/// <summary>
/// ラベリング
/// </summary>
/// <param name="sender"></param>
/// <param name="e"></param>
/// <example>
/// http://schima.hatenablog.com/entry/2015/08/19/215523
/// </example>
private void btnLabelling_Click(object sender, EventArgs e)
{
//画像準備
Mat matSrcLabelling = BitmapConverter.ToMat((Bitmap)picBoxDst.Image);
Mat matGrayLabelling = matSrcLabelling.CvtColor(ColorConversionCodes.BGR2GRAY);
Mat matBinary = matGrayLabelling.Threshold(0, 255, ThresholdTypes.Otsu | ThresholdTypes.Binary);
//ラベリング実行
Mat matLabel = new Mat();
int nLabels = Cv2.ConnectedComponents(matBinary, matLabel, PixelConnectivity.Connectivity8, MatType.CV_32SC1);
//色分け画像作成
Scalar[] colors = Enumerable.Range(0, nLabels + 1).Select(_ => Scalar.RandomColor()).ToArray();
colors[0] = Scalar.Black;
int rows = matBinary.Rows;
int cols = matBinary.Cols;
var dst = new Mat(rows, cols, MatType.CV_8UC3);
var labelIndexer = matLabel.GetGenericIndexer <int>();
var dstIndexer = dst.GetGenericIndexer <Vec3b>();
for (int y = 0; y < rows; y++)
{
for (int x = 0; x < cols; x++)
{
int labelValue = labelIndexer[y, x];
dstIndexer[y, x] = colors[labelValue].ToVec3b();
}
}
//描画
picBoxDst.Image = BitmapConverter.ToBitmap(dst);
}
public static void CalculoMatrizBinariaYRelleno(Mat imagen, out Mat imagenBinaria, out Mat imagenAberturaRelleno)
{
imagenBinaria = new Mat();
imagenAberturaRelleno = new Mat();
imagenBinaria = 255 - imagen;
Mat imagenBinariaProc = new Mat();
imagenBinaria.CopyTo(imagenBinariaProc);
Mat kernelMorfologicoElipse = Cv2.GetStructuringElement(MorphShapes.Ellipse, new Size(3, 3));
Mat kernelMorfologicoCruz = Cv2.GetStructuringElement(MorphShapes.Cross, new Size(3, 3));
Cv2.Erode(imagenBinariaProc, imagenBinariaProc, kernelMorfologicoElipse, iterations: 3);
Cv2.Dilate(imagenBinariaProc, imagenBinariaProc, kernelMorfologicoCruz, iterations: 2);
imagenBinariaProc = 255 - imagenBinariaProc;
Mat imagenTransfDistancia = new Mat();
Mat imagenTransfDistanciaBinaria = new Mat();
Cv2.DistanceTransform(imagenBinariaProc, imagenTransfDistancia, DistanceTypes.C, DistanceMaskSize.Mask5);
imagenTransfDistancia.ConvertTo(imagenTransfDistancia, MatType.CV_8U);
imagenTransfDistancia.CopyTo(imagenTransfDistanciaBinaria);
double maxDist, minDist, rangoDist;
imagenTransfDistanciaBinaria.MinMaxIdx(out minDist, out maxDist);
rangoDist = maxDist - minDist;
var indexadorImgDistancia = imagenTransfDistanciaBinaria.GetGenericIndexer <byte>();
for (int i = 0; i < imagenTransfDistancia.Rows; i++)
{
for (int j = 0; j < imagenTransfDistancia.Cols; j++)
{
if (indexadorImgDistancia[i, j] >= maxDist - rangoDist * 0.35 ||
indexadorImgDistancia[i, j] <= minDist + rangoDist * 0.15)
{
indexadorImgDistancia[i, j] = (byte)255;
}
else
{
indexadorImgDistancia[i, j] = (byte)0;
}
}
}
Mat marcadores = new Mat();
int numMarcadores = Cv2.ConnectedComponents(imagenTransfDistanciaBinaria, marcadores, PixelConnectivity.Connectivity8, MatType.CV_16UC1);
Dictionary <int, int> cantidadMarcador = new Dictionary <int, int>();
for (int i = 0; i < marcadores.Rows; i++)
{
for (int j = 0; j < marcadores.Cols; j++)
{
int areaMarcador = 0;
int m = marcadores.At <int>(i, j);
cantidadMarcador.TryGetValue(m, out areaMarcador);
if (areaMarcador > 0)
{
cantidadMarcador[m] += 1;
}
else
{
cantidadMarcador[m] = 1;
}
//try
//{
// cantidadMarcador[m] += 1;
//}
//catch
//{
// cantidadMarcador[m] = 1;
//}
}
}
int indiceMarcadorMax = cantidadMarcador.Select(v => new Tuple <int, int>(v.Key, v.Value))
.OrderByDescending(x => x.Item1).First().Item2;
Mat fondoMarcadores = new Mat();
marcadores.CopyTo(fondoMarcadores);
var indexador = fondoMarcadores.GetGenericIndexer <int>();
for (int i = 0; i < fondoMarcadores.Rows; i++)
{
for (int j = 0; j < fondoMarcadores.Cols; j++)
{
indexador[i, j] = (indexador[i, j] == indiceMarcadorMax) ? 255 : 0;
}
}
fondoMarcadores.ConvertTo(fondoMarcadores, imagenBinaria.Type());
Mat imagenBinariaSinFondo = new Mat();
Mat imagenRellenoHuecos = new Mat();
Cv2.Subtract(imagenBinaria, fondoMarcadores, imagenBinariaSinFondo);
Cv2.Add(imagenBinaria, imagenBinariaSinFondo, imagenRellenoHuecos);
Mat kernelMorfologicoApertura = Cv2.GetStructuringElement(MorphShapes.Rect, new Size(3, 3));
Cv2.MorphologyEx(imagenRellenoHuecos, imagenAberturaRelleno, MorphTypes.Open, kernelMorfologicoApertura, iterations: 2);
/* liberar memoria */
imagenBinariaProc.Release();
imagenBinariaSinFondo.Release();
imagenRellenoHuecos.Release();
kernelMorfologicoElipse.Release();
kernelMorfologicoCruz.Release();
kernelMorfologicoApertura.Release();
}
static void Main(string[] args)
{
//Thread capturaVideoThread = new Thread(new ThreadStart(Program.CapturarVideo));
//capturaVideoThread.Start();
VideoCapture captura = new VideoCapture("D:\\Dictuc\\out1.avi");
VideoWriter salida = new VideoWriter("D:\\Dictuc\\outSegmentado.avi", FourCC.XVID, 10.0, new Size(captura.FrameWidth, captura.FrameHeight), true);
Mat imagenProcesada = new Mat();
int numImg = 0;
while (true)
{
//captura.Read(imagen);
imagen = Cv2.ImRead("D:\\uvas2.jpg");
mutex.WaitOne();
imagen.CopyTo(imagenProcesada);
mutex.ReleaseMutex();
Mat imagenRuidoFiltrado = FiltradoRuido(imagenProcesada);
Mat imagenGrisContraste = EscalaGrisesEqualizada(imagenRuidoFiltrado);
Mat imagenGrisFrecAltasProc = FrecuenciasAltasPotenciadasContraste(imagenGrisContraste);
EdgeDetector edgeDetector = new EdgeDetector()
{
Threshold = (byte)18,
SparseDistance = 3,
WeightPreviousPoint = (float)2.0,
WeightCurrentPoint = (float)1.0,
WeightAfterPoint = (float)2.0,
};
EdgeDetector edgeDetector2 = new EdgeDetector()
{
Threshold = (byte)20,
SparseDistance = 5,
WeightPreviousPoint = (float)0.5,
WeightCurrentPoint = (float)1.0,
WeightAfterPoint = (float)0.5,
};
Mat imagenBordes = edgeDetector.EdgeImage(imagenGrisContraste);
Mat imagenBordes2 = edgeDetector2.EdgeImage(imagenGrisContraste);
Mat imagenBinaria, imagenAberturaRelleno;
CalculoMatrizBinariaYRelleno(imagenBordes2, out imagenBinaria, out imagenAberturaRelleno);
Mat mascaraInv = 255 - imagenAberturaRelleno;
Mat DistSureFg = new Mat();
Mat AreasSureFg = new Mat();
Mat Unknown = new Mat();
AreasSureFg += 1;
Cv2.DistanceTransform(imagenAberturaRelleno, DistSureFg, DistanceTypes.L1, DistanceMaskSize.Mask5);
int numAreas = Cv2.ConnectedComponents(imagenAberturaRelleno, AreasSureFg, PixelConnectivity.Connectivity8);
float[,] distValues = new float[DistSureFg.Rows, DistSureFg.Cols];
for (int i = 0; i < DistSureFg.Rows; i++)
{
for (int j = 0; j < DistSureFg.Cols; j++)
{
distValues[i, j] = DistSureFg.At <float>(i, j);
}
}
Segment[] segments = new Segment[numAreas];
for (int i = 0; i < AreasSureFg.Rows; i++)
{
for (int j = 0; j < AreasSureFg.Cols; j++)
{
int m = AreasSureFg.At <Int32>(i, j);
byte pixelSurrounding = 0;
float distance = (float)0;
//if (i >= 1)
//{
// distance = distValues[i - 1, j];
// if (distance == 2)
// {
// pixelSurrounding |= Segment.PIXEL_SURROUNDED_LEFT;
// }
//}
//if (i < AreasSureFg.Rows - 1)
//{
// distance = distValues[i + 1, j];
// if (distance == 2)
// {
// pixelSurrounding |= Segment.PIXEL_SURROUNDED_RIGHT;
// }
//}
//if (j >= 1)
//{
// distance = distValues[i, j - 1];
// if (distance == 2)
// {
// pixelSurrounding |= Segment.PIXEL_SURROUNDED_DOWN;
// }
//}
//if (j < AreasSureFg.Cols - 1)
//{
// distance = distValues[i, j + 1];
// if (distance == 2)
// {
// pixelSurrounding |= Segment.PIXEL_SURROUNDED_UP;
// }
//}
SegmentPixelData newPixel = new SegmentPixelData()
{
Distance = distValues[i, j],
CoordsXY = new int[] { i, j },
Concave = 0,
Indexes = new int[] { -1, -1 },
PixelsSurrounding = pixelSurrounding,
SubsegmentLabel = 0,
};
if (segments[m] == null)
{
segments[m] = new Segment()
{
SegmentId = m,
PixelData = new List <SegmentPixelData>(),
};
}
else
{
segments[m].MaxDistance = (segments[m].MaxDistance > newPixel.Distance) ? (int)segments[m].MaxDistance : (int)newPixel.Distance;
segments[m].PixelData.Add(newPixel);
}
}
}
Mat Centroides = new Mat();
imagenAberturaRelleno.CopyTo(Centroides);
var indexadorCentroides = Centroides.GetGenericIndexer <byte>();
var indexadorFiguras = AreasSureFg.GetGenericIndexer <Int32>();
foreach (var s in segments.Where(s => s.Circularity <= 0.9))
{
int distancia = 0;
if (s.Circularity > 0.7)
{
distancia = 5;
}
else if (s.Circularity > 0.5)
{
distancia = 5;
}
else if (s.Circularity > 0.25)
{
distancia = 6;
}
else
{
distancia = 6;
}
distancia = (distancia < s.MaxDistance) ? distancia : s.MaxDistance - 1;
foreach (var p in s.PixelData.Where(p => p.Distance <= distancia))
{
if (imagenAberturaRelleno.At <byte>(p.CoordsXY[0], p.CoordsXY[1]) != (byte)0)
{
indexadorCentroides[p.CoordsXY[0], p.CoordsXY[1]] = 0;
}
}
}
Cv2.Subtract(imagenAberturaRelleno + 255, Centroides, Unknown);
#region segmentStuff
//List<int> indexConcavos = segments.Where(s => s.Circularity > 1).Select(s => s.SegmentId).ToList();
//foreach (var s in segments.Where(s => s.Circularity < 1.1 && s.Circularity > 0.9))
//{
// foreach (var p in s.PixelData/*.Where(p => p.Distance == 1)*/)
// {
// if (imagenAberturaRelleno.At<byte>(p.CoordsXY[0], p.CoordsXY[1]) != (byte)0)
// {
// indexadorFiguras[p.CoordsXY[0], p.CoordsXY[1]] = 255;
// }
// }
//}
//foreach (var s in segments.Where(s => s.Circularity >= 1.1))
//{
// foreach (var p in s.PixelData/*.Where(p => p.Distance == 1)*/)
// {
// if (imagenAberturaRelleno.At<byte>(p.CoordsXY[0], p.CoordsXY[1]) != (byte)0)
// {
// indexadorFiguras[p.CoordsXY[0], p.CoordsXY[1]] = 255;
// }
// }
//}
//foreach (var s in segments)
//{
// s.SetPixelConcavity();
// s.Segmentation();
// foreach (var p in s.PixelData.Where(p => p.Distance == 1))
// {
// if (p.Concave == 1)
// {
// indexadorFiguras[p.CoordsXY[0], p.CoordsXY[1]] = 255;
// }
// if (p.Concave == -1)
// {
// indexadorFiguras[p.CoordsXY[0], p.CoordsXY[1]] = 255;
// }
// }
//}
//foreach (var s in segments)
//{
// //s.SetPixelConcavity();
// //s.Segmentation();
// foreach (var p in s.PixelData.Where(p => p.Distance == 2))
// {
// indexadorFiguras[p.CoordsXY[0], p.CoordsXY[1]] = 230;
// }
//}
//imagenAberturaRelleno.CopyTo(SureFg);
#endregion
Mat colormap = new Mat();
Mat Marcadores = new Mat();
Cv2.ConnectedComponents(Centroides, Marcadores);
Marcadores = Marcadores + 1;
var indexador2 = Marcadores.GetGenericIndexer <Int32>();
for (int i = 0; i < Unknown.Rows; i++)
{
for (int j = 0; j < Unknown.Cols; j++)
{
if (Unknown.At <byte>(i, j) == 255)
{
indexador2[i, j] = 0;
}
}
}
Marcadores.CopyTo(colormap);
colormap.ConvertTo(colormap, MatType.CV_8UC3);
Cv2.ApplyColorMap(colormap, colormap, ColormapTypes.Rainbow);
Cv2.ImWrite("D:\\Dictuc\\marcadores.png", Marcadores);
//Mat img1 = new Mat();
//imagen.CopyTo(img1);
Mat DistColor = new Mat();
//imagenGrisContraste = 255 - imagenGrisContraste;
Cv2.CvtColor(imagenAberturaRelleno, DistColor, ColorConversionCodes.GRAY2BGR);
DistColor.ConvertTo(DistColor, MatType.CV_8U);
Cv2.Watershed(DistColor, Marcadores);
Cv2.ImWrite("D:\\Dictuc\\watersheedIn.png", DistColor);
var indexador4 = imagen.GetGenericIndexer <Vec3i>();
//for (int i = 0; i < imagen.Rows; i++)
//{
// for (int j = 0; j < imagen.Cols; j++)
// {
// //if (Centroides.At<byte>(i, j) > 0)
// // indexador4[i, j] = new Vec3i(0, 0, 255);
// if (Marcadores.At<Int32>(i, j) == -1)
// indexador4[i, j] = new Vec3i(255, 20, 20);
// }
//}
for (int i = 0; i < imagen.Rows; i++)
{
for (int j = 0; j < imagen.Cols; j++)
{
//if (Centroides.At<byte>(i, j) > 0)
// indexador4[i, j] = new Vec3i(0, 0, 255);
if (imagenBordes.At <char>(i, j) > 0)
{
indexador4[i, j] = new Vec3i(255, 20, 20);
}
}
}
Mat seg = new Mat();
Marcadores.CopyTo(seg);
var indexador5 = seg.GetGenericIndexer <int>();
for (int i = 0; i < Marcadores.Rows; i++)
{
for (int j = 0; j < Marcadores.Cols; j++)
{
indexador5[i, j] = (Math.Abs(indexador5[i, j]) > 1) ? 255 : 0;
}
}
Mat kE1 = Cv2.GetStructuringElement(MorphShapes.Rect, new Size(1, 1));
Cv2.Erode(seg, seg, kE1, iterations: 3);
int thrs1 = 1500;
int thrs2 = 1800;
Mat edge1 = new Mat();
seg.ConvertTo(seg, MatType.CV_8U);
Cv2.Canny(seg, edge1, thrs1, thrs2, apertureSize: 5);
SimpleBlobDetector.Params params1 = new SimpleBlobDetector.Params()
{
MinThreshold = 0,
MaxThreshold = 255,
FilterByArea = true,
MinArea = 15,
FilterByCircularity = false,
MinCircularity = (float)0.01,
FilterByConvexity = false,
MinConvexity = (float)0.1,
FilterByInertia = false,
MinInertiaRatio = (float)0.01,
};
SimpleBlobDetector detectorBlobs = SimpleBlobDetector.Create(params1);
KeyPoint[] segmentosBlob = detectorBlobs.Detect(edge1);
Mat segmentosBlobMat = new Mat(1, segmentosBlob.Count(), MatType.CV_32FC1);
var indexador6 = segmentosBlobMat.GetGenericIndexer <float>();
for (int i = 0; i < segmentosBlob.Count(); i++)
{
indexador6[0, i] = segmentosBlob[i].Size;
}
Mat hist = new Mat();
Rangef[] ranges = { new Rangef(0, (float)segmentosBlob.Max(x => x.Size)) };
Cv2.CalcHist(new Mat[] { segmentosBlobMat }, new int[] { 0 }, null, hist, 1, new int[] { 100 }, ranges, uniform: true, accumulate: true);
float[] histAcumulado = new float[hist.Rows];
float[] histAcumuladoPorcentaje = new float[11];
histAcumulado[0] = hist.At <float>(0, 0);
for (int i = 1; i < hist.Rows; i++)
{
histAcumulado[i] = hist.At <float>(i, 0) + histAcumulado[i - 1];
}
int k = 1;
for (int i = 1; i < histAcumuladoPorcentaje.Count(); i++)
{
for (; k < hist.Rows; k++)
{
float porcentajeActual = histAcumulado[k] / segmentosBlob.Count() * 100;
float porcentajeAnterior = histAcumulado[k - 1] / segmentosBlob.Count() * 100;
float porcentajeRequerido = (float)((i < 10) ? i * 10 : 99.3);
if (porcentajeRequerido <= porcentajeActual)
{
float tamañoPorcentajeActual = (float)(k * (float)segmentosBlob.Max(x => x.Size) / 100.0);
float tamañoPorcentajeAnterior = (float)((k - 1) * (float)segmentosBlob.Max(x => x.Size) / 100.0);
float tasaVariacionTamañoPorcentaje = (tamañoPorcentajeActual - tamañoPorcentajeAnterior) / (porcentajeActual - porcentajeAnterior);
histAcumuladoPorcentaje[i] = tamañoPorcentajeAnterior + tasaVariacionTamañoPorcentaje * (i * 10 - porcentajeAnterior);
break;
}
}
}
for (int i = 0; i < histAcumuladoPorcentaje.Count(); i++)
{
Console.Write(histAcumuladoPorcentaje[i] + ",");
}
Console.WriteLine("");
// data1 = [];
// for i in range(0, len(keypoints1)):
// data1.append(keypoints1[i].size * coefTamano)
// #tamano.write(str(i)+'\t'+str(keypoints1[i].size*2*0.3)+'\n')
// cv2.line(im_with_keypoints1, (int(float(keypoints1[i].pt[0] - keypoints1[i].size)), int(float(keypoints1[i].pt[1]))), (int(float(keypoints1[i].pt[0] + keypoints1[i].size)), int(float(keypoints1[i].pt[1]))), (255, 0, 0), 1)
// cv2.line(im_with_keypoints1, (int(float(keypoints1[i].pt[0])), int(float(keypoints1[i].pt[1] - keypoints1[i].size))), (int(float(keypoints1[i].pt[0])), int(float(keypoints1[i].pt[1] + keypoints1[i].size))), (255, 0, 0), 1)
//# print(data1)
//n1, bins1, patches1 = hist(data1, 200,[0, max(data1)], normed = 100, cumulative = True, bottom = True, histtype = 'stepfilled', align = 'mid', orientation = 'vertical', rwidth = 1, log = False, color = "r")
// tamano = open(temp + "instancia_" + instancia + ".txt", "w")
// x = np.array(bins1)
// y = np.append([0], n1)
// xnew = [x[1], x[21], x[36], x[45], x[53], x[60], x[69], x[78], x[88], x[97], x[200]]
//ynew = [y[1], y[21], y[36], y[45], y[53], y[60], y[69], y[78], y[88], y[97], y[200]]
//tamano.write('INSERT INTO [dbo].[Granulometria](Cod_Instancia,Fecha,P_10,P_20,P_30,P_40,P_50,P_60,P_70,P_80,P_90,P_100, Filename) values (')
//tamano.write(instancia + ",CONVERT(datetime, '" + sys.argv[1][0:4] + "-" + sys.argv[1][4:6] + "-" + sys.argv[1][6:8] + ' ' + sys.argv[1][9:11] + ':' + sys.argv[1][11:13] + ':' + sys.argv[1][13:15] + "', 120)")
//for j in range(1, len(xnew)):
// #tamano.write (str(j)+'\t'+str(round(xnew[j],1))+'\t'+str(round(ynew[j]*100,2))+'\n')
// tamano.write(',' + str(round(xnew[j], 1)))
//tamano.write(",'" + sys.argv[1] + " - Resultado.jpg'")
//tamano.write(')')
//CvXImgProc.Thinning(mascaraInv, mascaraInv, ThinningTypes.ZHANGSUEN);
Mat imWithKeypoints1 = new Mat();
Cv2.DrawKeypoints(imagen, segmentosBlob, imWithKeypoints1, new Scalar(0, 0, 255), DrawMatchesFlags.DrawRichKeypoints);
var dataTamaños = segmentosBlob.Select(s => s.Size).ToArray();
Cv2.ImWrite("D:\\Dictuc\\output0" + numImg + ".png", imagen);
Cv2.ImWrite("D:\\Dictuc\\output1" + numImg++ + ".png", imWithKeypoints1);
Cv2.ImShow("Segmentado", imagen);
Cv2.ImShow("GrisContraste", imagenGrisContraste);
Cv2.ImShow("bordes90", imagenBordes);
Cv2.ImShow("bordes50", imagenBordes2);
salida.Write(imagen);
//System.Threading.Thread.Sleep(10);
Cv2.WaitKey(10);
imagenRuidoFiltrado.Release();
imagenGrisContraste.Release();
imagenGrisFrecAltasProc.Release();
imagenBordes.Release();
imagenBinaria.Release();
imagenAberturaRelleno.Release();
}
}
