Mineral Element Identification in Remote Sensing Imagery: A Fusion Approach Using CH-Tucker Decomposition and RFDNet

Abstract

In the realm of geological and mineral exploration, remote sensing technology has emerged as a pivotal high-tech instrument. However, the effective interpretation of remote sensing images, especially in the context of heterogeneous data processing, noise, and the identification of fine granularity, remains a challenge. In this study, a novel method for the identification of mineral elements within remote sensing imagery was introduced. Firstly, a heterogeneous feature tensor migration technique anchored on the Coupled Heterogeneous Tucker Decomposition (CH-Tucker decomposition) was presented. Through this technique, multi-source remote sensing data were effectively processed and fused. Notably, associated data features from varying resolutions and angles were seamlessly coupled. Subsequently, an optical remote sensing image processing model founded on the RFDNet network was established. This model demonstrated robustness against noise data, thereby enabling the identification of mineral elements with a higher degree of granularity. The proposed methodology exhibited the capacity to extract mineral element information comprehensively and with remarkable accuracy. Thus, this research offers both valuable theoretical insights and practical evidence for furtherance in geological research and mineral element exploration.