Research on Intelligent Interpretation Classification System for Multi Source Remote Sensing Big Data

The current classification systems are established from the perspective of terrain information that humans can collect on the surface, in order to meet different research purposes, research areas, research objects, or application and management requirements. The extraction of land use information contained in these distinctive classification systems mainly relies on remote sensing technology. However, due to the lack of strict connection between remote sensing technology and the current classification system and application, remote sensing technology has not been fully utilized in land use classification. Therefore, based on the multi-source remote sensing observation system’s representation of the ground in the spatiotemporal spectral dimensions and the principle of stable and classifiable of land cover, this study constructs a three-level remote sensing intelligent interpretation classification system: the first level -recognizable using "spectral characteristics ", the second level-identifiable by overlaying " temporal characteristics " or " temporal and spatial characteristics ", the third level-identifiable by overlaying "spectral physicochemical parameters". This classification system is a remote sensing intelligent interpretation product system driven by satellite remote sensing, which only includes remote sensing interpretable elements and does not mix management attributes. Taking the area of Beijing as an example, the first and second level was classified, demonstrating the feasibility of the technical approach of the classification system.