【摘要】 传统的时延估计方法受瑞利限限制,一些基于子空间的方法可以达到高分辨或超分辨,然而,子空间方法需要处理的数据量大,计算复杂度高。提出一种新的基于自适应稀疏分解的探地雷达水平分层介质时延估计方法,自适应稀疏分解将传统的参数估计问题转化为字典学习问题。自适应稀疏分解从过完备字典中选取少量原子,其对应的时间参数即为所需估计的时延。与传统的基于子空间方法相比,该算法在低信噪比时具有更高的估计正确率,并且直接在时域进行,减少了计算复杂度。仿真及实测结果表明,该算法与MUSIC算法对比在时延估计方面具有更大的优势。更多还原

【Abstract】 Since the traditional time delay estimation method is restricted by the Rayleigh limit,some subspace-based methods can reach high resolution or super-resolution. However,the subspace-based methods have large data size for processing and high computation complexity. A new horizontally- layered medium time delay estimation method of using ground penetrating radar(GPR)is proposed,which is based on adaptive sparse decomposition(ASD),and in which the traditional parameter estimation is turned to the dictionary learning by ASD. A few atoms are selected from the over-complete dictionary by means of ASD,and their corresponding time parameter is the estimating time delay. In comparison with the traditional subspace-based methods,this algorithm has higher estimation accuracy at low signal-to-noise ratio(SNR),and can directly run in time domain,so as to reduce computational complexity. The simulation and actual measured results show that the proposed algorithm has better advantage than the MUSIC(multiple signal classification)algorithm in the aspect of time delay estimation.更多还原