Temperature is a key parameter of the state of the atmosphere. Temperature data play an important role in such fields as atmospheric dynamics, climatology, meteorology, and chemistry. It is also an indispensable input parameter for remote sensing inversion of other parameters. As a remote sensing instrument, lidar has been used in the detection of meteorological elements (wind, temperature, Aerosol Optical Depth, etc). And lidar techniques for the remote sensing of atmospheric temperature profiles have reached the maturity stage for routine observations. Currently, there are some types of temperature lidars, such as Raman lidar (vibration and rotation), resonant fluorescence lidar and Rayleigh scattering lidar. However, a high power laser and a complicated background filter are required for Raman lidarto ensure the accuracy of the temperature, resonance fluorescent lidar cannot detect the temperature in the stratosphere, and most of lidar based on Rayleigh scattering can only measure the relative temperature of the atmosphere. That is to say, the definition of response functions and calibration procedures is necessary for temperature retrieval. The time resolution of the method of atmospheric temperature measurement based on solid cavity scanning F-P interferometer is low. In the lower atmosphere, Rayleigh scattering spectrum of molecule is influenced by the Brillouin scattering spectrum, the superposition of two signals to form Rayleigh-Brillouin scattering spectrum, so there is a large error in temperature obtained by measuring the full width at half maximum of echo spectrum, and the particles scattering has a great influence on the retrieval results when the temperature is inverted by the integral technique. In this paper, Fizeau interferometer and PMT array are proposed to measure the molecular Rayleigh-Brillouin scattering spectrum, and the parameter optimization design of free spectral range, solid cavity length, cavity reflectivity of Fizeau interferometer and the scanning interval were carried out. And the method of reducing particle scattering effect is proposed. The information of discrete points on the RB spectral was obtained by Fizeau interferometer that the parameters was optimized, and the square method was used to get the fitting line. The temperature retrieval was achieved by comparing the theoretical spectra obtained with the 1976 U. S. standard atmospheric model and the Tenti’s S6 model. The simulation results prove that the proposed method is feasible to reduce the influence of particle scattering, and the error of atmospheric temperature between the top of the boundary layer to the top of the tropopause is less than 1K without considering the influence of cloud and wind. This temperature retrieval method can detect the absolute temperature profile with high precision and temporal resolution. There is a reference significance for the investigation of filter system of similar lidar, providing a set of feasible spectroscopic system solutions and temperature retrieval methods for our country’s ground-based and spaceborne hyperspectral thermometry lidar.