Hyperspectral technology is considered a powerful tool for soil composition analysis due to its advantages of high resolution and non-destructive detection.However

the high dimensionality and redundancy of hyperspectral data increase the complexity of data processing

making feature band extraction a critical issue.This paper systematically reviews feature band extraction methods based on spectral characteristics and data-driven approaches

and analyzes the strengths and limitations of seven computational methods.Through comparative analysis of different methods’application effectiveness in soil composition inversion

several current research challenges are identified:insufficient optimization of feature band extraction methods

lack of physical interpretability in data-driven approaches

and the need for improved processing efficiency and model adaptability in hyperspectral data handling.Future development directions are proposed

suggesting that the integration of spectral feature-based and data-driven methods could potentially enhance the accuracy and stability of soil composition inversion.