The authors propose an accurate and robust method to reconstruct buried objects for half-space scenarios. First, the linear sampling method (lsm) was employed to estimate targets support. A relaxed threshold is define...
详细信息
The authors propose an accurate and robust method to reconstruct buried objects for half-space scenarios. First, the linear sampling method (lsm) was employed to estimate targets support. A relaxed threshold is defined for lsm to ensure all real targets are included. Despite the low accuracy of the estimation by lsm, the authors were more interested in its insensitivity to noise. Second, differential evolution (DE) optimisation was used to refine the lsm results. Since the magnitudes of lsm indicator functions are related to the probability of support belonging to real targets, different weights for gene elements are allocated according to their indicator values to control mutation and cross probability. A mask-beneficial contrast source inversion (CSI) is employed to evaluate cost function by encoding possible targets support and making it into a support mask. The mask is combined with CSI as a priori information and an accurate mask tends to derive a slight iteration error, whereas those incorrect ones will result in large errors. Thus, by adopting CSI iteration error as a cost function and taking advantage of superior optimisation performance of weighted DE, the authors find an optimum mask and reconstruct contrast with high fidelity. Effectiveness of the proposed method is validated by simulation results.
暂无评论