一种迭代的小光斑LiDAR波形分解方法

Iterative decomposition method for small foot-print LiDAR waveform

针对传统LiDAR波形数据分解方法受噪声影响严重、对复杂重叠及微弱回波分解能力不足的缺点,提出了一种新波形分解方法.通过计算滤波前后波形的幅值变化,估计波形的随机与背景噪声;采用逐层剥离的策略,从原始波形数据中不断分解出波形分量,直到剩余波形中最大峰值小于一定的阈值;利用L-BFGS算法优化初始参数,获得波形分量参数的最优解;最后对位置过近的波形分量进行合并.该方法计算速度快,探测微弱回波能力强,显著提高分解后点云的密度与精度.对大量LiDAR波形数据进行了分解,验证了其有效性.

LiDAR (Light Detection and Ranging) waveform decomposition is a key issue in remote sensing data processing. Traditional waveform decomposition methods can't detect weak sub-waveforms when sub-waveforms are overlapped in original data. Besides, these methods are time consuming and not robust to noise. To overcome the obstacle, this paper proposed a new method, which mainly includes four steps. The first is to estimate the errors by filtering the original waveform. Then, iteratively peeling off sub-waveforms from the waveforms till the value of maximum peak is less than a given threshold. The next step is to optimize the parameters of all sub-waveforms using L-BFGS method. At last, nearest sub-waveforms are combined. This new strategy can detect the weak peaks in the complex situations and is very robust to noise. Lots of experiments demonstrate the effectiveness of the proposed method.