Radiative Constraint-Based Scale Inversion for High-Altitude Aircraft From Three-Band Thermal Infrared Imagery

Read full text Summarise with VIRAT
Authors: Jianing Yu, Kefang Wang, Xiaoxuan Zhou, Zhuoyue Hu, Lixing Zhao, Xinyue Ni, Wenhao Pan, Xiaoyan Li, Fansheng Chen
Cited by: 0
References: 43

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Quantitative retrieval of high-altitude aircraft projected area from spaceborne long-wave infrared (LWIR) imagery is challenged by sub-pixel mixing, low thermal contrast, and atmospheric interference. This paper presents a radiatively constrained, geometry-aware inversion framework to estimate the pixel-coverage ratio and projected area from 30 m ground sampling distance tri-band LWIR data. The method synergistically integrates ADS-B flight data, ERA5 reanalysis, and MODTRAN radiative transfer modeling coupled with an aircraft skin heat-balance model. The processing pipeline incorporates scale-invariant segmentation, spectral-consistency filtering, certainty-weighted fusion, and viewing-geometry correction. Validation on a held-out dataset of 563 real observations across 57 aircraft types demonstrates robust performance: 84% of retrievals achieve ≤20% relative error (52% ≤10%), with an overall RMSE of 97.7 m² and MRE of 13.3%. Accuracy is high and stable within a defined operational domain (e.g., signal-to-clutter ratio ≥30). The technique advances passive TIR sensing from detection to quantitative scale estimation, providing valuable capability for situational awareness and cross-sensor cueing, particularly in covert or night-time scenarios where active sensors may be limited.

Fields of Study & Subjects

Infrared Target Detection Methodologies Advanced Neural Network Applications Remote-Sensing Image Classification

Publication details

Journal / source
IEEE Transactions on Geoscience and Remote Sensing
Volume
64
Pages
1–16
Publication date
1 January 2026
Language
EN