public void Stand_Evenkernel_OutputsError()
{
testImg.New("Quarters", new int[] { 6, 6 });
var stand = new LocalStandardization()
{
w1 = 2
};
double[,] image = testImg.Image.ToDouble();
Exception ex = Assert.Throws <Exception>(
delegate { stand.Standardize(ref image, "Reflect"); });
Assert.Equal("Kernel width is not odd!", ex.Message);
}
public void Stand_SmallKernelsOverride_Equalsreference()
{
testImg.New("Quarters", new int[] { 6, 6 });
var stand = new LocalStandardization(5, 3, 5, 1);
double[,] image = testImg.Image.ToDouble();
stand.Standardize(ref image, "Reflect");
double[,] refimage = new double[6, 6]
{
{ 0, -0.713520621472151, -1.0860460420492704, 1.1431629051894177, 0.84311553277412787, -2.2355540831313907E-15 },
{ -0.51110271561126808, -0.858138042569251, -1.2191098967853371, 0.88522909002589645, 0.50327268512720325, -0.79096598944498153 },
{ -0.90778462979852093, -1.1519031239765429, -1.4403470961301064, 0.52956642547784116, -0.023010004835671567, -1.0641502996862582 },
{ 1.0641502996862577, 0.023010004835671564, -0.52956642547784061, 1.4403470961301061, 1.1519031239765429, 0.90778462979852015 },
{ 0.79096598944498064, -0.50327268512720447, -0.88522909002589667, 1.2191098967853371, 0.85813804256925119, 0.51110271561126719 },
{ 0, -0.84311553277412821, -1.1431629051894172, 1.0860460420492706, 0.71352062147215034, 0 }
};
Assert.Equal(image, refimage);
}
/// <summary>
/// Calculates LBP features using LBPLibrary Nuget package.
/// Takes grayscale image as input.
/// Currently software inputs sum of mean and standard images of surface VOI.
/// </summary>
/// <returns>Feature array.</returns>
public static int[,] LBP(double[,] inputImage)
{
// Get default parameters
Parameters param = new Parameters();
// Grayscale standardization
var standrd = new LocalStandardization(param.W_stand[0], param.W_stand[1], param.W_stand[2], param.W_stand[3]);
standrd.Standardize(ref inputImage, param.Method); // standardize given image
// LBP calculation
LBPApplication.PipelineMRELBP(inputImage, param,
out double[,] LBPIL, out double[,] LBPIS, out double[,] LBPIR, out int[] histL, out int[] histS, out int[] histR, out int[] histCenter);
// Concatenate histograms
int[] f = Matrix.Concatenate(histCenter, Matrix.Concatenate(histL, Matrix.Concatenate(histS, histR)));
int[,] features = new int[0, 0];
return(Matrix.Concatenate(features, f));;
}
/// <summary>
/// Grading pipeline. Outputs the grades for each VOI as a string and results can be viewed from external grading window.
/// </summary>
/// <param name="zone">Name of the zone to be graded.</param>
/// <param name="sample">Sample name.</param>
/// <param name="mean">Mean image from the zone's VOI along depth-axis.</param>
/// <param name="sd">Standard deviation image from the zone's VOI along depth-axis.</param>
/// <param name="model_path">Path to grading model. Binary data file including values for PCA and regression.</param>
/// <param name="param_path">Path to LBP parameters.</param>
/// <returns></returns>
public static string grade_voi(string zone, string sample, double[,] mean, double[,] sd, string model_path, string param_path)
{
// Initialize Grading form
var grading = new GradingForm(); grading.Show();
grading.Text = zone;
// Load grading model
string path = LoadModel(out Model mod, out Parameters param, model_path, param_path);
// Sum mean and std images
var meansd = Elementwise.Add(mean, sd);
// Grayscale standardization
var standrd = new LocalStandardization(param.W_stand[0], param.W_stand[1], param.W_stand[2], param.W_stand[3]);
standrd.Standardize(ref meansd, param.Method); // standardize given image
// Replace NaN values from standardized image
bool nans = false;
var mu = LBPLibrary.Functions.Mean(meansd);
Parallel.For(0, meansd.GetLength(0), i =>
{
Parallel.For(0, meansd.GetLength(1), j =>
{
if (double.IsNaN(meansd[i, j]))
{
meansd[i, j] = 0;
nans = true;
}
});
});
if (nans)
{
Console.WriteLine("Input includes NaN values!");
}
// Show images to user
grading.UpdateModel(zone); grading.Show();
grading.UpdateMean(
DataTypes.DoubleToBitmap(mean),
DataTypes.DoubleToBitmap(sd),
DataTypes.DoubleToBitmap(meansd));
grading.Show();
//Get LBP features
grading.UpdateParameters(param);
double[,] features = LBP(meansd, param, out double[,] LBPIL, out double[,] LBPIS, out double[,] LBPIR, zone);
grading.UpdateLBP(
DataTypes.DoubleToBitmap(LBPIL),
DataTypes.DoubleToBitmap(LBPIS),
DataTypes.DoubleToBitmap(LBPIR));
grading.Show();
//Get grade
// Calculate PCA and regression
bool degenerated = FeaturesToGrade(features, mod, path, sample, out string grade);
Console.WriteLine(grade);
grading.UpdateGrade(grade, degenerated); grading.Show();
SaveResult(grade, path, sample);
return(grade);
}
