Research On Technology Of Chemical Gas Detection With Infrared Hyperspectral Images

The impact on the human being and nature caused by gas disaster has beengradually increased, due to the physical and chemical properties of gas; theconventional detection method is difficult to effective implementation for thepurpose of gas detection. Long-wave infrared(LWIR) hyperspectral remote sensingas one of the most effective means of the gas disaster detection can provide thetemperature and spectral information, is of great value in military and civilian. Thisdissertation starts from the study the characteristics of LWIR hyperspectral data, wewill achieve the infrared hyperspectral gas data modeling and detection based on thegas radiative transfer imaging model, to provide help and guidance for the buildinggas disaster monitoring system. This thesis mainly work on the following three partsto achieve above goals:Firstly, this paper expounds the four basic laws of thermal infrared radiation,and focus on the analysis of these laws, and summarize some rules and inference ofthe infrared hyperspectral imaging, and then analyzes the radiation effect, scatteringeffect and absorption effect on the radiative transfer imaging process based on theimpact on the radiative transfer by atmospheric composition, and analyzes theinfrared radiation imaging model and the hyperspectral imaging model.Secondly, we research on the technology of LWIR hyperspectral gas imagingmodel and simulation. The temperature difference of the gas and the environmentaround is the foundation of infrared detection, and the spectral absorption of energyby gas is the basis of the spectrum characteristic imaging. Based on theLambert-Beer’s law, this paper researches on the effect on the spectral radiationtransmission by gas, and analyzes infrared radiative transfer model with gas target,and based on this, we build the gas spectral library,the atmospheric environmentsimulation,and the temperature distribution model, and finally through the analysisof the basic law of infrared radiation, we eventually complete the long-waveinfrared hyperspectral image simulation with pure gas and mixed gas in8μm to14μm.Finally, this paper researches on the gas detection applications of the LWIRhyperspectral image. Based on the fundamental laws of infrared radiation, weconstruct the single linear detection model and the linear mixed detection model bythe analysis on the gas signal detection model, and detect the single gas infraredhyperspectral simulation data by matched filter, adaptive coherence/cosine estimatorand adaptive matched filter. Experimental results show that the traditional hyperspectral detection methods are suitable for the gas data analysis and detection,and verify the robustness to noise, spectral resolution and atmospheric sceneestimation error. Finally we realize the detection of mixed gas and abundanceextraction of the LWIR hyperspectral simulation mixed gas data.