Wu Renbiao

An iterative segmentation algorithm of SAR image based on support vector machine

In this paper, an iterative algorithm, which is based on support vector machine (SVM), is proposed for synthetic aperture radar (SAR) image segmentation. The proposed method considers the SAR image segmentation as the pixel classification. The pixels of the previous segmented image are regarded as the training samples for SVM, which is used to re-segment the image. These iterations are repeated until the convergence, which is determined by checking the relative change of the entropy between two consecutive segmented images. Experimental results show that compared with, the proposed algorithm can achieve much better segmented results than the Markov random field (MRF) algorithm, and the proposed method dramatically reduces the influence of initial segmentation on the final result.

Multilayered diffraction tomography algorithm for ground penetrating radar

A multilayered diffraction tomography algorithm for ground penetrating radar is derived in this paper. The algorithm considers the planar air-soil interface and uses the thought of the equivalent speed to solve the problem of multilayered medium. The first Born approximation and the dyadic Greens function are adopted to construct a linear forward model that inverted by fast Fourier transform to meet the requirement of small calculation, so we can real-time reconstruct the buried objects. Simulation results are provided to illustrate performance of this algorithm.