A Hierarchy System for Automatic Target Recognition in SAR Images
A novel SAR image automatic target hierarchy recognition (ATHR) system based on SVM and D-S evidence theory is proposed in this chapter. This system has three hierarchies corresponding to three features. PCA, LDA and NMF features are extracted from images without preprocessing, and are fed to SVM classifier. However, not all features are used in each recognition process. At each recognition process, a threshold is used to determine the used features and hierarchy depth. Experiments on MSTAR public data set demonstrate that the proposed system outperforms the system combining the outputs of three features directly.
KeywordsSAR ATR Hierarchy recognition SVM D-S evidence theory
This work is supported by Fundamental Research Funds for the Central Universities under Projects ZYGX2009Z005.
- 1.Novak LM, Owirka GJ, Weaver AL (1999) Automatic target recognition using enhanced resolution SAR data. IEEE Trans AES 35(1):157–175Google Scholar
- 2.Kaplan ML (2001) Analysis of multiplicative apeckle models for template- based SAR ATR. IEEE Trans AES 31(4):1424–1432Google Scholar
- 3.O’Sullivan JA, Devore MD (2001) SAR ATR performance using a conditionally Gaussian model. IEEE Trans AES 37(1):91–108Google Scholar
- 4.Bhanu B, Yingqiang Lin (2000) Recognition of occluded targets using stochastic models. In: Proceedings of IEEE workshop on computer vision beyond the visible spectrum: method and applications, Singapore, pp 73–82Google Scholar
- 5.Gong Cheng, Wei Zhao, Jinping Zhang (2006) A practical kernel criterion for feature extraction and recognition of MSTAR SAR images. In: Proceedings of ICIP, Singapore, vol 4Google Scholar
- 6.Changzhen Qiu, Hao Ren, Huanxin Zou (2009) Performance comparison of target classification in SAR images based on PCA and 2D-PCA features. In: Proceedings of 2nd APSAR, Singapore, pp 868–871Google Scholar
- 8.Xin Yu, Yukuan Li, LC Jiao (2011) SAR target recognition based on classifiers fusion. In: Proceedings of M2RSM, Singapore, pp 1–5Google Scholar