Dangerous gas leakage may cause severe consequences in industrial fields. Hence, it is significant to develop a reliable and rapid method for gas leakage recognition. Infrared imaging technology has a privilege of visibility for some invisible objects. However, infrared radiation signals of the gases are extremely weak. Therefore, it is hard to identify the gas leakage by normal infrared imaging device with wide wavelength range in the condition of atmospheric dispersion. In this research, a series of enhancement methods to detect the gas leakage with conventional wide-wavelength thermal imaging device was proposed by introducing auxiliary excitation including background enhancement and external thermal pulse excitation to improve the infrared imaging for gas leaking in the atmosphere. CO2 gas leakage was tested with FLIR-DuoR infrared camera under different auxiliary excitation modes. The experiments showed that it was impossible to recognize the leaking gases by infrared imager without any auxiliary means even there had large temperature difference between the leaking gases and background. Proper backboard did make the gas leaking visible except the white board for gas leaking at room temperature. However, the auxiliary background board with different materials demonstrated varied performance for imaging enhancement. Moreover, the auxiliary excitation with light-thermal pulse excitation also enhanced the infrared imaging for gas leakage. Finally, a leakage recognition method with machine learning model was proposed based on infrared images obtained from the experiments. The result indicated that overall prediction accuracy was above 95%. Therefore, the gas leakage recognition method based on normal infrared imaging with auxiliary excitation means and machine learning model is a potentially good tool to detect and distinguish the gas leakage. [ABSTRACT FROM AUTHOR]