Hyperspectral image clustering method, device and product

The invention discloses a hyperspectral image clustering method, device and product, and relates to the field of hyperspectral image clustering. The hyperspectral image clustering method comprises the following steps: converting an obtained hyperspectral image into superpixels, and constructing an undirected graph based on the superpixels; de-noising the original node features of the undirected graph by using a low-pass graph filter to obtain smooth node features; a parameter non-shared MLP encoder is adopted to encode smooth node features, comparison graph node embedding feature learning is carried out, and two corresponding enhanced views are obtained; fusing the two enhanced views in a linear mode to obtain clustering-oriented node features; and clustering the clustering-oriented node features by using a K-means clustering method to obtain a clustering result. According to the invention, the clustering efficiency of the hyperspectral image can be improved. 本发明公开一种高光谱图像聚类方法、设备及产品，涉及高光谱图像聚类领域。所述高光谱图像聚类方法包括将获