Prev: fsolve problem
Next: TriScatteredInterp
From: Arnaud Quintart on 19 May 2010 11:05
Hello,

I'm working on a vision based tracking system for UAVs. I am using Shawn Lankton's implementation of the Sparse Field Method for Active Contours (http://www.mathworks.co.uk/matlabcentral/fileexchange/23847-sparse-field-methods-for-active-contours) and I would like to adapt it to use it in Simulink.

This implementation calls external functions written in C++, to make it faster. I have asked around me for the most efficient way to use that code in Simulink but I'm getting different answers. I can either rewrite it in Matlab code (in an embedded m-file for instance) but I'm afraid the code will be far too slow for real time processing. Otherwise, I can use the existing C++ files and interface them with Simulink. In this case, what would be the best solution, knowing that the original files are written with the MEX protocol and that the code requires calling functions defined in various files? I have already tried an embedded m-file (using the eml.ceval function) without succeeding and I'm now working on an S-function, with the S-function builder. The best way would be to find something capable of calling the C++ mex function already written, but I haven't found any block capable of
doing that. Finally, if I manage to interface the C++ code with Simulink, should I expect a slower processing than in Matlab?

Thank you for your answers
AQ
From: Ralph Schleicher on 19 May 2010 18:23
"Arnaud Quintart" <arnaudquintart(a)hotmail.fr> writes:

> I'm now working on an S-function, with the S-function builder. The
> best way would be to find something capable of calling the C++ mex

Hello Arnaud,

you used the wrong wording, you want to say "find _someone_ capable"

> Finally, if I manage to interface the C++ code with Simulink, should
> I expect a slower processing than in Matlab?

Not if you do it right.

--
Ralph Schleicher <http://ralph-schleicher.de>

Development * Consulting * Training
Mathematical Modeling and Simulation
Software Tools
From: Mike on 1 Jun 2010 19:35
Arnaud - I have a similar problem. I am disturbed by the reply given by the Matlab representative below which is unhelpful and concerning.

If there is anyone who can assist I would be very grateful.

Mike H



Ralph Schleicher <rs+usenet(a)ralph-schleicher.de> wrote in message <871vd7cyzc.fsf(a)bravo.mueller-schleicher.i>...
> "Arnaud Quintart" <arnaudquintart(a)hotmail.fr> writes:
>
> > I'm now working on an S-function, with the S-function builder. The
> > best way would be to find something capable of calling the C++ mex
>
> Hello Arnaud,
>
> you used the wrong wording, you want to say "find _someone_ capable"
>
> > Finally, if I manage to interface the C++ code with Simulink, should
> > I expect a slower processing than in Matlab?
>
> Not if you do it right.
>
> --
> Ralph Schleicher <http://ralph-schleicher.de>
>
> Development * Consulting * Training
> Mathematical Modeling and Simulation
> Software Tools
From: Walter Roberson on 1 Jun 2010 20:21
Mike top-posted (now repaired)

> Ralph Schleicher <rs+usenet(a)ralph-schleicher.de> wrote in message
> <871vd7cyzc.fsf(a)bravo.mueller-schleicher.i>...
>> "Arnaud Quintart" <arnaudquintart(a)hotmail.fr> writes:
>>
>> > I'm now working on an S-function, with the S-function builder. The
>> > best way would be to find something capable of calling the C++ mex
>>
>> Hello Arnaud,
>>
>> you used the wrong wording, you want to say "find _someone_ capable"
>>
>> > Finally, if I manage to interface the C++ code with Simulink, should
>> > I expect a slower processing than in Matlab?
>>
>> Not if you do it right.
>>
>> --
>> Ralph Schleicher <http://ralph-schleicher.de>
>>
>> Development * Consulting * Training
>> Mathematical Modeling and Simulation
>> Software Tools

> Arnaud - I have a similar problem. I am disturbed by the reply given by
> the Matlab representative below which is unhelpful and concerning.

I suggest you read more closely, Mike. Ralph does not work for
Mathworks: he is an independent consultant who answered on his own
initiative. I have seen very few postings from him.
 | 
Pages: 1
Prev: fsolve problem
Next: TriScatteredInterp