static void Test()
{
List<GenericDecisionTreeTrainingData<GrayPixel>> trainingData =
new List<GenericDecisionTreeTrainingData<GrayPixel>>();
GenericDecisionTree<GrayPixel> decisionTree = new GenericDecisionTree<GrayPixel>();
GenericTrainedDecisionTree trainedTree = null;
BinaryFormatter formatter = new BinaryFormatter();
using (FileStream fin = File.OpenRead("C:/users/ben/desktop/trained.dat"))
{
trainedTree = formatter.Deserialize(fin) as GenericTrainedDecisionTree;
}
Bitmap lowRes = (Bitmap)Bitmap.FromFile("C:/users/ben/desktop/feynman2.jpg");
BitmapData lowResBitmapData = lowRes.LockBits(
new Rectangle(0, 0, lowRes.Width, lowRes.Height),
ImageLockMode.ReadOnly,
lowRes.PixelFormat);
byte[] lowResBytes = new byte[lowResBitmapData.Stride * lowResBitmapData.Height];
Marshal.Copy(lowResBitmapData.Scan0, lowResBytes, 0, lowResBytes.Length);
lowRes.UnlockBits(lowResBitmapData);
Bitmap newBitmap = new Bitmap(lowRes.Width, lowRes.Height, lowRes.PixelFormat);
for (int y = 0; y < lowRes.Height; y++)
{
for (int x = 0; x < lowRes.Width; x++)
{
GrayPixel pixel = new GrayPixel();
pixel.Coordinate = new Point(x, y);
pixel.ImageData = lowResBytes;
pixel.ImageWidth = lowRes.Width;
pixel.ImageHeight = lowRes.Height;
int pixelClass = decisionTree.ClassifyWithTree(new GrayscaleDecisionTreeBuilder(),
trainedTree, pixel);
Color lowResColor = lowRes.GetPixel(x,y);
int pixelValue = lowResColor.R + pixelClass;
if(pixelValue < 0) pixelValue = 0;
if(pixelValue > 255) pixelValue = 255;
newBitmap.SetPixel(x, y, Color.FromArgb(pixelValue, pixelValue, pixelValue));
}
}
newBitmap.Save("C:/users/ben/desktop/new.jpg");
}
static void Train()
{
Bitmap lowRes = (Bitmap)Bitmap.FromFile("c:/users/ben/desktop/feynman.jpg");
Bitmap highRes = (Bitmap)Bitmap.FromFile("c:/users/ben/desktop/feynman2.jpg");
BitmapData lowResBitmapData = lowRes.LockBits(
new Rectangle(0, 0, lowRes.Width, lowRes.Height),
ImageLockMode.ReadOnly,
lowRes.PixelFormat);
byte[] lowResBytes = new byte[lowResBitmapData.Stride * lowResBitmapData.Height];
Marshal.Copy(lowResBitmapData.Scan0, lowResBytes, 0, lowResBytes.Length);
lowRes.UnlockBits(lowResBitmapData);
List<GenericDecisionTreeTrainingData<GrayPixel>> trainingData =
new List<GenericDecisionTreeTrainingData<GrayPixel>>();
for (int y = 0; y < lowRes.Height; y++)
{
for (int x = 0; x < lowRes.Width; x++)
{
GenericDecisionTreeTrainingData<GrayPixel> trainingDataPoint =
new GenericDecisionTreeTrainingData<GrayPixel>();
int originalIntensity = highRes.GetPixel(x, y).R;
int lowResIntensity =lowRes.GetPixel(x, y).R;
trainingDataPoint.Data = new GrayPixel();
trainingDataPoint.Data.Coordinate = new Point(x, y);
trainingDataPoint.Data.ImageData = lowResBytes;
trainingDataPoint.Data.ImageWidth = lowRes.Width;
trainingDataPoint.Data.ImageHeight = lowRes.Height;
trainingDataPoint.Class = (originalIntensity - lowResIntensity);
trainingData.Add(trainingDataPoint);
}
}
GrayscaleDecisionTreeParameters parameters = new GrayscaleDecisionTreeParameters
{
MaxRecursionLevels = 20,
SufficientGainLevel = 0,
MaximumOffset = 50,
MinimumOffset = -50,
NumberOfQuestions = 1000
};
Stopwatch sw = new Stopwatch();
sw.Start();
GenericDecisionTree<GrayPixel> decisionTree = new GenericDecisionTree<GrayPixel>();
GenericTrainedDecisionTree tree = decisionTree.Train(
new GrayscaleDecisionTreeBuilder(), parameters, trainingData);
BinaryFormatter formatter = new BinaryFormatter();
using (FileStream fout = File.Create("C:/users/ben/desktop/trained.dat"))
{
formatter.Serialize(fout, tree);
}
sw.Stop();
Console.WriteLine("Took " + sw.Elapsed.TotalSeconds + " seconds.");
Console.ReadLine();
return;
}
