Image reconstruction of impenetrable cylinders using cubic B-splines and genetic algorithms

摘要

We discuss a new method for the shape reconstruction of perfectly conducting cylinders item measured data based on the cubic uniform B-splines [1] and genetic algorithms [2]. The perimeter of the cylinder is modeled by cubic B-splines. A B-spline is defined by an ordered set of control points which determine the shape of the corresponding curve. With an appropriate choice of control points, smooth curves as well as sharp edged contours can be modeled. The direct problem is formulated using standard integral equations which are numerically solved using the method of moments. Although the cylindrical contour is discretized with segments less than a tenth of a wavelength in size, the inversion algorithm deals only with the control points whose number is typically an order of magnitude less than the total number of the segments. The inverse problem is formulated based on an optimization of the spatial coordinates of the control points corresponding in the optimal shape matching the measured echo width. Since a fairly complicated closed curve can be easily and accurately modeled by a B-spline with a limited number of control points, the search space is drastically reduced. A simple genetic algorithm is applied for the inversion of the measured data. We present sample test results obtained by applying the proposed algorithm on real experimental data. Both principal polarizations (TM and TE) are treated separately and in combination depending on the availability and the nature of the measured data. The performance and limitations of the inversion algorithm on full as well as limited angular views are discussed in detail. It is shown that the proposed method is less sensitive In chive than these using Fourier expansions [3].