Infrared (IR) technologies have been universally acknowledged as a valuable pedagogical tool for exploring novel and abstract scientific subjects in science education. This study explores the roles of IR images played in middle school students' Evidence-based Reasoning (EBR) process in support of the understanding of the heat radiation process. Specifically, we implement image processing algorithms explicitly for the visual artifacts mentioned in students' descriptions of the radiation phenomenon to obtain the numeric representations of their corresponding features. Meanwhile, the quality of those descriptions is further coded with the guidance of the EBR framework for indicating students' understanding levels of the phenomenon. Finally, the associations between the numerical image features and the quality of descriptions are analyzed to examine the effectiveness of the IR visual artifacts in helping students understand the heat radiation process. The analytical results found that the image features are further positively correlated with the quality of the descriptions generated by students for the heat radiation. The results further suggest the IR images have the potential of driving students to think proactively and explore detailed procedural changes in learning the heat radiation process. Finally, our study calls for the integration of interdisciplinary instructional approaches in science education to reduce students' cognitive load and guide learning attention, for example, incorporating visualization and relevant processing approaches to present and analyze the otherwise invisible abstract process to help students make sense related knowledge more easily.