To address the challenge of accurately measuring flame temperature and spectral emissivity in pulverized coal-fired boilers, this paper proposes a multispectral analysis method that simultaneously determines both parameters without requiring predefined emissivity models. In laboratory experiments, spectral data were acquired using a spectrometer and normalized across wavelength bands by combining blackbody-calibrated temperatures with spectral radiation signals. A sum of squared errors (SSE) algorithm was employed to iteratively retrieve the actual flame temperature and spectral emissivity, with validation tests conducted in a power plant boiler furnace. The results show the temperature inversion error of this method is less than 1%, meeting the accuracy requirements of industrial flame monitoring. Under constant load, increasing furnace height reduces both flame temperature and emissivity due to decreasing solid particle concentrations. At fixed heights, higher loads elevate emissivity, but an inverse relationship exists between average emissivity and temperature, driven by rapid combustion dynamics and particle density effects. Flames exhibit near-gray body behavior within the 600～940 nm wavelength range, where emissivity remains wavelength-independent. This research provides critical technical support for precise temperature and emissivity measurements, enhancing the safety, efficiency, and low-emission operation of coal-fired boilers.