Dimensionality reduction is an important research area for hyperspectral remote sensing\nimages due to the redundancy of spectral information. Sparsity preserving projection (SPP) is a\ndimensionality reduction (DR) algorithm based on the l1-graph, which establishes the relations of\nsamples by sparse representation. However, SPP is an unsupervised algorithm that ignores the label\ninformation of samples and the objective function of SPP; instead, it only considers the reconstruction\nerror, which means that the classification effect is constrained. In order to solve this problem,\nthis paper proposes a dimensionality reduction algorithm called the supervised sparse embedded\npreserving projection (SSEPP) algorithm. SSEPP considers the manifold structure information of\nsamples and makes full use of the label information available in order to enhance the discriminative\nability of the projection subspace. While maintaining the sparse reconstruction error, the algorithm\nalso minimizes the error between samples of the same class. Experiments were performed on an\nIndian Pines hyperspectral dataset and HJ1A-HSI remote sensing images from the Zhangjiang estuary\nin Southeastern China, respectively. The results show that the proposed method effectively improves\nits classification accuracy.
Loading....