Angle of Linear Polarization (AoLP) Explained

• AoLP is defined as half the arctan2 of U and Q derived from Stokes parameters, offering a precise measurement of polarization orientation.
• It is central to imaging polarimetry and spectroscopy, enabling detailed analysis of material structures, biological tissues, and atmospheric phenomena.
• Accurate AoLP mapping requires careful calibration, demosaicking, and noise management to ensure reliable results across diverse applications.

The Angle of Linear Polarization (AoLP) is a fundamental observable in polarimetric sensing that quantifies the orientation of the plane of vibration of the electric field vector in linearly polarized light, as projected onto an image or reference plane. It is defined mathematically from the linear Stokes parameters and encodes rich physical and structural information about materials, biological tissues, atmospheric phenomena, and engineered surfaces, making it a critical variable in diverse polarimetric imaging, spectroscopy, and remote sensing applications.

1. Mathematical Definition and Formalism

AoLP is universally defined via the Stokes parameters, specifically as half the two-argument (quadrant-resolving) arctangent of the ratio of the U and Q components: $\mathrm{AoLP} = \frac{1}{2}\arctan2(U, Q)$ or in expanded form using intensity images at fixed analyzer angles (e.g., 0°, 45°, 90°, 135°): $$S_0 = I_{0} + I_{90}, \quad S_1 = Q = I_{0} - I_{90}, \quad S_2 = U = I_{45} - I_{135}$$

$\mathrm{AoLP} = \frac{1}{2}\arctan2(S_2, S_1)$

The output domain is typically mapped to $[0, \pi)$, reflecting the physical invariance of linear polarization orientation under 180° rotation (Alighieri, 2016, Žurauskas et al., 6 Nov 2025, Duhovic et al., 2024). In radioastronomy and CMB applications, this definition corresponds to Stokes coordinate conventions and, depending on context, angles may be reported in degrees or radians over $[0^\circ, 180^\circ)$ or $(-90^\circ, 90^\circ]$.

2. Coordinate System Conventions and Historical Context

There exist two coordinate conventions for reporting AoLP, distinguished by their angular zero point and sense of increase:

• IAU Convention (adopted in 1973): Zero at celestial North, angles increase counter-clockwise (CCW) when facing the source. AoLP is measured from 0° to 180° towards East (Alighieri, 2016).
• CMB/WMAP Convention: Zero at celestial South, angles increase clockwise (CW) when facing the source. Used in CMB experiments post-WMAP and propagating to Planck.

Conversion between these conventions requires a 180° flip and a sign inversion of the U parameter: $\chi_{\text{CMB}} = 180^\circ - \chi_{\text{IAU}}$ This distinction is critical, as misapplication can induce artificial polarization frame rotations, E–B leakage, or spurious astrophysical signals (Alighieri, 2016).

3. Measurement Methodologies in Imaging Polarimetry

Polarization Camera Architectures

Modern imaging polarimeters employ micropolarizer arrays (also known as division-of-focal-plane polarimeters) that overlay each superpixel with wire-grid polarizers at fixed axes (commonly 0°, 45°, 90°, 135°). In a single exposure, this yields four spatially interleaved images, $I_{0}$, $I_{45}$, $I_{90}$, $$S_0 = I_{0} + I_{90}, \quad S_1 = Q = I_{0} - I_{90}, \quad S_2 = U = I_{45} - I_{135}$$0, from which Stokes parameters are reconstructed per-pixel via interpolation ("demosaicking") and calibration (Žurauskas et al., 6 Nov 2025, Duhovic et al., 2024, Saur et al., 10 Nov 2025).

Processing steps universally include:

• Flat-field and dark-frame calibration to correct pixel gain and micro-polarizer nonuniformities.
• Demosaicking to reconstruct aligned Stokes images at full spatial resolution.
• Computation of $$S_0 = I_{0} + I_{90}, \quad S_1 = Q = I_{0} - I_{90}, \quad S_2 = U = I_{45} - I_{135}$$1, $$S_0 = I_{0} + I_{90}, \quad S_1 = Q = I_{0} - I_{90}, \quad S_2 = U = I_{45} - I_{135}$$2, $$S_0 = I_{0} + I_{90}, \quad S_1 = Q = I_{0} - I_{90}, \quad S_2 = U = I_{45} - I_{135}$$3 at each pixel.
• Masking on low $$S_0 = I_{0} + I_{90}, \quad S_1 = Q = I_{0} - I_{90}, \quad S_2 = U = I_{45} - I_{135}$$4 or low DoLP to avoid numerical instability in AoLP evaluations in low-signal or unpolarized regions.
• Smoothing or denoising, often with small Gaussian kernels, to suppress pixel-level noise before Stokes computation (Žurauskas et al., 6 Nov 2025, Duhovic et al., 2024, Saur et al., 10 Nov 2025).

Intensity-Based Polarimetric Spectroscopy

In spectral regimes (e.g., terahertz), intensity-only polarimetry is achieved using fixed polarizer/analyzer stacks or frequency-selective surfaces (FSS) at prescribed axes. Sequential or multiplexed measurement at multiple axes enables computational Stokes estimation and AoLP extraction, as demonstrated with four-band THz polarimetric imaging using a rotating PS-FSS wheel (Ahmad et al., 7 Feb 2026). In this architecture, the per-frequency AoLP is computed identically by the half-angle formula.

4. Statistical Estimation and Uncertainty Quantification

Naïve and Maximum-Likelihood Estimation

Given observed $$S_0 = I_{0} + I_{90}, \quad S_1 = Q = I_{0} - I_{90}, \quad S_2 = U = I_{45} - I_{135}$$5, $$S_0 = I_{0} + I_{90}, \quad S_1 = Q = I_{0} - I_{90}, \quad S_2 = U = I_{45} - I_{135}$$6, the basic AoLP estimator is simply

$$S_0 = I_{0} + I_{90}, \quad S_1 = Q = I_{0} - I_{90}, \quad S_2 = U = I_{45} - I_{135}$$7

For canonical noise (equal, uncorrelated variance in $$S_0 = I_{0} + I_{90}, \quad S_1 = Q = I_{0} - I_{90}, \quad S_2 = U = I_{45} - I_{135}$$8 and $$S_0 = I_{0} + I_{90}, \quad S_1 = Q = I_{0} - I_{90}, \quad S_2 = U = I_{45} - I_{135}$$9), this estimator is unbiased at high signal-to-noise ratio (SNR), but for anisotropic or correlated noise, a small bias proportional to the noise covariance appears in $\mathrm{AoLP} = \frac{1}{2}\arctan2(S_2, S_1)$0. The maximum-likelihood (ML) estimator coincides with the naïve estimator under canonical noise (Montier et al., 2014).

Bayesian Estimation and Confidence Intervals

A Bayesian approach yields a posterior in the true angle $\mathrm{AoLP} = \frac{1}{2}\arctan2(S_2, S_1)$1 given observed $\mathrm{AoLP} = \frac{1}{2}\arctan2(S_2, S_1)$2. For flat priors, the MAP estimator is again the observed angle, while credible (highest-posterior-density, HPD) intervals are computed numerically or via fitting formulae as a function of the observed SNR (Maier et al., 2014): $\mathrm{AoLP} = \frac{1}{2}\arctan2(S_2, S_1)$3 for $\mathrm{AoLP} = \frac{1}{2}\arctan2(S_2, S_1)$4 and analogous expressions for 95% and 99.7% coverage. For $\mathrm{AoLP} = \frac{1}{2}\arctan2(S_2, S_1)$5, intervals are broad and non-Gaussian, requiring careful treatment of the posterior's support.

Bias and Distribution Shape

At high SNR ($\mathrm{AoLP} = \frac{1}{2}\arctan2(S_2, S_1)$6), all estimators (naïve, ML, asymptotic, Bayesian mean/posterior) converge, and confidence intervals are symmetric and Gaussian. At low SNR, distributions become highly non-Gaussian, with significant width. The practical recommendation is to employ MAP or Bayesian mean estimation with HPD bands for low SNR, reverting to the naïve formula for $\mathrm{AoLP} = \frac{1}{2}\arctan2(S_2, S_1)$7 (Montier et al., 2014, Maier et al., 2014).

5. Physical Interpretation and Application Domains

Material and Structural Contrast

AoLP encodes the orientation of anisotropic domains—fibers, grains, layered structures—when illuminated by polarized light. In fiber orientation imaging, for example, the per-pixel AoLP in specular or diffusely reflected light provides a direct map of surface (or near-surface) fiber alignment, validated to within tens of degrees compared to micro-CT in complex composites (Duhovic et al., 2024).

In biological and medical imaging, AoLP reveals tissue anisotropy, as in the scleral collagen network of the eye, where submillimeter variations in AoLP furnish a stable subject-specific "fingerprint" for eye-tracking even under challenging occlusion or illumination conditions (Žurauskas et al., 6 Nov 2025).

Remote Sensing and Environmental Applications

In atmospheric and astrophysical contexts, AoLP, extracted from scattering or resonance line wings, serves as a probe of the orientation, strength, and filling fraction of unresolved magnetic fields or scattering geometries. In solar spectropolarimetry (e.g., Ca I 4227 Å), the MO-induced rotation of AoLP in line wings scales nearly linearly with the product of field strength and magnetic filling factor, providing crucial context unattainable from circular (Zeeman) polarization alone (Capozzi et al., 2021).

Machine Vision and Classification

Polarization-resolved descriptors, especially AoLP, enable robust classification of microstructures. In in situ microplastic identification, per-particle AoLP maps, visualized or input to convolutional neural networks, yield higher robustness to contextual noise and feature degradation than the degree of linear polarization, providing superior discrimination between morphologically similar polymer types (Saur et al., 10 Nov 2025).

In advanced imaging (e.g., SPIDeRS), the projection and recovery of structured AoLP patterns replace conventional intensity modulation for depth, normal, and reflectance estimation, leveraging specular AoLP information to associate pixels and recover surface geometry under invisible structured light (Ichikawa et al., 2023).

6. Instrumentation-Specific Processing and Calibration

Device/Method	Polarizer Axes	Calibration Steps
DoFP Encoding Cameras	0°, 45°, 90°, 135°	Flat-field, axis check, demosaicking, bias correction
THz PS-FSS Imaging	0°, 45°, 90°, 135°	Mechanical alignment, spectral calibration
SPIDeRS SLM Projector	Continuous 0°–90°	Per-pixel AoLP mapping, LC/birefringence calibration

Corrections for sensor nonidealities—including cross-talk, extinction ratio calibration, and axis misalignment—are universally required, often employing test targets with known polarization, diffuse unpolarized illumination for offset correction, and dark-frame subtraction (Duhovic et al., 2024, Žurauskas et al., 6 Nov 2025).

7. Practical Recommendations and Limitations

• Masking and thresholding: Only compute AoLP where intensity or DoLP is above relevant SNR thresholds, suppressing spurious angle maps in uninformative regions (Žurauskas et al., 6 Nov 2025, Duhovic et al., 2024).
• Noise propagation: Apply Bayesian or ML angle estimation in low-DoLP or low-SNR domains; confidently use classical estimators at high SNR (Montier et al., 2014, Maier et al., 2014).
• Material-specific contrast: AoLP is robust for discriminating materials or fibers with orientation-dependent scattering; less sensitive to purely isotropic or depolarizing surfaces (Duhovic et al., 2024, Saur et al., 10 Nov 2025).
• Depth ambiguity: AoLP provides primarily surface orientation; true 3D fiber distributions require volumetric imaging (e.g., CT) for validation (Duhovic et al., 2024).
• Conventions and coordinate reporting: Always specify the adopted convention (IAU vs. CMB/WMAP) for astrophysical applications to ensure unambiguous angle interpretation and inter-dataset consistency (Alighieri, 2016).

AoLP, as a Stokes-derived observable, remains a core quantity for optical, THz, radar, and remote sensing polarimetry, and is increasingly central in data-driven material classification, non-contact metrology, and bioimaging systems. Its canonical definition, robust statistical estimation, and versatility across modalities underscore its foundational status in polarimetric research.