---
title: Weighted Procrustes Method
url: https://www.emergentmind.com/topics/weighted-procrustes-method
type: topic
---

# Weighted Procrustes Method

The weighted Procrustes method is a family of techniques for optimal alignment of vector sets, operators, or matrices under orthogonal, affine, or more general linear transformations, subject to weighting of individual correspondences or rows. Originating from classical shape analysis and statistical alignment, the weighted Procrustes problem now encompasses operator theory in Hilbert spaces, robust computer vision pose estimation, and modern conic optimization frameworks. Recent developments include fully differentiable solvers for deep learning pipelines and probabilistic relaxations for large-scale geometric estimation.

## 1. General Formulation of the Weighted Procrustes Problem

The weighted Procrustes problem seeks a transformation (often rigid, affine, or orthogonal) that best aligns two data sets, matrices, or operators, assigning a possibly non-uniform weight to each correspondence or data element.

- **Finite-dimensional, matrix setting:** Given $A \in \mathbb{R}^{p \times m}$, $B \in \mathbb{R}^{p \times n}$, and $W \in \mathbb{R}^{p \times p}$ (usually diagonal, $W \succeq 0$), the weighted Frobenius-norm Procrustes formulation is:
  $$
  \text{minimize}_{X \in C} \;\| W^{1/2}(A X - B) \|_F^2
  $$
  subject to structural constraint $X \in C$, e.g., orthogonality or obliqueness [2304.14961].

- **Operator-theoretic (Hilbert-space) formulation:** Given $A \in L(\mathcal{H})$ with closed range, $B \in L(\mathcal{H})$, and positive $W \in L(\mathcal{H})^{+}$ with $W^{1/2} \in S_p$ ($p$-Schatten class), the problem is:
  $$
  \min_{X \in L(\mathcal{H})} \| A X - B \|_{p,W}, \quad \text{where} \;\; \| X \|_{p,W} := \| W^{1/2}X \|_p
  $$
  [1610.00558].

- **Registration/alignment version:** Given paired points or keypoints $\{\mathbf{p}_n, \mathbf{q}_n\}_{n=1}^N$ (possibly in $\mathbb{R}^d$), with weights $w_n \ge 0$, the rigid weighted Procrustes objective is:
  $$
  \min_{R \in SO(d), t \in \mathbb{R}^d} \sum_{n=1}^N w_n \| R\mathbf{p}_n + t - \mathbf{q}_n \|^2
  $$
  [2502.20078, 2507.18541].

## 2. Mathematical Solutions and Characterization

### 2.1. Closed-form Solution (Rigid/Orthogonal Transform, Euclidean Case)

For point clouds or keypoint alignment with non-negative weights:

- Compute weighted centroids:
  $$
  \bar{\mathbf{p}} = \frac{1}{W_{\text{tot}}} \sum_n w_n \mathbf{p}_n, \quad
  \bar{\mathbf{q}} = \frac{1}{W_{\text{tot}}} \sum_n w_n \mathbf{q}_n
  $$
  with $W_{\text{tot}} = \sum_n w_n$.

- Compute centered, weighted cross-covariance:
  $$
  S = \frac{1}{W_{\text{tot}}} \sum_n w_n (\mathbf{p}_n - \bar{\mathbf{p}})(\mathbf{q}_n - \bar{\mathbf{q}})^\top
  $$

- Obtain the optimal rotation via SVD:
  $$
  S = U \Sigma V^\top, \quad R^* = U \, \mathrm{diag}(1, \det(U V^\top)) V^\top
  $$
  The reflection-correction via the determinant enforces $R^* \in SO(d)$ [2502.20078, 2507.18541].

- The translation is $t^* = \bar{\mathbf{q}} - R^* \bar{\mathbf{p}}$.

### 2.2. Weighted Least-Squares and Operator-theoretic Characterization

When $A$ and $B$ are bounded operators and $W$ a positive operator:

- The minimum exists if and only if $\mathrm{R}(B) \subseteq \mathrm{R}(A) + \mathrm{R}(A)^{\perp_W}$, with $\mathrm{R}(A)^{\perp_W} = W^{-1}(\mathrm{R}(A)^\perp)$.
- The solution is characterized as the $W$-inverse of $A$ in $R(B)$ and satisfies the normal equation:
  $$
  A^* W (A X - B) = 0
  $$
  The minimum-norm solution (when $A^* W A$ invertible) is $X^* = (A^* W A)^{-1}A^* W B$ [1610.00558].

### 2.3. Numerical/Optimization Algorithms

| Approach              | Setting               | Complexity/Remarks                                   |
|-----------------------|----------------------|------------------------------------------------------|
| SVD (closed form)     | Rigid, d=2,3         | $O(N)$ keypoints, $O(1)$ SVD per instance            |
| Normal equations      | General least-squares | Solve $A^*WA X = A^*W B$; Cholesky/QR direct solvers |
| Conic SDP reform.     | Structured $X$ constraints | $O(N^3)$ per SDP iteration; practical for moderate $N$    |
| Iterative/EM variant  | Probabilistic coupling| Alternates E-step (weights) and M-step (transf.)     |

For deep learning or large-scale cases, differentiability with respect to weights and locations (enabled by autodiff through SVD) is essential [2502.20078]. For outlier-robust cases, iterative reweighting or soft assignment (EM-type) algorithms are effective [2507.18541].

## 3. Generalizations: Constraints, Weighting Schemes, and Conic Optimization

Weighted Procrustes problems admit a wide range of generalizations:

- **Structured constraints:** The transformation $X$ or $R$ may be restricted to lie on a Stiefel manifold (orthogonality), be oblique (diagonal of $X^TX$ is 1), or subject to other algebraic constraints [2304.14961].
- **Conic SDP relaxation:** Many structured versions can be reformulated as rank-constrained semidefinite programs, enabling the encoding of additional linear, quadratic, or semidefinite side-constraints on $X$. The original objective (weighted least squares) is encoded as a linear trace over an auxiliary PSD block. Convex relaxations (dropping the rank constraint) provide lower bounds, while log-det or convex iteration heuristics yield near-optimal, feasible solutions. Practical computation is typically limited to moderate dimensions by cubic cost per SDP iteration [2304.14961].
- **Operator-norm and alternative norms:** While the Frobenius norm is classical, generalization to Schatten $p$-norms, $\ell_1$ or $\ell_\infty$ matrix norms, and operator norms is available within the same framework [1610.00558, 2304.14961].

## 4. Differentiable and Probabilistic Variants in Modern Applications

### 4.1. Differentiable Weighted Procrustes Solver

For integration with deep learning models, all components of the weighted Procrustes pipeline are designed to be fully differentiable, enabling end-to-end gradient-based training:

- Weighted centroids, covariance, SVD, and the final transformation mapping are all differentiable when supported by modern autodiff frameworks [2502.20078].
- Example: In monocular visual odometry (BEV-DWPVO), keypoints are extracted and weighted by learned confidences; matching and pose estimation is achieved via a single differentiable weighted Procrustes call per frame pair, providing stable, interpretable, and end-to-end scale-anchored odometry without explicit depth supervision.

### 4.2. Probabilistic Weighted Procrustes (EM/Softweight)

For large-scale or noisy correspondence problems, probabilistic formulations augment the weighted Procrustes objective by allowing the correspondence weights themselves to adapt:

- Weights $\gamma_l \ge 0$ are introduced for each pair, with an entropic regularizer $\epsilon \sum_l \gamma_l \ln \gamma_l$ and a "dustbin" parameter $\delta$ absorbing outlier mass.
- The algorithm alternates between computing soft assignments (E-step) and updating the transformation (M-step) via weighted Procrustes minimization.
- Gradients of the weighted objective with respect to transformation parameters can be computed analytically for efficient optimization [2507.18541].

This approach enables robust alignment of point clouds and poses in high-noise or partially ambiguous data, as in joint optimization of 3D Gaussian Splatting (3DGS) and global reconstructions from unposed image sets.

## 5. Operator-theoretic and Infinite-dimensional Perspectives

In the abstract Hilbert-space setting, the weighted Procrustes problem generalizes classical least-squares approximations:

- The existence and uniqueness of a minimizer are characterized by compatibility conditions between the weight (positive operator $W$) and the image/range of $A$ and $B$.
- The solution set is described by $W$-inverses and is equivalent to the solution of the normal equation $A^*W A X = A^*W B$ when the latter exists and is unique on the relevant subspace [1610.00558].
- When $W$ is singular or semi-definite, compatibility conditions are stricter, and the Kreĭn–Anderson–Trapp shorted operator is used to express the minimum [1610.00558].

This abstract framework supports extensions to weighted pseudoinverses, structured low-rank approximation, and $p$-Schatten class regularizations.

## 6. Applications and Empirical Evidence

Weighted Procrustes methodology appears extensively in statistical shape analysis, multidimensional scaling, factor analysis, computer vision, geometric point set registration, system identification, 3D scene reconstruction, and data fusion:

- **Monocular visual odometry:** Differentiable weighted Procrustes ensures stable and scale-anchored metric pose estimation, demonstrated across challenging driving datasets (NCLT, Oxford, KITTI) with state-of-the-art accuracy [2502.20078].
- **3D Gaussian Splatting and global point cloud registration:** Robust probabilistic weighted Procrustes enables accurate global alignment and pose estimation across hundreds of unposed images, handling tens of millions of points and outlier rejection efficiently [2507.18541].
- **Constrained alignment for multivariate data:** Conic optimization methods permit principled simultaneous handling of orthogonality, obliqueness, and other constraints, with documented success in moderate-scale numerical experiments and side-constraint satisfaction within tight residual bounds [2304.14961].

Weighted Procrustes algorithms thus form a core component in robust, scalable, and interpretable data alignment and registration across a wide range of computational and applied domains.

Source: https://www.emergentmind.com/topics/weighted-procrustes-method