/// <summary>
/// Inverse Haar Transform of a 2D image to a Matrix.
/// This is using the tensor product layout.
/// Performance is also quite fast.
/// Note that the input array must be a square matrix of dimension 2n x 2n where n is an integer
/// </summary>
/// <param name="image">2D array</param>
/// <param name="disableMatrixDimensionCheck">True if matrix dimension check should be turned off</param>
/// <returns>Matrix with inverse haar transform</returns>
public static Matrix InverseHaarWaveletTransform(double[][] image, bool disableMatrixDimensionCheck = false)
{
Matrix imageMatrix = new Matrix(image);
// Check that the input matrix is a square matrix of dimension 2n x 2n (where n is an integer)
if (!disableMatrixDimensionCheck && !imageMatrix.IsSymmetric() && !MathUtils.IsPowerOfTwo(image.Length))
{
throw new Exception("Input matrix is not symmetric or has dimensions that are a power of two!");
}
double[] imagePacked = imageMatrix.GetColumnPackedCopy();
HaarTransform.haar_2d_inverse(imageMatrix.Rows, imageMatrix.Columns, imagePacked);
Matrix inverseHaarMatrix = new Matrix(imagePacked, imageMatrix.Rows);
return(inverseHaarMatrix);
}
/// <summary>
/// TEST02 tests HAAR_2D and HAAR_2D_INVERSE.
/// </summary>
public static void test02()
{
int m = 16;
int n = 4;
int seed;
double[] u;
double[] v;
double[] w;
Console.Write("\n");
Console.Write("TEST02\n");
Console.Write(" HAAR_2D computes the Haar transform of an array.\n");
Console.Write(" HAAR_2D_INVERSE inverts the transform.\n");
//
// Demonstrate successful inversion.
//
seed = 123456789;
u = HaarTransform.r8mat_uniform_01_new(m, n, ref seed);
HaarTransform.r8mat_print(m, n, u, " Input array U:");
v = HaarTransform.r8mat_copy_new(m, n, u);
HaarTransform.haar_2d(m, n, v);
HaarTransform.r8mat_print(m, n, v, " Transformed array V:");
w = HaarTransform.r8mat_copy_new(m, n, v);
HaarTransform.haar_2d_inverse(m, n, w);
HaarTransform.r8mat_print(m, n, w, " Recovered array W:");
u = null;
v = null;
w = null;
return;
}
public static void TestHaarTransform()
{
double[][] mat = Get2DTestData();
Matrix matrix = new Matrix(mat);
//matrix.Print();
double[] packed = matrix.GetColumnPackedCopy();
HaarTransform.r8mat_print(matrix.Rows, matrix.Columns, packed, " Input array packed:");
HaarTransform.haar_2d(matrix.Rows, matrix.Columns, packed);
HaarTransform.r8mat_print(matrix.Rows, matrix.Columns, packed, " Transformed array packed:");
double[] w = HaarTransform.r8mat_copy_new(matrix.Rows, matrix.Columns, packed);
HaarTransform.haar_2d_inverse(matrix.Rows, matrix.Columns, w);
HaarTransform.r8mat_print(matrix.Rows, matrix.Columns, w, " Recovered array W:");
Matrix m = new Matrix(w, matrix.Rows);
//m.Print();
}
public static void RunTests()
{
HaarTransform.timestamp();
Console.Write("\n");
Console.Write("HaarTransformTest\n");
Console.Write(" C# version\n");
Console.Write(" Test the HAAR library.\n");
test01();
test02();
//
// Terminate.
//
Console.Write("\n");
Console.Write("HaarTransformTest\n");
Console.Write(" Normal end of execution.\n");
Console.Write("\n");
HaarTransform.timestamp();
return;
}
/// <summary>
/// TEST01 tests HAAR_1D.
/// </summary>
public static void test01()
{
int i;
int n;
int seed;
double[] u;
double[] v;
double[] w;
Console.Write("\n");
Console.Write("TEST01\n");
Console.Write(" HAAR_1D computes the Haar transform of a vector.\n");
//
// Random data.
//
n = 16;
seed = 123456789;
u = HaarTransform.r8vec_uniform_01_new(n, ref seed);
v = HaarTransform.r8vec_copy_new(n, u);
HaarTransform.haar_1d(n, v);
w = HaarTransform.r8vec_copy_new(n, v);
HaarTransform.haar_1d_inverse(n, w);
Console.Write("\n");
Console.Write(" i U(i) H(U)(i) Hinv(H(U))(i)\n");
Console.Write("\n");
for (i = 0; i < n; i++)
{
Console.Write(" ");
Console.Write("{0,2}", i);
Console.Write("{0,2}", " ");
Console.Write("{0,10:N6}", u[i]);
Console.Write("{0,4}", " ");
Console.Write("{0,10:N6}", v[i]);
Console.Write("{0,4}", " ");
Console.Write("{0,10:N6}", w[i]);
Console.Write("{0,4}", "\n");
}
u = null;
v = null;
w = null;
//
// Constant signal.
//
n = 8;
u = HaarTransform.r8vec_ones_new(n);
v = HaarTransform.r8vec_copy_new(n, u);
HaarTransform.haar_1d(n, v);
w = HaarTransform.r8vec_copy_new(n, v);
HaarTransform.haar_1d_inverse(n, w);
Console.Write("\n");
Console.Write(" i U(i) H(U)(i) Hinv(H(U))(i)\n");
Console.Write("\n");
for (i = 0; i < n; i++)
{
Console.Write(" ");
Console.Write("{0,2}", i);
Console.Write("{0,2}", " ");
Console.Write("{0,10:N6}", u[i]);
Console.Write("{0,4}", " ");
Console.Write("{0,10:N6}", v[i]);
Console.Write("{0,4}", " ");
Console.Write("{0,10:N6}", w[i]);
Console.Write("{0,4}", "\n");
}
u = null;
v = null;
w = null;
//
// Linear signal.
//
n = 16;
u = HaarTransform.r8vec_linspace_new(n, 1.0, (double)n);
v = HaarTransform.r8vec_copy_new(n, u);
HaarTransform.haar_1d(n, v);
w = HaarTransform.r8vec_copy_new(n, v);
HaarTransform.haar_1d_inverse(n, w);
Console.Write("\n");
Console.Write(" i U(i) H(U)(i) Hinv(H(U))(i)\n");
Console.Write("\n");
for (i = 0; i < n; i++)
{
Console.Write(" ");
Console.Write("{0,2}", i);
Console.Write("{0,2}", " ");
Console.Write("{0,10:N6}", u[i]);
Console.Write("{0,4}", " ");
Console.Write("{0,10:N6}", v[i]);
Console.Write("{0,4}", " ");
Console.Write("{0,10:N6}", w[i]);
Console.Write("{0,4}", "\n");
}
u = null;
v = null;
w = null;
//
// Quadratic data.
//
n = 8;
u = new double[n];
u[0] = 25.0;
u[1] = 16.0;
u[2] = 9.0;
u[3] = 4.0;
u[4] = 1.0;
u[5] = 0.0;
u[6] = 1.0;
u[7] = 4.0;
v = HaarTransform.r8vec_copy_new(n, u);
HaarTransform.haar_1d(n, v);
w = HaarTransform.r8vec_copy_new(n, v);
HaarTransform.haar_1d_inverse(n, w);
Console.Write("\n");
Console.Write(" i U(i) H(U)(i) Hinv(H(U))(i)\n");
Console.Write("\n");
for (i = 0; i < n; i++)
{
Console.Write(" ");
Console.Write("{0,2}", i);
Console.Write("{0,2}", " ");
Console.Write("{0,10:N6}", u[i]);
Console.Write("{0,4}", " ");
Console.Write("{0,10:N6}", v[i]);
Console.Write("{0,4}", " ");
Console.Write("{0,10:N6}", w[i]);
Console.Write("{0,4}", "\n");
}
u = null;
v = null;
w = null;
return;
}
