Face movement detection for C# code

hello everyone

I'm university student and I'd love to have help from you people with experience in the field , I never worked on matlab

anyway I'm working on a project developed on C# programming language ( a program to detect face movement and according to the detection moving the mouse cursor )

I've written a code for detecting the movement in C# but when I move the cursor using the face the pointer's accuracy is really low the pointer is shaky..

and I've been told that Matlab guy's can help you ...

thank you ,I would really appropriate your help .

Answer 1

[Image Analyst]

We can help with MATLAB code.

[Hadeel Arslan]

aha cool so it's doable !! but how so!!

[Walter Roberson]

The point is that you are working in C# but this forum is for MATLAB. Are you planning to switch to MATLAB? There is no direct way to interface between MATLAB and C#, if I recall correctly.

[Image Analyst]

You don't need to mark this as an answer. It was actually a gentle hint that we won't help with your C# code, BUT if you post your translation of it into MATLAB code, we'd be happy and able to help. As of now we don't really have anything to go on. No real algorithm (which we don't help with much anyway), and no code at all. Nothing to help with. Your turn to supply more information so that we can help you. Help us to help you.

[Hadeel Arslan]

excuse me I'm not really aware the site. anyway what exactly does posting my translation of it into Matlab code!! and excuse me again I am Asking too many questions

[Walter Roberson]

If you are able to rewrite your code as MATLAB code then we might be able to help you with that code. There are other forums for assistance with C# code.

Note: if you are doing hardware access to do the tracking, then it might be difficult to translate your code to MATLAB; we would need to know more about how you get the positions to say how might be done in MATLAB.

[Hadeel Arslan]

well I'm not using any hardware except the webcam . I never used the MATLAB before but I'm gonna give it a try and learn alittle about it.

Answer 2

[Walter Roberson]

Examine the frequency of the movement in the conditions where you want the motion to settle down. Filter the position changes to block that range of frequencies, predicting the revised positions.

[Hadeel Arslan]

reasonable , though I have no Idea how to do it

[Walter Roberson]

Use your existing software, with a small modification to record the coordinates to a file, and use it in a series of trials to move the cursor to various locations. Best would be if you could also add a mechanism to allow you to insert end-of-trial markers in the file.

Then import the data to any handy analysis program (such as MATLAB), break it up into sub-trials, and run something like STFT (Short Time Fourier Transform) to produce a spectrogram to allow you to examine how the movement frequencies evolve as the subject moves around, and how it differs when the subject is settled and trying to focus in one area. Is the movement high frequency? Low frequency? Once you know, you can start designing filters to dampen the effective cursor movement under those circumstances.

If you do use MATLAB for the analysis, you will want the Signal Processing Toolbox for STFT (though you might be able to find a File Exchange contribution that implements it without that toolbox), and you might want the add of a utility such as this one that plots the data in different ways.

[Hadeel Arslan]

I think I have a hint of what should I do now. I'm gonna try it and let know what I come up with. thanks alot Walter.