private void borderFollow(ushort j)
{
cvsim.Source = images[j];
cvsbmp = UtilFn.BitmapImage2Bitmap(images[j]);
Bitmap resbmp = UtilFn.ResizeBitmap(cvsbmp, (int)cvsim.Width, (int)(cvsim.Width * images[j].Height / images[j].Width));
Grayscale gfilter = new Grayscale(0.2125, 0.7154, 0.0721);
Bitmap grayImage = gfilter.Apply(cvsbmp);
BorderFollowing bf = new BorderFollowing();
List <Accord.IntPoint> contour = bf.FindContour(resbmp);
imageInfo.Text = contour.Count + ": ";
foreach (var c in contour)
{
Ellipse dot = new Ellipse();
dot.Width = 4;
dot.Height = 4;
dot.StrokeThickness = 2;
dot.Stroke = System.Windows.Media.Brushes.Violet;
Canvas.SetTop(dot, c.Y - 1 + (int)((cvs.Height - resbmp.Height) / 2));
Canvas.SetLeft(dot, c.X - 1);
imageInfo.Text += "(" + c.X + ", " + c.Y + ");";
cvs.Children.Add(dot);
}
imagesEdited.Add(UtilFn.rtbToBitmapImage(UtilFn.ExportToPng(null, cvs)).Clone());
cvs.Children.RemoveRange(1, cvs.Children.Count);
}
private void faceDetect(ushort j)
{
cvsbmp = UtilFn.BitmapImage2Bitmap(images[j]);
var cascade = new Accord.Vision.Detection.Cascades.FaceHaarCascade();
var detector = new Accord.Vision.Detection.HaarObjectDetector(cascade, minSize: 50,
searchMode: Accord.Vision.Detection.ObjectDetectorSearchMode.Average);
System.Drawing.Rectangle[] rectangles = detector.ProcessFrame(cvsbmp);
List <Bitmap> listBitmap = new List <Bitmap>();
foreach (System.Drawing.Rectangle r in rectangles)
{
listBitmap.Add(UtilFn.CropImage(cvsbmp, r.X, r.Y, r.Width, r.Height));
}
if (rectangles.Count() == 0)
{
imageInfo.Text = "No faces detected!";
}
else
{
//imagesEdited[j] = (converter.Convert(UtilFn.MergeImages(listBitmap), Type.GetType("BitmapImage"), null, null) as BitmapImage).Clone();
foreach (var x in listBitmap)
{
imagesEdited.Add((converter.Convert(x, Type.GetType("BitmapImage"), null, null) as BitmapImage).Clone());
}
}
}
private void edgeDetect(ushort j)
{
cvsbmp = UtilFn.BitmapImage2Bitmap(images[j]);
Grayscale gfilter = new Grayscale(0.2125, 0.7154, 0.0721);
Bitmap grayImage = gfilter.Apply(cvsbmp);
GaborFilter filter = new GaborFilter();
Bitmap output = filter.Apply(grayImage);
imagesEdited.Add(converter.Convert(output, Type.GetType("BitmapImage"), null, null) as BitmapImage);
}
private void trainSVMim(object sender, RoutedEventArgs e)
{
imageInfo.Text = "Training SVMim!";
if (bow == null || imagesEdited.Count == 0)
{
MessageBox.Show("NO TRAINED BoW MODEL or NO IMAGES LOADED");
return;
}
Bitmap[] trainIms = new Bitmap[imagesEdited.Count];
ushort z = 0;
foreach (BitmapImage b in imagesEdited)
{
trainIms[z++] = UtilFn.BitmapImage2Bitmap(b);
}
int[] labels = new int[imagesEdited.Count];
for (z = 0; z < imagesEdited.Count; ++z)
{
labels[z] = selected[z] ? +1 : -1;
}
if (bow == null)
{
MessageBox.Show("NO TRAINED BoW MODEL");
return;
}
double[][] features = bow.Transform(trainIms);
var teacher = new SequentialMinimalOptimization <Linear>()
{
Complexity = 70 // margin hardness
};
svmIm = teacher.Learn(features, labels);
imageInfo.Text = "Done";
}
private void PredictImages(object sender, RoutedEventArgs e)
{
imageInfo.Text = "Predicting";
if (bow == null)
{
MessageBox.Show("No BoW model!");
return;
}
if (svmIm == null)
{
MessageBox.Show("No SVM model!");
return;
}
Bitmap[] trainIms = new Bitmap[imagesEdited.Count];
ushort z = 0;
foreach (BitmapImage b in imagesEdited)
{
trainIms[z++] = UtilFn.BitmapImage2Bitmap(b);
}
double[][] features = bow.Transform(trainIms);
output = svmIm.Decide(features);
if (output != null)
{
if (output[j] && !cvs.Children.Contains(rectSel2))
{
cvs.Children.Add(rectSel2);
}
else if (!output[j] && cvs.Children.Contains(rectSel2))
{
cvs.Children.Remove(rectSel2);
}
}
imageInfo.Text = "Done";
btnCorrect.IsEnabled = true;
}
private void cluster(ushort j)
{
cvsbmp = UtilFn.BitmapImage2Bitmap(images[j]);
var imageToArray = new ImageToArray(min: -1, max: +1);
var arrayToImage = new ArrayToImage(cvsbmp.Width, cvsbmp.Height, min: -1, max: +1);
int kk;
double[][] pixels; imageToArray.Convert(cvsbmp, out pixels);
try
{
kk = Int16.Parse(kCluster.Text);
}
catch (Exception)
{
return;
}
if (kk < 1)
{
return;
}
KMeans kmeans = new KMeans(k: kk)
{
Distance = new SquareEuclidean(),
Tolerance = 0.05
};
var clusters = kmeans.Learn(pixels);
int[] labels = clusters.Decide(pixels);
double[][] replaced = pixels.Apply((x, i) => clusters.Centroids[labels[i]]);
Bitmap result; arrayToImage.Convert(replaced, out result);
imagesEdited.Add(converter.Convert(result, Type.GetType("BitmapImage"), null, null) as BitmapImage);
}