Self-Supervised Topologically Invariant Manifold Learning for Railway Image Quality Assessment

2026-08-15

AI Summary

This study addresses the limited generalization of existing blind image quality assessment methods in real-world scenarios by proposing a self-supervised topologically invariant manifold learning framework. The approach introduces a self-constrained quality manifold and an elite evaluator pool, integrating progressive background dilution and monotonic divergence filtering to establish stable quality references for cross-paradigm decoupling. Experimental results demonstrate that this framework achieves superior zero-shot transfer performance across multiple benchmarks without requiring labeled data. Notably, it attains a manifold similarity exceeding 0.999 on railway datasets and maintains a 100% survival rate under extreme stress conditions. These findings indicate significant improvements in both topological robustness and practical utility for unlabeled image quality assessment, effectively overcoming domain adaptation challenges through intrinsic geometric consistency rather than supervised fine-tuning.

Abstract

Existing blind image quality assessment (BIQA) methods typically rely on synthetic distortions and subjective annotations, limiting generalization in real-world domains. To address this, we propose a fully self-supervised BIQA framework based on topologically invariant manifold learning under boundary constraints, which constructs a stable quality reference without manual labels. The framework generates progressive background dilution scales via repeated random cropping around each target; exploiting the monotonic degradation of target information density across these scales, it establishes a self-constrained quality manifold. A linearized spatial moment projection eliminates geometric distortions from random cropping; then a monotonicity divergence filter prunes background-sensitive evaluators, isolating an elite pool \(\mathcal{M}_{\text{elite}}\). A robust M-estimator with a principal component stabilizer fuses the metrics into an asymptotically efficient pseudo-ground truth \(q_{\text{PGT}}\), contracting variance toward the Cramér-Rao lower bound. Extensive evaluations demonstrate that the elite evaluator pool, distilled from 11 baseline metrics, secures superior zero-shot transferability across standard synthetic and wild benchmarks (CSIQ, LIVEC, LIVE-2). Concurrently, deployments on the CQU Railway Rolling Stock Surveillance Dataset (2,797 images) yield a manifold cosine similarity \(>0.999\) and a 100.0\% survival rate under industrial extreme stresses, robustly validating its cross-paradigm decoupling and topological resilience.

Problem

• Blind Image Quality Assessment
• Generalization
• Real-world Domains
• Self-supervised Learning

Innovation

• Self-Supervised BIQA
• Topologically Invariant Manifold Learning
• Pseudo-Ground Truth
• Zero-Shot Transferability
• Railway Image Quality Assessment