小波去噪激光雷达数据树冠高度反演

Estimating forest canopy height using Lidar data based on wavelet denoising

针对光学遥感图像记录的地表物体反射太阳辐射能量的高低不能较好地反映地表林木冠层高度这一问题,该文利用小波去噪对激光雷达数据进行去噪处理,分别计算小波去噪和高斯去噪后信号的信噪比(SNR)、均方根误差(RMSE);评价去噪效果,发现利用小波去噪得到的信号SNR均大于13db、RMSE均小于0.007 0,去噪效果优于高斯去噪;根据树木垂直结构以及回波原理利用回波分析法,确定回波开始以及地面回波位置来反演树木冠顶高度;将反演得到的树高进行对比分析,并利用实测数据对反演结果进行检验,发现本文反演结果与实测数据具有较好的相关性,反演结果稳定、精度较高。

Aiming at the problem that optical remote sensing image cannot reflect the canopy height well, Li- dar data was denoised by both wavelet denoising and Gaussian denoising after which SNR and REMS of signals in each method were respectively calculated to evaluate the denoising effect in this paper. It turned out that the denois- ing effect was better when Lidar data was processed by wavelet denoising in which SNR was greater than 13db and RMSE was less than 0. 0070. Forest canopy height was estimated by locating start position and ground peak in ac- cordance with echo analysis, which was based on the forests vertical structure and the principle of Lidar backscat- ter. Forest canopy height was compared with the result of standard deviation of the mean and the ratio and tested by measured data. Precision test showed that canopy heights estimated by echo analysis correlated significantly with measured data which was better than estimating canopy the ratio, which meant the proposed method was much height via calculating standard deviation of the mean and better in terms of stability, rationality and accuracy.