Multi-channel optical gas imaging: quantitative detection of gas column concentration and temperature

Optical gas imaging (OGI) is widely used for detecting gas leaks but is limited by qualitative testing methods, lacking accurate quantitative data such as gas column concentration and temperature. For this, a multi-channel OGI (MOGI) method employing two absorption spectral channels and one non-absorption reference channel is proposed. An absorption model is developed, enabling gas transmission equation derivation, and a residual networks-based multi-task regression model is trained on simulated data. This method achieves high-accuracy predictions for four gases (NH₃, C₂H₄, C₃H₆, and C₂H₃Cl), with minimal errors in column concentration (46.32 ppm·m) and temperature (0.30 °C), significantly enhancing gas detection.