Abstract: Mine ecological restoration is an essential guarantee for sustainable mineral resource development, and vegetation biomass serves as a key indicator to evaluate the effectiveness of ecological restoration. Traditional biomass estimation relies on field survey data, which suffers from limitations such as high time costs and labor intensiveness. Multi-spectral and LiDAR sensors mounted on drones, combined with hand-held LiDAR scanning, were used to collect parameters of evergreen and deciduous trees at the Pingshu Coal Mine in Shouyang County, Jinzhong City, Shanxi Province. By applying Pearson correlation analysis to select model variables, above-ground biomass models for evergreen and deciduous trees were developed at the individual tree scale using multiple linear stepwise regression and random forest methods. Results show that the random forest-based above-ground biomass model achieved the highest accuracy. Specifically, the evergreen tree model yielded an R² of 0.78 and an RMSE of 11.043 kg/tree, while the deciduous tree model resulted in an R² of 0.74 and an RMSE of 33.29 kg/tree. The study also applied the Maximum Likelihood Classification (MLC) algorithm to multispectral imagery with feature combinations for tree species identification. Results indicate that incorporating the near-infrared band, red-edge band, and the Normalized Difference Vegetation Index (NDVI) significantly improved the accuracy of tree species identification. Additionally, the random forest-based single-tree above-ground biomass model, combined with the watershed segmentation algorithm for canopy area estimation, was used to calculate single-tree biomass density. Subsequently, an above-ground biomass inversion of the study area was performed on a per-pixel basis according to land categories. The study provides a reference for monitoring and analyzing above-ground biomass in mining areas and offers data support for quantitatively evaluating the effectiveness of ecological restoration and environmental protection in mining regions.