Adaptive quasiconformal kernel principle component analysis for MSTAR SAR recognition

作者：Lin Chuang^*; Sun Xirui; Wang Binghui; Kong Xiangjie

来源：Information Technology Journal, 2012, 11(12): 1751-1757.

DOI：10.3923/iti.2012.1751.1757

• Data sets
• Feature space
• Kernel principle component analysis
• Linear classification
• Maximum margin criterions
• Quasiconformal kernel
• SAR recognition

In this study, an Adaptive Quasiconformal Kernel Principle Component Analysis (AQKPCA) algorithm is proposed. We apply a quasiconformal kernel to generate a new feature space (quasiconformal feature space) to maximize the distance of each class in terms of Maximum Margin Criterion (MMC) and then, the quasiconformal kernel based KPCA and corresponding classifiers are adopted for linear classification in the feature space. The extensive experiments are conducted on MSTAR SAR dataset. The experimental results indicate the superiority of our method compare with other algorithms.

• 出版日期2012