探地雷达信号预处理及成像技术

发布时间：2018-10-16 11:13

【摘要】：探地雷达(GPR)利用了电磁波在不同电磁特性物体分界面上的折射和反射原理,实现对地下目标场景的反演。由于探地雷达的探测环境较为复杂,回波信号中不仅存在着有效信号分量,还有收发天线之间的直耦波分量、经地表直接反射波分量、噪声分量以及射频干扰分量等,如果不进行GPR信号预处理技术,有效回波信号容易被淹没在能量较强的直耦波和地表直接反射波当中,导致无法有效探测目标。本文针对GPR回波信号模型,研究了针对直达波去除和噪声抑制的信号预处理技术,结合实测数据分析了信号预处理技术的有效性。GPR成像是将GPR探测的地下目标回波信息转化为图像信息以更加直观的方式显示地下场景的信号处理技术。由于分层介质场景、双站工作模式的复杂性,本文从时域后向投影(BP)成像、频域距离偏移(RM)成像以及压缩感知(CS)成像三类成像算法入手,研究适合分层介质场景以及双站模式的GPR成像算法。其中,针对BP算法在GPR场景中计算复杂度较高的问题,研究了计算效率改进的BP算法,同时针对旁瓣和干扰能量较高的问题,本文提出了基于互相关信息的BP改进算法,并将此方法与计算效率改进的查找表BP算法结合起来,在提高计算效率的同时又保证了旁瓣和干扰的抑制能力,理论分析并实测数据以及仿真数据验证了算法的有效性。然后,针对传统RM算法在分层介质以及双站模式中应用的局限性,研究了基于分层介质的RM(LRM)算法,并将其推广到双站模式的LRM(BLRM)算法,实测数据以及仿真数据验证了算法在分层介质以及双站模式中应用的有效性。最后,研究了基于压缩感知(CS)理论的GPR成像算法,结合GPR成像模型以及地下目标的稀疏环境,利用三种稀疏重构算法实现了压缩感知理论下的成像结果,从聚焦度角度获得了比传统算法更好的成像效果,并从成像位置成功概率角度研究三种压缩感知算法随信噪比(SNR)、压缩数据比以及目标之间间隔的变化情况。
[Abstract]:The ground penetrating radar (GPR) (GPR) uses the principle of refraction and reflection of electromagnetic waves on the interface of objects with different electromagnetic characteristics to realize the inversion of the underground target scene. Because the detection environment of GPR is more complex, there are not only effective signal components in echo signal, but also direct-coupled wave components between transceiver antennas, direct reflection wave components, noise components and radio frequency interference components through the surface, etc. If the GPR signal preprocessing technique is not carried out, the effective echo signal is easily submerged in the direct coupling wave and the surface direct reflection wave, which makes it impossible to detect the target effectively. In this paper, for the GPR echo signal model, the signal pre-processing technology for direct wave removal and noise suppression is studied. The effectiveness of the signal preprocessing technique is analyzed by combining the measured data. GPR imaging is a signal processing technique that converts the echo information of the underground target detected by GPR into the image information to display the underground scene in a more intuitive way. Due to the complexity of layered media scene and bistatic operation mode, this paper starts with three imaging algorithms: time-domain backward projection (BP) imaging, frequency-domain range offset (RM) imaging and compression sensing (CS) imaging. The GPR imaging algorithm suitable for layered media scene and bistatic mode is studied. In order to solve the problem of high computational complexity of BP algorithm in GPR scene, the improved BP algorithm is studied. At the same time, aiming at the problem of high sidelobe and interference energy, an improved BP algorithm based on cross-correlation information is proposed in this paper. This method is combined with the improved lookup table (BP) algorithm, which not only improves the computational efficiency, but also ensures the ability of sidelobe and interference suppression. The effectiveness of the algorithm is verified by theoretical analysis, measured data and simulation data. Then, in view of the limitation of traditional RM algorithm in layered medium and bistatic mode, the RM (LRM) algorithm based on layered medium is studied and extended to the LRM (BLRM) algorithm of bistatic mode. The effectiveness of the algorithm in layered media and bistatic mode is verified by the measured data and simulation data. Finally, the GPR imaging algorithm based on compression sensing (CS) theory is studied. Combined with GPR imaging model and sparse environment of underground target, three sparse reconstruction algorithms are used to realize the imaging results of compression sensing theory. From the angle of focusing degree, the imaging effect is better than the traditional algorithm, and the variation of three compression sensing algorithms with SNR (SNR), compression data ratio and the interval between targets is studied from the point of view of image location success probability.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN957.51

【参考文献】