J4 ›› 2015, Vol. 42 ›› Issue (3): 15-21+89.doi: 10.3969/j.issn.1001-2400.2015.03.003

• Original Articles • Previous Articles     Next Articles

Novel method for attributed feature extraction from SAR imagery

LI Fei;JIU Bo;WANG Yinghua;LUO Tao;LIU Hongwei   

  1. (National Key Lab. of Radar Signal Processing, Xidian Univ., Xi'an  710071, China)
  • Received:2014-01-02 Online:2015-06-20 Published:2015-07-27
  • Contact: LI Fei E-mail:xdlifei@gmail.com

Abstract:

Aiming to extract attributed features of overlapped attributed scattering centers, a new method for attributed feature extraction of the SAR target is proposed based on the fully polarimetric attributed scattering center model by considering the sparsity of the scattering center in the space domain and scattering mechanism domain. According to this sparsity, the row sparse constraint and matrix sparse constraint are imposed on the polarimetric decomposition coefficient matrix of the target, respectively. Since the polarimetric scattering mechanism dictionary contains an unknown parameter, the coordinate decent technique is employed to optimize the polarimetric decomposition coefficient matrix and polarimetric scattering mechanism dictionary for attributed feature extraction (attributed scattering center and polarimetric signature). Numerical results on electromagnetic computation data verify the validity of the proposed algorithm.

Key words: synthetic aperture radar automatic target recognition, polarimetry, feature extraction, sparse

CLC Number: 

  • TN957