can VC++ create convolution functions

[J_expoler2]

Hi
i'm study convolution functions but in problem
how to make function receive floatingpoint input array
and return output floating array too i'm try to create it but id doesn't work


float x[10] ---->| H(Z) | --------> float y[10]


:?: :?: :?:
Thank

 

[Hughes]

You may use pointer(or reference) as the function's parameters and modify the array itself. For example,

conv(x, h) will calculate the convolution of x * h and store the results to x.

conv(x, h, y) stores the results to y.

y=conv(x, h) returns the results pointer. The results array should be dynamic allocated by memory allocation functions.

 

[wagnert]

It's not safe returning object arrays. You can do that, but you should know well the language.

2 ways of doing "float x[10] ---->| H(Z) | --------> float y[10] ":

--
float function( float )
{
}

int nMember;
float x[ 20 ], y[ 20 ];

for( nMember = 0; nMember < 20; ++nMember )
y[ nMember ] = function( x[ nMember ] );

-- or --

bool function( float x[], float y[], int nMaxMembers )
{
int nMember;

for( nMember = 0; nMember < nMaxMembers; ++nMember )
y[ nMember ] = compute_function( x[ nMember ] );

return( true );
}

float x[ 20 ], y[ 20 ];

function( x, y, 20 );
--

Wagner.

 

[dynoboy]

Pointers will do the trick i remember having implemented almost all possible DSP Algos using C++ , so definately its possible in VC.

I am a big fan of pointers so i guess you will require pointers to functions and then make the function return the pointer...(even ** i.e. 2 level ptrs will be really useful)try it...
i'll see if i could send you the codes (prob is they are copyrighted if my seniors agree i'll send it across)

 

[siosavin]

You also can use a matrix class similar to matlab & write some function such as convolution for it.