Using MatLab Add-in for MS Visual Studio 6

• Download source files - 93.6 Kb

Introduction

Following the line of the article posted by A.Riazi, Solving engineering problem using MATLAB C API, I will show in this small example how to use M Functions inside our VC++ 6 project.

The project calculates the convolution of two vectors, and shows the resulting vector in the MsChart control. The function that performs the convolution is written in MatLab and then through the MatLab Add-in converted to C code ready to be used inside our app.

Requirements

You need Visual Studio 6 and MatLab Release 12 installed with the MatLab C Library and Matlab Compiler.

Installing the MatLab Add-in

The add-in automates the integration of M-files into Visual C++ projects. The add-in for Visual Studio is automatically installed on your system when you run either mbuild -setup or mex -setup and select Microsoft Visual C/C++ version 6. In order to use the add-in you must follow these steps:

• Start MatLab, in the prompt type mbuild -setup
• Follow the menus and choose MS Visual C++6.0.
• Type the following commands in the matlab prompt :
  
  cd (prefdir)
  mccsavepath

These commands save the MatLab path to a file called mccpath in your user preferences directory (prefdir), ussually inside your documents and settings file. The path is used by the add-in because it runs outside Matlab and there is no other way for it to determine your Matlab path. If you add new directories to your Matlab path you will have to rerun this command if you want the add-in to see them.

• Configure the Matlab Add-in for Visual Studio 6 to work within MSVisual C++.
• Open MSVisualC++.
• Select Tools -> Customize from the MSVC menu.
• Click on the Add-ins and Macro Files tab.
• Check MATLAB for Visual Studio on the Add-ins and Macro Files list and click Close.

The floating MATLAB add-in for Visual Studio toolbar appears. The checkmark directs MSVC to automatically load the add-in when you start MSVC again.

Calculating the convolution in Matlab

We will write a simple function in MatLab that will perform the convolution of two vectors:

In1 and In2 and store the result in out.

We actually can use any matlab built-in function or toolbox function inside our function here we use only the function conv for simplicity.

function out=MyFunc(In1,In2)
%Returns the convolution of vector IN1 and IN2
out=conv(In1,In2);

Save the function with the name MyFunc.m

Writing the Application

Use the MFC AppWizard (exe) option to generate a Dialog Based Application and call it conv. Create a Button which will calculate the convolution of two given vectors.

Add the code below to the button's Message Handler.

//x1 and x2 are the input array to do the computation
//res will have the result, we already know that the length 
//of the convolution of 2 vectors is (length(x1)+length(x2)-1)=19
double x1[]={1,2,3,4,5,6,7,8,9,0};
double x2[]={4,5,6,7,4,5,8,9,0,7};
double res[19];

//create the arrays that will be passed to MyFunc
mxArray* In1;
mxArray* In2;
mxArray* Out;

//we make them Real and have 10 values
In1=mxCreateDoubleMatrix(1,10,mxREAL);
In2=mxCreateDoubleMatrix(1,10,mxREAL);

//this make In1=x1;In2=x2
memcpy(mxGetPr(In1),x1,10*sizeof(double));
memcpy(mxGetPr(In2),x2,10*sizeof(double));

//Call to MyFunc.m that return the convolution of In1 and In2
Out=mlfMyFunc(In1,In2);

//now we have the result(Out) in a double array 'res'
memcpy(res,mxGetPr(Out),19*sizeof(double));

//Destroy matrices
mxDestroyArray(Out);
mxDestroyArray(In1);
mxDestroyArray(In2);

Copy MyFunc.m into your VC++ project directory. Now go to the MatLab Add-in and click the .m++ (Add m-files to current project) button:

Select Windows Console Exe from the Combo Box and check Generate main file and debug mode. Press OK.

Then select the MyFunc.m file and click Open. After a while you will see new files added to your project.

Under MATLAB M-files you will see our MyFunc.m, you can even edit it from inside Visual C++. Under MATLAB C/C++ are MyFunc.c and MyFunc_main.c, generated by the MatLab Compiler. And all the headers needed by the compiler.

We can't build the app yet, we must add the following lines to convDlg.cpp:

#include "matlab.h"
#include "MyFunc.h"

Now the project should be built without any problem.   

Points of Interest

The steps described here don't include the MSChart part, you can see it from the source code, but you can see that the calculation is done by placing a  breakpoint inside the button message handler and checking the value of the res vector after the line:

memcpy(res,mxGetPr(Out),19*sizeof(double));

Well I hope this simple article be interesting for MatLab lovers, this is the first article I wrote, so I apologize if it is not clear enough. I will be glad to explain anything if you ask me. Greetings.