穿墙雷达因为具有墙体透视效果好、受环境影响小等特点，在军事和民用两方面都有广阔的用途，所以近些年发展速度非常快。同时该技术在应用中也存在很多问题，例如体积大、重量重、功耗高、操作复杂等，限制了穿墙雷达在一些条件较为苛刻场景中的使用，所以穿墙雷达的微小型化是目前其发展的一个重要方向，即出现了便携式穿墙雷达，但是由于硬件和使用环境的限制，其成像效果并不理想。本文针对该问题，在对穿墙雷达成像技术进行深入研究和分析的基础上，提出一种可用于便携式穿墙雷达成像的有效方法。

本文首先对不同频率信号下墙体的介电参数和人体目标的信号进行了实验研究，分析总结了墙体的介电参数随频率变化的特点，以及人体目标回波信号的特征，为穿墙雷达成像过程中检测和辨识目标信号奠定了基础，对接下来面向成像方法的研究和改进也具有重要指导意义。然后在分析便携式穿墙雷达的使用条件后，发现后向投影(BP)成像算法较为适合在该雷达系统中使用，其具有墙体补偿精确度高、抑制杂波效果好及对雷达移动路径没有严格要求等优点。但是经典BP算法的计算复杂度为O(MLN),存在计算量大，成像速度慢的问题，所以针对该问题提出一种基于非均匀快速傅立叶变换(NUFFT)的快速BP算法，通过选择合适的窗函数，将均匀的采样数据变换为非均匀数据，减少计算的数据量，从而降低计算复杂度。最后用FPGA开发系统和穿墙雷达实测数据对快速BP算法的硬件可实现性和实际成像效果进行了验证。

成像算法经过改进后，基于非均匀快速傅立叶变换(NUFFT)的快速BP算法可以将计算复杂度降低为，通过与经典BP算法的MATLAB仿真结果对比，证明了所提算法能在保证成像质量的同时提高成像速度。经过实验，验证了快速BP算法可以在FPGA开发系统上成功实现，并且在条件参数相同的情况下，其结果与MATLAB仿真结果具有很好的一致性，说明所提算法具有良好的硬件可实现性。通过实验室研制的便携式穿墙雷达系统对木板和砖墙后人体目标进行了探测，并利用快速BP算法对测得的数据进行成像，其结果可以准确反映目标的位置信息，证明快速BP成像算法能够对目标实现准确快速的穿墙成像。

The through-wall radar has a wide range of applications in both military and civil fields, because it has the characteristics of good wall perspective and little environmental impact. And the through-wall radar has developed very fast in recent years. At the same time there also exist many problems in the technology, such as big volume, heavy weight, high power consumption, complex operation and so on. These problems limit the usage scenario of the through-wall radar in some times. So research on compact through-wall radar, namely the portable through-wall radar, is an important direction in this field. But due to limitations in hardware and environment, the imaging is not ideal. In this thesis, an effective method for the portable through-wall radar imaging is proposed, which is based on the research and analysis of the through-wall radar imaging technology.

Firstly, on the basis of experiments, characteristics of the wall's dielectric parameters and the human target echo signals are studied. This is important to detect signals, identify targets, and improve imaging methods. Then, after analyzing the operating conditions of portable through-wall radar, we find that the back projection (BP) imaging algorithm is more suitable for this radar system. It has the advantages of accurate wall compensation, good noise suppression, and less strict requirements on the radar movement path. However, the computational cost of the classical BP algorithm is O(MLN). And there is a problem of slow imaging speed. So, a fast BP algorithm based on non-uniform fast Fourier transform (NUFFT) is proposed. Through selecting an appropriate window function , the uniform sampled data is transformed into non-uniform data. This way can reduce the computational cost by reducing the amount of data. Finally, the hardware implementation and actual imaging effect of the fast BP algorithm are verified by the FPGA and the measured data.

The fast BP algorithm based on non-uniform fast Fourier transform (NUFFT) can reduce the computational cost to . By comparing with the simulation results of classical BP algorithm in MATLAB, it is proved that the proposed algorithm can improve the imaging speed and ensure the imaging quality. Experiments show that the fast BP algorithm can be successfully implemented on the FPGA, and the results are in good agreement with the simulation results of MATLAB under the same conditions, which indicate that the proposed algorithm has good hardware implementation. The portable through-wall radar system developed in our laboratory is used to detect the human target behind the wooden and brick walls, and the measured data is used to image by the fast BP algorithm. The results show that the target can be accurately positioned, which proves that we can achieve accurate and fast through-wall imaging by the fast BP imaging algorithm.