Skip to content

7-Parameter Helmert Transformation

The 7-parameter Helmert transformation performs a translation in the three principal directions of an earth-centered, earth-fixed coordinate system, as well as rotations around those axes and a scaling of the coordinates.

Type

type: helmert_7param

Parameter Estimation

This operator uses non-linear least squares estimation with the full rotation matrix (no small-angle approximation). The rotation matrix is composed of elementary rotations about the x, y, and z axes, resulting in an inherently non-linear problem that is solved iteratively using the Levenberg-Marquardt algorithm.

This method is suitable for transformations involving large rotations where the small-angle approximation is not valid. For small rotations, Helmert7ParamSmallAngle provides a faster closed-form solution.

Options

Option Type Required Default Description
convention string Yes - Rotation convention (position_vector or coordinate_frame)
x float No 0.0 Translation in X direction (meters)
y float No 0.0 Translation in Y direction (meters)
z float No 0.0 Translation in Z direction (meters)
rx float No 0.0 Rotation about X axis (arc seconds)
ry float No 0.0 Rotation about Y axis (arc seconds)
rz float No 0.0 Rotation about Z axis (arc seconds)
s float No 0.0 Scale factor (ppm)

Example

operators:
- name: ITRF to ETRS89 (Full Rotation)
  type: helmert_7param
  convention: coordinate_frame
  x: 0.5
  y: 1.2
  z: -0.9
  rx: 0.001
  ry: 0.002
  rz: 0.003
  s: 0.005

Mathematical Background

Forward Transformation

A coordinate in vector \(\mathbf{V}_a\) is transformed into vector \(\mathbf{V}_b\) using:

\[ \mathbf{V}_b = \mathbf{T} + (1 + s \cdot 10^{-6}) \mathbf{R} \mathbf{V}_a \]

Where:

  • \(\mathbf{T} = (x, y, z)^T\) is the translation vector
  • \(s\) is the scale factor in parts per million (ppm)
  • \(\mathbf{R}\) is the \(3 \times 3\) rotation matrix

Rotation Matrix

The rotation matrix is given as

\[ \mathbf{R} = \mathbf{R}_3(r_z) \mathbf{R}_2(r_y) \mathbf{R}_1(r_x) \]

Where \(r_x, r_y, r_z\) are the rotation angles in radians and:

\[ \mathbf{R}_1(r_x) = \begin{bmatrix} 1 & 0 & 0 \\ 0 & \cos(r_x) & -\sin(r_x) \\ 0 & \sin(r_x) & \cos(r_x) \end{bmatrix}, \quad \]
\[ \mathbf{R}_2(r_y) = \begin{bmatrix} \cos(r_y) & 0 & \sin(r_y) \\ 0 & 1 & 0 \\ -\sin(r_y) & 0 & \cos(r_y) \end{bmatrix}, \quad \]
\[ \mathbf{R}_3(r_z) = \begin{bmatrix} \cos(r_z) & -\sin(r_z) & 0 \\ \sin(r_z) & \cos(r_z) & 0 \\ 0 & 0 & 1 \end{bmatrix} \]

Rotation Convention

Two conventions exists for the rotation matrix: Position Vector and Coordinate Frame. The above formulation is using the Position Vector convention. Transposing the rotation matrix \(\mathbf{R}\) changes the convention:

\[ \mathbf{R}_{position vector} = \mathbf{R}_{coordinate frame}^T \]

Inverse Transformation

The inverse transformation is given by:

\[ \mathbf{V}_a = -\mathbf{T} + \frac{1}{s} \cdot \mathbf{R}^{-1} \mathbf{V}_b \]

Where \(\mathbf{R}^{-1} = \mathbf{R}^T\) for proper rotation matrices.

Parameter Estimation

The 7-parameter Helmert transformation relates source coordinates \(\mathbf{S}\) to target coordinates \(\mathbf{T}\) through:

\[ \mathbf{T} = \mathbf{T}_{translation} + (1 + s \cdot 10^{-6}) \cdot \mathbf{R} \cdot \mathbf{S} \]

The estimation is done using the full rotation matrix (not the small-angle approximation), making this a non-linear problem solved iteratively using the Levenberg-Marquardt algorithm.

Methodology

  1. Coordinate Normalization: To improve numerical stability, coordinates are centered around their weighted centroids:

    • \(\bar{x} = \sum(w_i \cdot x_i) / \sum(w_i)\)
    • Normalized coordinates: \(X = S - \bar{x}\), \(Y = T - \bar{y}\)

  2. Non-linear Least Squares: The transformation model in normalized coordinates:

    \(\mathbf{Y} = \mathbf{t} + k \cdot \mathbf{X} \cdot \mathbf{R}^T + \varepsilon\)

    Where \(k = 1 + s \cdot 10^{-6}\) and \(\beta = [T_x, T_y, T_z, k, r_x, r_y, r_z]^T\) is the parameter vector.

  3. Iterative Solution: The minimization problem \(\min ||r(\beta)||^2\) is solved using Levenberg-Marquardt (scipy.optimize.least_squares with method='lm').

  4. Coordinate System Transformation: After convergence, translation parameters are transformed back to the original coordinate system:

\[ \mathbf{T}_{orig} = \mathbf{t} - k \cdot \mathbf{R} \cdot \bar{x} + \bar{y} \]

References