A hybrid Fuzzy–PSO framework for across-track radiometric beam pattern extraction from multi-sector multibeam backscatter

Abstract

Radiometric correction of multibeam echosounder (MBES) backscatter is critical for quantitative seabed interpretation. MBES beam patterns are pre-calibrated but drift over time. Hiroji’s framework extracts relative beam patterns from measured data by separating seafloor angular response from system response, performing well under large roll and pitch. However, under limited roll variation, the retrieved across-track pattern suffers from discontinuities, gain jumps, and stitching errors at sector boundaries. To improve stability, we propose a hybrid Fuzzy–PSO soft-computing framework. It preserves the GRA–SRA separation principle, replaces hard-threshold outlier rejection with fuzzy reliability weighting, and replaces local common-GRA offset estimation followed by chain-wise stitching with PSO-based global alignment. In this paper, we conduct a 2 2 ablation experiment using Kongsberg EM2040 field datasets to evaluate the individual and coupled effects of the two modules and explore the applicability boundaries of the proposed framework, specifically that under severe conflicting boundary constraints, improving the dominant boundary may lead to degradation of secondary-boundary consistency. Furthermore, generalization validation was conducted on 32 datasets spanning different platforms, water depths, and sea-state conditions, and the experimental results show that the dominant-boundary consistency improvement rate surpasses 90% for low-to-medium baseline cases, while retaining about 50% positive improvement even under severe fracture scenarios. Additionally, sensitivity analysis uncovers a complementary interplay between the smoothness weight λ and the Huber threshold δ , which guarantees robust repair performance across an extensive parameter space. The framework effectively preserves intra-sector radiometric patterns without introducing noticeable statistical noise or nonphysical fluctuations. 2D backscatter imagery correction on the representative dataset demonstrates that the framework effectively suppresses systematic intensity variations while preserving textural features of seafloor morphology and geology.