Polarimetric Scattering-Structure Contrastive Learning for Few-Shot HRRP Recognition

Fangyi Wang; Peng Zhao; Haipeng Wang

cs.CVNov 3, 2026

Polarimetric Scattering-Structure Contrastive Learning for Few-Shot HRRP Recognition

Authors: Fangyi Wang, Peng Zhao, Haipeng Wang

Organizations: Fudan University, China

Abstract

Few-shot high-resolution range profile (HRRP) recognition is challenged by strong aspect-dependent variation and sparse novel-class support. We propose polarimetric scattering-structure contrastive learning (PSS-CL), which supplies an episodic HRRP encoder with a structured physical training view. Dominant local responses are represented by their soft affinities to canonical polarimetric mechanisms, normalized range locations, and energy contributions. A symmetric cross-view objective transfers this scattering structure to the learned HRRP representation, while episodic prototype learning supports novel-class inference without the physical branch. We conduct Random-aspect and Cross-aspect experiments on a six-class measured dataset. In Cross-aspect 1-, 3-, and 5-shot recognition, PSS-CL achieves 81.73%, 86.92%, and 90.98% mean class accuracy (MCA), respectively, outperforming the best competing results by 5.62%, 3.09%, and 3.56%.

Figures & tables

Figure 1: HRRP formation and the Cross-aspect few-shot recognition challenge. Dominant scattering centers form a one-dimensional range profile, while aspect-dependent scattering causes support–query prototype mismatch.

Figure 2: Overview of PSS-CL. The structured physical branch provides training-only cross-view supervision for episodic HRRP learning.

Figure 3: Six canonical polarimetric scattering-response families used in the Cameron implementation.

General few-shot learning
Method	Random-aspect			Cross-aspect
Method	1-shot	3-shot	5-shot	1-shot	3-shot	5-shot
Matching Nets [ 24 ]	72.42	79.16	84.28	60.83	67.76	70.69
ProtoNet [ 22 ]	76.28	82.36	84.55	63.52	72.32	73.85
RelationNet [ 23 ]	74.76	81.55	85.39	63.19	70.93	75.46
MAML [ 9 ]	70.35	74.16	76.60	58.70	63.98	67.25

Table 1: MCA (%) on the measured HRRP dataset.

Figure 4: Cross-aspect confusion matrices for the fishing–tanker–reflector fold: (a) 1-shot; (b) 3-shot; (c) 5-shot.

Configuration	$\mathcal{L}_{\mathrm{sc}}$	$\mathcal{L}_{\mathrm{ca}}$	View	1-shot	3-shot	5-shot
Episodic baseline	–	–	–	64.50	72.39	76.82
Scattering alignment	✓	–	$\bar{\mathbf{g}}_{i}$	75.14	81.75	85.51
Aspect regularizer	–	✓	–	71.76	78.49	82.61
Global scattering view	✓	✓	$\bar{\mathbf{g}}_{i}$	78.33	84.49	87.68
PSS-CL (Ours)	✓	✓	$\mathbf{g}_{i}$	81.73	86.92	90.98

Table 2: Ablation results under Cross-aspect evaluation (MCA, %).

Figure 5: Post-selection Cross-aspect 1-shot sensitivity to (a)

R

and

\lambda_{\mathrm{sc}}

; (b)

\tau_{\mathrm{ca}}

and

\eta

Few-Shot Learning

Few-shot learning develops methods that recognize or adapt to new tasks from only a handful of labeled examples.

See all 218 papers

P-SER: Physical Consistency-Constrained Scattering Expandable Representation for Class-Incremental HRRP Target Recognition
Nov 3, 2026Fangyi Wang, Peng Zhao, Haipeng WangClass-Incremental Learning SAR ATR
Key Laboratory of Information Science of Electromagnetic Waves (MOE), and the Discipline and Technology Center of Microwave Vision Intelligent Sensing, Fudan University, Shanghai 200433, China
High-Resolution Range Profile Classifiers Require Aspect-Angle Awareness
Feb 16, 2026Edwyn Brient, Santiago Velasco-Forero, Rami KassabClassification
CMM · ARC, Thales Land and Air Systems Limours, France
Cross-polarization pairing for label-free few-shot thin-section classification
Oct 17, 2026Bingchen YanFew-Shot Image Classification Microscopy Image Analysis
School of Computer Science and Information Security, Guilin University of Electronic Technology, Guilin, 541004, China

Nov 3, 2026J/K move · Enter open · S save

P-SER: Physical Consistency-Constrained Scattering Expandable Representation for Class-Incremental HRRP Target Recognition

Fangyi Wang, Peng Zhao, Haipeng Wang

Key Laboratory of Information Science of Electromagnetic Waves (MOE), and the Discipline and Technology Center of Microwave Vision Intelligent Sensing, Fudan University, Shanghai 200433, China

Class-Incremental Learning SAR ATR Physics-Informed ML

Open paper AskPDF arXiv

High-resolution range profiles (HRRPs) provide compact radar target representations but exhibit strong aspect sensitivity, complicating class-incremental recognition when historical data are unavailable. Changes in viewing aspect shift scattering-peak locations and amplitudes, while reflector groups may produce responses resembling ships. Existing dynamic expansion methods can also incur substantial parameter and memory overhead. To address these challenges, we propose P-SER, a Physical Consistency-Constrained Scattering Expandable Representation framework for exemplar-free class-incremental HRRP recognition. P-SER reconstructs dominant scattering points from complex-valued sum- and azimuth-difference-channel HRRPs and observation geometry. The reconstructed spatial structure, target extent, and energy distribution are used to derive class-level scattering prototypes and sample-level saliency priors, providing physically grounded constraints for feature learning and old-class knowledge retention. A primitive scatterer dictionary further characterizes local strong-scattering mechanisms and calibrates uncertain ship--reflector predictions at inference. Meanwhile, lightweight residual adapters are incrementally introduced into a shared HRRP backbone to learn new classes with limited model growth. Experiments on simulated and measured HRRP datasets show that P-SER delivers competitive recognition performance without retaining old-class exemplars, while maintaining physical consistency, interference robustness, knowledge retention, and model scalability.

Abstract

Few-Shot Learning

Explore similar work

P-SER: Physical Consistency-Constrained Scattering Expandable Representation for Class-Incremental HRRP Target Recognition

High-Resolution Range Profile Classifiers Require Aspect-Angle Awareness

Cross-polarization pairing for label-free few-shot thin-section classification