Notes

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using System.Drawing;
using AForge.Imaging.Filters;
using AForge;

namespace Segmentation
{
public class HandSegmentation
{
private bool isSkin(Color c)
{
return (c.R > 95 && c.G > 40 && c.B > 20) &&
(Math.Max(c.R, Math.Max(c.G, c.R)) -
Math.Min(c.R, Math.Min(c.G, c.B))) > 15 &&
(Math.Abs(c.R - c.G) > 15 && c.R > c.G && c.R > c.G);
}

public Bitmap Apply(Bitmap originalImage)
{

//reduce image size so that less, bicubic resizes with less breakage
ResizeBicubic resizeObject = new ResizeBicubic(200, 200);
Bitmap smallOriginalImage = resizeObject.Apply(originalImage);
Bitmap copiedImage = (Bitmap)smallOriginalImage.Clone();

// to get the colour of the pixel passed as parameter
for (int x = 0; x < smallOriginalImage.Width; x++)
{
for (int y = 0; y < smallOriginalImage.Height; y++)
{
if (!isSkin(copiedImage.GetPixel(x, y)))
{
copiedImage.SetPixel(x, y, Color.Black);
}
}
}
copiedImage = Grayscale.CommonAlgorithms.BT709.Apply(copiedImage);
Threshold bwObj = new Threshold(50);
copiedImage = bwObj.Apply(copiedImage);


//applying closing to remove small black spots(closing holes in the image) i.e dilusion followed by erosion
AForge.Imaging.Filters.Closing filter = new Closing();
copiedImage = filter.Apply(copiedImage);
//pictureBox2.Image = copiedImage;

//extracting the biggest blob or a blob to get only the palms, here we get the bounding box
//bounding box is the smallest box having the image, hence we see only the palms
ExtractBiggestBlob biggestblobObject = new ExtractBiggestBlob();
copiedImage = biggestblobObject.Apply(copiedImage);



//we need to get the coordinates of the bounding box
IntPoint point = biggestblobObject.BlobPosition;

//create a rectangle to pass to the crop class, it takes x,y,height,width
Rectangle rect = new Rectangle(point.X, point.Y, copiedImage.Width, copiedImage.Height);

Crop cropObject = new Crop(rect);

//we pass the original image because that cohtains noise, we remove the background and have only palms
Bitmap croppedImage = cropObject.Apply(smallOriginalImage);


//we still have a lot of background which need to be removed as the background between the fingers have background
//hence we do a logical and between original image and the cropped image with pixels having white pixel
//this operation is called as masking
for (int x = 0; x < copiedImage.Width; x++)
{
for (int y = 0; y < copiedImage.Height; y++)
{
Color c = copiedImage.GetPixel(x, y);
if (c.R == 0 && c.G == 0 && c.B == 0)
{
croppedImage.SetPixel(x, y, Color.Black);

}
}
}


//it takes time because each pixel is checked and the image is huge,
//so we need to resize, hence we do smallOriginalImage


//we need to resize all objects to a standard size

croppedImage = resizeObject.Apply(croppedImage);
//pictureBox2.Image = croppedImage;

//convert the image to edge detected for the next step
croppedImage = Grayscale.CommonAlgorithms.BT709.Apply(croppedImage);
CannyEdgeDetector cannyObj = new CannyEdgeDetector(0, 0, 1.4);
croppedImage = cannyObj.Apply(croppedImage);
Threshold threshObj = new Threshold(20);
croppedImage = threshObj.Apply(croppedImage);
return (croppedImage);
}
}
}