConvert_16u32f_C1R function to convert your data in format appropriate for ipps10Log10_32s_Sfs function. After that you should use ippiConvert_32f16u_C1R function.

Thanks,

Art

Logarithm base 10

Andreoli__Carlo — Mon, 01 Feb 2010 07:34:23 GMT

Of course you are right, my image is one channel 16 bit....i surely will follow your suggestion...still it will get 3 passage before i get my final result...but by now i think it's the only way i can follow...
thanks Artem Vorobiev