Land cover classification of high-resolution remote sensing images based on improved spectral clustering DOI Creative Commons
Song Wu,

Jingmiao Cao,

Xinyu Zhao

et al.

PLoS ONE, Journal Year: 2025, Volume and Issue: 20(2), P. e0316830 - e0316830

Published: Feb. 6, 2025

Applying unsupervised classification techniques on remote sensing images enables rapid land cover classification. Using imagery from the ZY1-02D satellite’s VNIC and AHSI cameras as basis, multi-source feature information encompassing spectral, edge shape, texture features was extracted data source. The Lanczos algorithm, which determines largest eigenpairs of a high-order matrix, integrated with spectral clustering algorithm to solve for eigenvalues eigenvectors. results indicate that this method can quickly effectively classify cover. accuracy significantly improved by incorporating information, kappa coefficient reaching 0.846. Compared traditional methods, demonstrated better adaptability distribution superior performance. This suggests has strong recognition capabilities pixels complex spatial shapes, making it high-performance, approach.

Language: Английский

Land cover classification of high-resolution remote sensing images based on improved spectral clustering DOI Creative Commons
Song Wu,

Jingmiao Cao,

Xinyu Zhao

et al.

PLoS ONE, Journal Year: 2025, Volume and Issue: 20(2), P. e0316830 - e0316830

Published: Feb. 6, 2025

Applying unsupervised classification techniques on remote sensing images enables rapid land cover classification. Using imagery from the ZY1-02D satellite’s VNIC and AHSI cameras as basis, multi-source feature information encompassing spectral, edge shape, texture features was extracted data source. The Lanczos algorithm, which determines largest eigenpairs of a high-order matrix, integrated with spectral clustering algorithm to solve for eigenvalues eigenvectors. results indicate that this method can quickly effectively classify cover. accuracy significantly improved by incorporating information, kappa coefficient reaching 0.846. Compared traditional methods, demonstrated better adaptability distribution superior performance. This suggests has strong recognition capabilities pixels complex spatial shapes, making it high-performance, approach.

Language: Английский

Citations

0