Non-contact respiratory signal extraction method with anti-interference capability

By detecting the nasal ROI in thermal infrared image sequences and combining it with a two-stage signal completion and reconstruction strategy, the problem of missing ROI in thermal infrared image respiration detection under motion and environmental changes is solved, and high-quality respiration signal recovery and feature extraction are achieved.

CN121301882BActive Publication Date: 2026-06-02HEFEI UNIV OF TECH
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Patents(China)
Current Assignee / Owner
HEFEI UNIV OF TECH
Filing Date
2025-09-30
Publication Date
2026-06-02

AI Technical Summary

Technical Problem

Existing non-contact respiration detection technologies based on thermal infrared images suffer from motion sensitivity defects in ROI detection when faced with motion and environmental changes. They also lack explicit modeling and compensation mechanisms for missing ROI states, which leads to interference with respiration detection results.

Method used

A thermal infrared nasal ROI detection method based on single-frame missing judgment is adopted, combined with a two-stage signal completion and reconstruction strategy, including master frequency reconstruction and respiratory generative adversarial network BreathGAN, to process mixed respiratory signal sequences to obtain high-quality reconstructed respiratory signals.

Benefits of technology

It effectively completes and reduces the noise of respiratory signals, ensuring the continuity and integrity of the signals, and improving the accuracy of respiratory signal feature extraction, providing accurate and reliable data support for health analysis and physiological monitoring.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121301882B_ABST
    Figure CN121301882B_ABST
Patent Text Reader

Abstract

The application provides a non-contact respiratory signal extraction method with anti-interference capability, and relates to the field of respiratory signal extraction.In the application, a thermal infrared nose ROI detection method is used to traverse a thermal infrared image sequence, and a mixed respiratory signal sequence containing effective respiratory signals and missing intervals is obtained through frame-by-frame missing judgment.On this basis, a two-stage signal completion reconstruction strategy is proposed, a signal completion method based on main frequency reconstruction is designed in the first stage, different length missing intervals are filled and completed in a targeted manner, preliminary completion and reconstruction are realized, in the second stage, a respiratory generative adversarial network (BreathGAN) is designed, noise reduction of the signal is further carried out under time-frequency domain fusion, more accurate filling and reconstruction are completed, and finally a high-quality reconstructed respiratory signal sequence is recovered.In addition, the application also carries out multi-dimensional feature analysis on the recovered high-quality reconstructed signal, and scientifically measures the quality of the extracted respiratory signal through direct and indirect ways.
Need to check novelty before this filing date? Find Prior Art