SAR target recognition based on improved sparse LSSVM

Xiangrong Zhang, Yifan Zhang, Licheng JiaoProceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE 2009 · DOI

Abstract

An Improved Fast Sparse Least Squares Support Vector Machine (IFSLSSVM) is proposed for Synthetic Aperture Radar (SAR) target recognition. Least Squares Support Vector Machine (LSSVM) is a least square version of Support Vector Machine (SVM), but it lacks the sparseness compared with SVM. IFSLSSVM, which combines the incremental learning and decremental learning, selects those important samples as the support vectors, and implements pruning by a certain condition, can solve the non-sparse problem of LSSVM effectively. Benchmarking UCI datasets are firstly used for testing the performance of our algorithm, followed by SAR target recognition. Experimental results on MSTAR SAR dataset show that IFSLSSVM is an effective SAR target recognition approach (SAR-ATR), which not only reduces the number of support vectors but also enhances the recognition rate.
BibTeX
@inproceedings{zhang2009sar,
  author = {Xiangrong Zhang and Yifan Zhang and Licheng Jiao},
  title = {{SAR target recognition based on improved sparse LSSVM}},
  booktitle = {Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE},
  year = {2009},
  doi = {10.1117/12.832466},
}

Loading related content...

Figure