【作者】 胡永忠；

【导师】 周正欧；

【作者基本信息】 电子科技大学 ， 信号与信息处理， 2010， 博士

【摘要】 微波阵列天线成像技术在国防安全、工业生产、医学成像等诸多领域都有着广泛而重要的应用。其在不同的应用领域既有一些共同的技术特点,也存在一些各自领域独特的技术问题。本论文针对微波阵列天线在穿墙成像和医学断层成像应用中的干扰抑制等问题进行了研究,主要的工作及创新点如下:1.针对线性阵列天线BP算法成像结果中存在虚假目标图像的问题,提出使用成像概率函数滤波的抑制方法。利用成像计算中目标单元信号是同相叠加,而非目标单元信号是异相叠加的特点,以此来定义成像概率函数,通过选用适当的滤波门限可有效消除虚假目标图像。应用中,由于部分像素单元的成像概率函数计算存在误差,导致使用成像概率函数滤波后,仍然会残留部分虚假目标图像。针对这一问题,提出利用图像位置信息反向求解滤波的方法,可进一步抑制残留的虚假目标图像。仿真成像处理的结果表明,所提出的抑制方法效果显著,实现方便,不受墙体参数误差的影响,可有效提高线性阵列天线穿墙成像的质量。2.针对射频干扰对微波医学成像系统的影响,利用系统激励源与干扰源的非相关性,提出对系统激励信号初始相位按照均匀分布在[-π,π]的随机相位序列编码的方法,将射频干扰转变为零均值的随机序列,进而再使用均值滤波的方法进行抑制。以乳腺肿瘤二维微波断层成像系统的应用为例,给出了随机相位编码方法的实现过程,通过仿真成像实验验证了该方法的有效性。3.针对环形阵列天线信号测量中,传统的直接测量模式数据存在动态范围广、参数灵敏度差异大和信噪比特性较差等不足,提出了测量天线对信号差值的测量模式。以16天线的乳腺肿瘤环形阵列成像系统为例,通过数值分析和成像实验的方法,对差值测量模式的数据特性进行了对比研究。结果表明,差值测量模式能有效改善直接测量模式的不足,具备更优越的测量数据特性。4.分析了基于非线性迭代方法的微波断层成像计算中,矩阵方程的阶数与有限元网格密度的内在关系。分析表明,随着计算网格密度的提高,求解方程的计算复杂度和计算时间,都呈指数级数增长。为便于微波断层成像技术的普及应用,针对乳腺肿瘤二维微波断层成像在个人计算机上的实现,提出了一种二次成像的快速方法。其基本原理是对感兴趣的图像区域采用高密度网格计算以保证成像精度,对于不太感兴趣的图像区域采用较低密度网格计算以减少计算单元数量,提高成像效率。介绍了二次成像快速方法的实现过程。经理论分析和成像实验的结果表明,在乳腺肿瘤二维微波断层成像应用系统中,所提出的二次成像方法可以对常见的乳腺肿瘤目标进行成像检测,与传统的使用全局精细网一次成像方式相比,这种二次成像方法可以节省成像时间80%左右,同时可以获得相近似的目标成像质量。5.针对手指关节的微波断层成像检测应用,提出利用X射线成像结果来获取微波断层成像的先验信息,通过优化迭代计算的初值分布,来改善迭代成像的效果。基于所设计的手指关节简化模型进行了仿真成像实验。仿真成像结果表明,利用X射线图像获得的手指关节结构先验信息来成像,可以对手指关节模型中2mm宽的关节区域进行成像检测,模拟的病态关节参数变化特征可以在成像结果中检测出来。更多还原

【Abstract】 Microwave array imaging (MAI) technology is widely used in national security, industrial measurement, medical diagnosis, and etc. It is playing an important role in variety applications. There are common technical features in the different areas, and there are also some unique technical problems in their respective field. This thesis focuses on the study of interference suppressing and fast imaging method to solve the problems encountered in through-wall-imaging and microwave medical tomography applications. The main contents and innovations of this thesis are as follows:1. To suppress ghost image brought in by backward projection (BP) imaging algorithm, the imaging probability function (IPF) method is proposed to reduce the ghost images. Due to the addition operation in target pixel unit is in phase, and the addition operation in non-target pixel unit is out of phase, we define the imaging probability function based on imaging addition operation, and suppress the ghost images using proper gate value of IPF to filter. In practice, there are calculation errors of IPF in part of pixel unit. These errors make part of ghost images can’t be suppressed using IPF to filter. To solve this problem, using location inverse solution (LIS) method is proposed to reduce residual ghost images. The simulation results show that the proposed methods can’t be affected by errors of wall parameters, and can effectively suppress ghost images.2. To reduce the effect of radio frequency interference (RFI) in microwave medical tomography system, according to the non-correlation feature between RFI sources and system illuminating source, a phase random coding method is proposed to transform RFI into zero mean random sequence, then the RFI can be suppressed simply using traditional mean-value filter algorithm. We conducted a set of 2D breast cancer imaging experiments using simulated data to test and verify the proposed method. Simulation results show that the proposed method is feasible and effective.3. Due to the measured data obtained by direct measurement pattern in circular antenna array is wide in dynamic range, unbalanced in parameter sensitivity and poor in signal to noise ratio (SNR), a difference measurement pattern is proposed to improve those deficiencies. Theoretical analysis and imaging experiments were conducted to evaluate the performance of proposed difference measurement pattern on 16-antenna microwave tomography system. Results demonstrate that difference measurement pattern have better performance of measured data.4. Analysis shows that the computational complexity and time-cost of imaging algorithm, based on nonlinear iterative method, will evidently increase with a speed of exponential progression while the density of computational mesh increases. To facilitate the spread of 2D microwave tomography for breast cancer detection, a fast imaging method is proposed in the realization of a personal computer, which uses a coarse mesh to image and locate the site of target firstly, then builds a locally refined fine mesh to image secondly. Imaging experiments were conducted using simulated and experimental data for breast cancer detection, the imaging results demonstrate that compared with the conventional imaging method using an overall fine mesh, the proposed twice imaging method can remarkably reduce the complexity of computation and save time-cost up to 80% while the equivalent image quality of target can be obtained.5. To detect finger arthritis using microwave tomography, an imaging method using priori information obtained from X-ray image is proposed, which optimize initial distribution of imaging parameters to improve quality of image. Imaging experiments were conduced using simulated data based on a simplified finger joint model. Experimental results show that a 2mm-wide finger joint can be quantitatively imaged based on its microwave property, and the pathological feature of finger joint can be detected from imaging result.更多还原