研究目的
To propose a novel local tangent space alignment (LTSA) variant, kernel extended (KE)-LTSA for synthetic aperture radar (SAR) image classification, aiming to extract local geometric structures and maximize global interclass separability.
研究成果
KE-LTSA greatly outperforms LLTSA and other traditional methods including PCA, LDA, and KPCA. It shows meaningful improvement compared with two newly proposed methods and demonstrates high robustness with respect to both target signature variability and neighbourhood size selection.
研究不足
The technical and application constraints of the experiments, as well as potential areas for optimization, are not explicitly mentioned in the abstract.
1：Experimental Design and Method Selection:
The study proposes KE-LTSA, a kernel version of linear extended local tangent space alignment (LE-LTSA), for SAR image classification. It combines local geometric structures and global interclass separability.
2：Sample Selection and Data Sources:
SAR images from the MSTAR public database are used, including ten targets.
3：List of Experimental Equipment and Materials:
Not explicitly mentioned.
4：Experimental Procedures and Operational Workflow:
The algorithm involves determining the kernel function, computing local coordinates, constructing global coordinates, and feature transformation.
5：Data Analysis Methods:
The performance of KE-LTSA is compared with PCA, LDA, KPCA, LLTSA, L1/2-NMF, and another method using 1-NN classifier, SVM, and 5-NN classifier.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
