Skip to Main Content
In this paper, a new approach to speckle filtering of synthetic aperture radar (SAR) images is presented, in which principal component analysis (PCA) is applied to sub-aperture images for RCS reconstruction. To describe a pixel, we define a parameter vector, the covariance of which is decomposed into two orthogonal subspaces: the signal subspace and the noise subspace. By projecting the variant part of the vector of the current pixel onto the signal subspace, the intrinsic structural features of the scene can be well obtained. Then, the RCS can be estimated. Experimental results show that our method compares favorably to several other de-speckling methods. It preserves details such as edges and small objects much better while its speckle inhibiting degree is not any worse. The effectiveness of this approach is demonstrated by using 1 m × 1 m X-band airborne SAR data.