Extract faces from live webcam video by c#

My target is make a program which take a webcam live video and save only extracted image of faces from that video.

So i use a very strong dotnet library name emguCV . what u can get from here. in short emgucv is

Emgu CV is a cross platform .Net wrapper to the Intel OpenCV image processing library. Allowing OpenCV functions to be called from .NET compatible languages such as C#, VB, VC++, IronPython etc. The wrapper can be compiled in Mono and run on Linux / Mac OS X

So first step is download the emgucv dll files because we are going to use it in our code. After download just add the dll in reference of ur project solution.

Now we need a very important file which is used as template to detect face we called it face_template, It is a xml file, u can get it from here . secondary link is here.

After download the xml file saved in “bin” folder and named it “face_template.xml” .

Now prepare a simple gui, A windows from, two picture box and a timer which is enabled at interval not less than 100.

first step is declare the library just add following

using Emgu.CV;
using Emgu.CV.Structure;
using Emgu.CV.CvEnum;

second step is initialize the web cam and load my detection template. U can just write this simple code for that

private Capture webcamvideo;
private HaarCascade facedetection;

 public Form1()
 {InitializeComponent();
 webcamvideo = new Capture(0); 
 facedetection = new HaarCascade("face_template.xml");
 }

Now timerclick event which will take the webcam captured image one after one. first see the full function code then analyze it.

private void timer1_Tick(object sender, EventArgs e)
{
 using (Image<Bgr, byte> Frame = webcamvideo.QueryFrame())
 {
  if (nextFrame != null)
{
 Image<Gray, byte> grayframe = Frame.Convert<Gray, byte>();
var faces =grayframe.DetectHaarCascade(facedetection , 1.4, 4,HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,new Size(Frame.Width/8, Frame.Height/8))[0];
foreach (var face in faces)
{
 Frame.Draw(face.rect, new Bgr(Color.Red), 3);
 Bitmap c = Frame.ToBitmap();
 Bitmap bmp = new Bitmap(face.rect.Width, face.rect.Height);
 Graphics g = Graphics.FromImage(bmp);
 g.DrawImage(c, 0, 0, face.rect, GraphicsUnit.Pixel);
 pictureBox2.Image = bmp;
 bmp.Save("myface"+DateTime.Now.Second.ToString() + ".jpeg"); 
 }
 pictureBox1.Image = nextFrame.ToBitmap();
}
 }
}

Now see what code does

using (Image<Bgr, byte> Frame = webcamvideo.QueryFrame())

it takes a frame from current web cam video stream. after .1 seconds it gets another frame because of timer function interval set to 100, if we want to fastened it we have to use threading.

if (Frame != null)

by this code we certain that we get a picture and can start analysis.

This analysis can divide in two steps.

• detact face
• extract the face

In detect face step we first make the image in greyscale using the following code

Image<Gray, byte> grayframe = Frame.Convert<Gray, byte>();

Now we take a special variable ‘var’ which a special type of Emgu.cv.ml. it can takes all the face location . and we use grayframe.DetectHaarCascade() function which take our template file , edggedetection algorithom type, scale factor and size of the part to except . So the code is like

var faces =grayframe.DetectHaarCascade( facedetection , 1.4, 4,

HAAR_DETECTION_TYPE.DO_CANNY_PRUNING,

new Size(Frame.Width/8, Frame.Height/8)

)[0];

now we have to create extracted image of face. “faces” has all of faces information if multiple face have. To extract we use

foreach (var face in faces)
{
Frame.Draw(face.rect, new Bgr(Color.Red), 3);
Bitmap c = Frame.ToBitmap();
Bitmap bmp = new Bitmap(face.rect.Width, face.rect.Height);
Graphics g = Graphics.FromImage(bmp);
g.DrawImage(c, 0, 0, face.rect, GraphicsUnit.Pixel);
pictureBox2.Image = bmp;
bmp.Save("myface"+DateTime.Now.Second.ToString() + ".jpeg"); 
 }

here frame.draw() draw the rectangle in the main frame.Graphics g is used to draw the extracted face in new image file which is bmp.

So using this way we can extract the face from live video like this image

hope, u find it cool. this is the most easiest way to detect and extract faces using web-cam or cc-cam