Knowledge graph embedding spectral unmixing

Selecting effective endmembers from a set of endmembers is important in the process of spectral unmixing. However, the selection of endmembers will be affected by the spectral variability of endmembers, which results in a certain uncertainty in the results of selection and the accuracy of unmixing. This study combines geoscience prior knowledge with sparse unmixing to solve this problem, and a Knowledge Graph Embedding Spectral Unmixing (KGESU) algorithm is proposed. While utilizing spectral features, certain prior knowledge is introduced to further improve the reliability of endmember selection. The implementation steps of the KGESU algorithm involve two issues the embedding training of geoscience knowledge graph and spectral unmixing with priori knowledge. The embedding training of geoscience knowledge graph transforms geoscience knowledge into a structured expression form through knowledge graph. Then, the TransE model is used for graph embedding. We perform knowledge reasoning according to the knowledge graph embedding to address the second issue. Then, a reasoning – weighting sparse unmixing algorithm is developed to integrate the process of reasoning and unmixing. Experiments are conducted to validate the effectiveness of the proposed method. The prior knowledge is instantiated with the aid of auxiliary data such as Landsat 8 and GDEMV2. The spectral unmixing data are GF-5 satellite data. The GF-2 data with a resolution of 1 m after graphic fusion are used for verification. Compared with the traditional pixel-by-pixel evaluation, this study expands the evaluation window. The sensitivity of different resolution images to registration errors is reduced by increasing the overlap area between pixels and allocating the residuals. The root mean square error of each endmember, the mean of the root mean square error of each endmember, and the overall root mean square error of the image are used as evaluation indexes to evaluate the unmixing results. Results demonstrate that the KGESU algorithm outperforms the state-of-the-art algorithms. By the guidance of geo-prior knowledge in the unmixing process, the uncertainty caused by factors such as data itself and external noise can be reduced. The ability to discriminate endmembers can be improved to a certain extent. At the same time, the method proposed in this study combines the advantages of knowledge reasoning and numerical computation. Furthermore, we use geoscience knowledge and spectral characteristics to select endmembers. The unmixing result can be more reliable. In the future, the research has the following issues that need further consideration. (1) In this study, a knowledge graph is constructed only from the perspective of land use classification, and prior knowledge is introduced. In the follow-up work, secondary and even more precise classification can be considered to highlight the advantages of hyperspectral data. (2) In the future work, we will consider more complex relationships between ground objects, introduce more abundant geoscience knowledge, and further build a more perfect geoscience knowledge graph. (3) Knowledge reasoning based on graph embedding is a relatively good method to integrate reasoning results into spectral unmixing at present. With the continuous development of technology, we will further attempt to introduce knowledge through other knowledge reasoning mechanisms. © 2024 Science Press. All rights reserved.