Raman spectroscopy, with its advantages of simplicity, rapidity, repeatability, and non-destructiveness, has been widely applied in various fields. When utilizing Raman spectroscopy for natural gas component analysis, the introduction of noise signals and fluorescence during the Raman spectrum acquisition process is inevitable. Noise and fluorescence can significantly affect the accuracy of natural gas component analysis results. Therefore, this study proposes a new segmented baseline correction method and presents an iterative S-G (Savitzky-Golay) smoothing-based baseline interval recognition method. The study achieves spectral segmentation and determines the optimal sliding window length, polynomial order, and iteration step through experimental methods, using mean square error as the evaluation metric. A comparative experiment of baseline correction for three simulated signals is conducted using the proposed method and iterative polynomial fitting method. The experimental results demonstrate the accurate spectral segmentation of the proposed method, with mean square errors of 1.6138E-4, 1.1785E-4, and 1.9469E-4 for the three signals after correction, indicating excellent baseline correction performance.