Recently, a collaborative effort between Professor Shi Shengxian's research group at Shanghai Jiao Tong University, Researcher Zeng Fei from the Hunan Power Machinery Research Institute (affiliated with the China National Aviation Development Engine Corporation), and Professor Tsai Dinping of the City University of Hong Kong has led to the development of a light-field hyperspectral radiation thermometry technique grounded in dispersive metalenses (DMT). The research findings have been published in Nature Communications. This innovative technology harnesses the power of metalenses to precisely manipulate light fields on a microscopic level, effectively segregating thermal radiation signals into their broadband components and encoding them into compressed, dispersive images. By integrating conventional convex optimization with advanced deep learning methodologies, the technique enables the reconstruction of temperature data from a minimum of 21 spectral channels using just a single image. Within the temperature spectrum spanning from 1673K to 2973K, the measurement inaccuracy remains below 0.32%, marking a nearly sixfold enhancement in precision compared to traditional approaches. During experimental trials, the DMT system exhibited measurement errors consistently below 2% across three distinct types of high-temperature ceramic materials—namely, alumina, zirconia, and silicon carbide, which possess emissivities ranging from 0.2 to 0.9—even in the presence of interference from high-temperature gases. This underscores the system's exceptional robustness and accuracy. The research paves the way for the creation of compact, highly integrated radiation thermometers, with promising applications in temperature monitoring and combustion diagnostics for high-temperature components, including aero-engine/gas turbine blades and combustion chambers.