Abstract: Accurate interpretation of open-pit mining areas is crucial for the rational planning of mineral resources, dynamic environmental monitoring, and the effective regulation of illegal mining. However, due to the inherent complexity and high variability of mine scenes, models relying exclusively on limited single-view remote sensing data often struggle to yield satisfactory recognition performance. To address this limitation, this paper proposes a novel remote sensing classification approach for mine areas driven by the feature fusion of conventional scenes and natural mine images. This approach synergistically leverages conventional remote sensing datasets, natural mine images, and optical mine remote sensing imagery to boost classification accuracy in complex mining environments. The model is developed upon a two-stage training paradigm. In the first stage, the backbone network is pre-trained on a large-scale conventional remote sensing dataset. This allows the network to acquire generalized surface priors, endowing it with robust foundational capabilities for spatial feature extraction. In the second stage, the parameters of the backbone network are frozen, shifting the focus toward feature enhancement and target scene classification. Specifically, a feature distillation module is incorporated concurrently with the supervised scene classification of the optical mine remote sensing images. Treating natural mine images as auxiliary cues, this distillation mechanism directs the network to extract rich fine-grained textures and facade visual details, which are subsequently transferred into the remote sensing feature space. This feature-level distillation and alignment significantly bolster the scene representational capacity of the remote sensing data. Extensive experiments reveal that the proposed two-stage model successfully bridges the semantic gap across different viewpoints, yielding a substantial improvement in classification accuracy. Consequently, this study offers a promising technical framework for the intelligent monitoring of open-pit mines.