论文部分内容阅读
压缩感知成像要求信号在某个域上能满足稀疏性要求,地下多目标在空域上降低了信号的稀疏性,导致成像出现散焦和虚像。扩大成像背景保证了稀疏性要求但又使得成像计算量上升,实时性不足。提出一种根据探地雷达回波特征预提取出潜在目标位置的压缩感知成像方法。通过对数据进行去噪、滑动矩阵过滤来确定目标的水平位置,再对水平位置处的几道A-Scan数据进行极值搜索,从而可以提取出成像区域目标位置信息,进而在建立成像冗余字典时只需考虑目标位置处的字典元素,无目标处字典元素直接剔除,减少字典建立所需的元素,降低了压缩感知求解计算量。该方法由于只对潜在目标区域进行成像,因此在保证成像实时性的同时也保证了成像精度。实验结果表明算法可行、有效。
Compressed sensing imaging requires signals to satisfy the sparsity requirement in a certain domain. The underground multi-objective reduces the signal sparsity in the airspace, resulting in defocus and virtual images. Extending the imaging background ensures sparsity but increases imaging computational complexity and real-time performance. A compressed sensing imaging method based on the echoes of GPR is proposed to extract the potential target position. Through the data denoising, sliding matrix filtering to determine the horizontal position of the target, and then the horizontal position of several A-Scan data for extreme search, which can extract the imaging area target location information, and then establish the imaging redundancy Dictionary only need to consider the target location of the dictionary elements, no target dictionary elements directly removed, reducing the elements required to establish the dictionary, reducing the computational complexity of perception. Because this method only images the potential target area, the imaging accuracy is ensured while ensuring the real-time imaging. Experimental results show that the algorithm is feasible and effective.