seiko's profile - activity

Hello. Let me explain a little what i want and need; I have an application where i need to calculate stress posture of a hand, intuitivness of position, stress and so on. But this should be very fast, for video processing. I am recognizing the hand, i have the exactly shape but is not enough for what i want. I want to check the current hand position with a set of images, and for that images i know the stress, position, etc. What i want to ask, is in open cv a function that can tell me wich of the stored image is the same or the closest with my current image wich is processed? I dont need a very good one but a fast one is preffered. Or, if you have another idea how to solve that, please tell me. Thanks!

Hello. I am a new user of OpenCV and i am wondering if it is possible. If it is not, what is the easiest way to implement such algorithm? I have done some simple algorithms but i need something more complex, i can do that but is not exptimized at all. I need to be fast, for video processing.

Thanks for answers!

Maybe somebody thoutght to implement this function? Open CV does not have something like this, just minEnclosingCircle. I need to send an array of points (contour) and return the maximum enclosing circle. Thank you!

I know that way...but when you close/open the palm the center is not the same and i need a stable point, as much is possible.

I am detecting the hand, using Defects but i need only the palm without fingers, the center of the palm without fingers. I wrote this piece of code:

int ht, //height left 
    hb, //height right
    wl, //width top
    wr; //width botton
bool ok = true;
label1.Text = label1.Text + ",";
for(int i = 10; i < skin.Height - 10; i += 10)
    for(int j = 10; j < skin.Width - 10; j+=10)
    {
        wl = wr = j;
        ht = hb = i;
        if(skin[i, j].MCvScalar.v0 == 0)
            ok = true;
            while(ok)
            {
                wl--;
                wr++;
                ht--;
                hb++;
                if( (wl <= 0 || wr >=skin.Height || ht <= 0 || hb >=skin.Width) || 
                    ( skin[j, ht].MCvScalar.v0 != 0 || skin[j, hb].MCvScalar.v0 != 0 ||
                      skin[wl, j].MCvScalar.v0 != 0 || skin[wr, j].MCvScalar.v0 != 0) )
                {
                    if(maxCircleRadius < (wr - wl) / 2)
                    {
                        maxCircleRadius = (wr - wl) / 2;
                        palmCenter.X = (wr + wl) / 2;
                        palmCenter.Y = (ht + hb) / 2;
                        currentFrame.Draw(new CircleF(new PointF(palmCenter.X, palmCenter.Y), 3), new Bgr(200, 125, 75), 2);
                        label1.Text = maxCircleRadius.ToString();
                    }
                    ok = false;
                }
            }
    }

but it takes around 0.5 sec for each image to detect the palm wich is way too much for me. Is there any other solution to detect the palm without fingers?