论文部分内容阅读
介绍几种超分辨成像方法,包括线性预测数据外推的离散傅立叶变换(LPDEDFT)法、动态优化线性最小二乘(DOLLS)法和Hopfjeld神经网络非线性最小二乘(HNNNLS)法,并给出B-52飞机缩比金属模型微波暗室转台实测数据和Boeing-727飞机外场实测数据的成像结果。比较了这些方法在分辨能力、计算复杂性等方面的优劣。采用这些超分辨成像方法,与普通的傅立叶方法相比,在相同信号带宽和总转角的条件下可以获得更高的图像分辨率,或用较小的信号带宽和总转角可以获得相同质量的图像。
Several super-resolution imaging methods are introduced, including LPDEDFT (extrapolation of linear prediction data), linear least squares (DOLLS) method and Hopfjeld neural network nonlinear least squares (HNNNLS) B-52 aircraft reduction model metal anechoic chamber turret measured data and Boeing-727 aircraft field measured data imaging results. The advantages and disadvantages of these methods in terms of resolution ability and computational complexity are compared. With these super-resolution imaging methods, higher image resolution can be achieved with the same signal bandwidth and total corner angle compared to the conventional Fourier method, or images of the same quality can be obtained with a smaller signal bandwidth and total angle of rotation .