一种渐进剥离与整体拟合的子波形提取方法

A Method of Sub-waveform Extraction Using Progressive Stripping and Global Fitting

针对全波形分解中弱子波形提取困难以及子波形提取精度不高的难点,提出了一种在总波形约束下的子波形渐进剥离分解方法,通过计算回波局部最大值来确定高斯分量的峰值和中心位置以及利用高斯拐点匹配确定左、右拐点求取半宽值,将剩余波形重复迭代剥离直至子波形的最大峰值小于振幅阈值,渐进获取各子波形参数的初值,进而在总波形约束下实现各子波形的整体拟合与参数精化。通过对ICESat-GLAS回波信号进行波形分解表明,该方法能有效提取弱子波形,与3种传统的波形分解方法相比,提取子波形的数目和参数初值合理,整体拟合后的均方根误差分别下降75%、66.85%和64.1%。

Aiming at the difficulty of extracting weak sub-waveforms and the low accuracy of sub-waveform extraction in full waveform decomposition,this paper proposes a method of progressive stripping decomposition of sub-waveforms under the constraints of global waveform.The peak and center position of Gaussian components are determined by calculating the local maximum of echo,and the left and right inflection points are determined by Gaussian inflection point matching to obtain the half width value.The remaining waveform is repeatedly and iteratively stripped until the maximum peak value of the sub-waveform is smaller than the amplitude threshold,and the initial values of each sub-waveform parameter are gradually obtained,and then the global fitting and parameter refinement of each sub-waveform is achieved under the constraints of global waveform.By decomposing the ICESat-GLAS echo signal and comparing it with three traditional waveform decomposition methods,the experimental results show that the proposed method can extract the weak sub-waveform effectively,the number of extracted sub-waveforms and the initial value of the parameters are reasonable,and the root mean square error is decreased by 75%,66.85%and 64.1%after global fitting respectively.