Skip to main navigation Skip to search Skip to main content

Reconstructing 3D Flame Temperature Fields via Light Field Compression, Denoising and Feature Extraction

  • Qingran Wang*
  • , Pengfei Gao
  • , Yatao Ren
  • , Hong Qi
  • *Corresponding author for this work
  • School of Energy Science and Engineering, Harbin Institute of Technology

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Accurate 3D temperature reconstruction is vital for combustion diagnostics under noisy conditions. This study develops a light field imaging framework incorporating a radiative transfer model and a Light Field Compression and Noise Reduction (LFCNR) method based on low-rank approximation and eigenvalue decomposition. The LFCNR algorithm separates signal and noise subspaces and applies Singular Value Decomposition (SVD) for structural compression, reducing computational complexity. Comparative results show that the customized 1D denoising with SVD truncation achieves superior accuracy (ARE = 2.8770%, NRMSE = 0.0408, SSIM = 0.9288), outperforming wavelet thresholding method (ARE = 2.9920%, NRMSE = 0.0429, SSIM = 0.9230). The framework ensures robust, high-resolution temperature reconstruction in complex combustion fields.

Original languageEnglish
Title of host publicationProceeding of 2025 IEEE 4th International Conference on Computing, Communication, Perception and Quantum Technology, CCPQT 2025
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9798331525835
DOIs
StatePublished - 2025
Externally publishedYes
Event4th IEEE International Conference on Computing, Communication, Perception and Quantum Technology, CCPQT 2025 - Ordos, China
Duration: 24 Oct 202526 Oct 2025

Publication series

NameProceeding of 2025 IEEE 4th International Conference on Computing, Communication, Perception and Quantum Technology, CCPQT 2025

Conference

Conference4th IEEE International Conference on Computing, Communication, Perception and Quantum Technology, CCPQT 2025
Country/TerritoryChina
CityOrdos
Period24/10/2526/10/25

Keywords

  • Combustion diagnostics
  • Light field imaging
  • Low-rank approximation
  • Temperature tomography

Fingerprint

Dive into the research topics of 'Reconstructing 3D Flame Temperature Fields via Light Field Compression, Denoising and Feature Extraction'. Together they form a unique fingerprint.

Cite this