【作者】 周定法； 薄亚明；

【机构】 江南大学通信与控制工程学院；

【摘要】 在电磁逆散射问题中,需交替求解非线性的散射方程和逆散射方程,以达到对全场和待反演介质分布的精确近似。传统的Tikhonov正则化方法仅考虑了数据项误差的影响,而截断完全最小二乘(TTLS)正则化方法则同时考虑了逆散射方程系数矩阵和数据项误差。与Tikhonov正则化相比,TTLS方法不仅适用于不适定性较弱的情况,也适合于不适定性较强的情况。数值算例表明,配合适当的正则化参数,TTLS方法能够更好地消除噪声、提高反演的质量与可靠度。更多还原

【Abstract】 Linear forward and inverse scattering equations must be solved alternately to provide the total field estimation and objected dielectric distribution in resolving electromagnetic inverse scattering problem. In this process, the traditional Tikhonov regularization method deals with the effect of data item error , however, the truncated total least squares(TTLS) regularization method gets involved in both coefficient matrix and data item error of the inverse scattering equation. TTLS method is suitable for equations with poor inadaptability and with strong inadaptability as compared with the Tikhonov regularization method. TTLS method, in conjunction with appropriate regularization parameters, is used to filter noise efficiently and improve quality and reliability of reconstruction image.更多还原