Infrared thermography (IRT) is a new type of digital nondestructive testing technology that has developed rapidly and has found wide use. However, it still has some shortcomings, such as low accuracy, image noise, and requires a large amount of data. In this study, the phase change processes of paraffin and water in porous materials are characterized by IRT technology at pore scale, and the temperature measurement accuracy and phase interface characterization are optimized. First, the reflective surface (wrinkled aluminum foil) is added to obtain the environmental reflective temperature in the test process to correct the target object temperature; subsequently, guided filtering and principal component thermography (PCT) are used to reduce the noise of IRT images to improve the detection accuracy of the infrared thermal imager for temperature field and phase interface. The experimental results show that the temperature measured by the environmental reflection temperature correction method is closer to that measured by a thermocouple after eliminating the influence of changing environmental temperature; after selecting an appropriate filter radius and filter parameters, the contour of phase interface is clearer, and a better noise reduction effect is obtained. In addition to the good noise reduction effect, PCT makes the phase interface clearer and reduces the amount of processed data by four orders of magnitude. These three processing methods provide better optimization methods and theoretical support for infrared thermal imaging technology to characterize the phase transition process of porous media at the pore scale.