全波形机载激光数据分解研究

Research on decomposition of full-waveform airborne laser data

针对进行全波形机载激光数据分解时采用重力中心(COG)方法和高斯脉冲拟合(GPF)方法难以检测到全波形数据中的叠加波和弱波脉冲,从而造成地物信息丢失的问题,提出了一种改进的严格高斯检测(RGD)算法。该算法首先采用高斯滤波提高信噪比,然后通过脉冲检测估计初始参数,并利用Trust Region参数优化方法进行高斯拟合,最终有效地分解出波形中的叠加波和弱波。利用提取到的每个单次回波的标准差、振幅、位置等模型参数能够得到半高波宽(FWHM)、后向散射截面和回波强度等波形信息。多次试验表明,与RGD算法相比,该方法可以有效地提高叠加波的识别率,减小波形拟合误差。

Based on the analysis of the present techniques of decomposing full-waveform airborne laser data, a modified rigorous Gaussian detection（RGD） algorithm was proposed to solve the problem that in waveform decomposition the center of gravity（COG） method and the Gaussian pulse fitting（GPF） method can not extract weak pulses and over- lapped pulses in the full-waveform data, so some earth surface feature information can not be achieved. Firstly, the algorithm uses the Gaussian filtering to improve the signal to noise ratio. Then initial parameters are estimated through pulse detection, and the trust region parameters optimization method is adopted to perform the Gaussian fit- ting. Finally, the weak pulses and the overlapped pulses are effectively decomposed. By using the model parameters such as standard deviation, amplitude and the position of each single echo extracted from decomposition, the full width at half maximum（ FWHM ）, backscattered cross-section and echo intensity, etc. , can be obtained. The numer- ous experiments were performed, and the results show that the proposed algorithm can effectively improve the recog- nition rate of overlapped pulses and reduce the error of waveform fitting compared with the rigorous Gaussian detec- tion algorithm.