激光雷达信号的可变间隔阈值经验模式分解去噪法

针对采用经验模式分解直接阈值(EMD-DT)和经验模式分解间隔阈值(EMD-IT)在激光雷达回波信号的去噪应用中会产生的模态混叠现象,采用一种可变间隔阈值的经验模式分解(EMD-SIT)的去噪方法。首先,对信号进行经验模式分解。然后,采用过零率方法将分解出的含有噪声的固有模态函数分离。最后,应用过零点阈值,设立一个新的可变阈值,将EMD-IT和EMD-DT有效融合对信号进行去噪。通过与多种阈值的仿真对比以及激光雷达的回波信号去噪实验,结果表明该方法可以有效地去除噪声,抑制模态混叠,较EMD-IT和EMD-DT更具有优越性,因此有着很好的应用前景。

Denoising method using empirical mode decomposition with switchable interval threshold for lidar signals

Using the empirical mode decomposition direct threshold (EMD-DT) and empirical mode decomposition interval threshold (EMD-IT) to de-noise lidar return signals, aiming to solve the problem of mode-mixing, a method of EMD switch interval threshold (EMD-SIT) is applied. Firstly, the signal is processed by the use of EMD. Then the decomposed intrinsic mode functions (IMF) with noise are removed by zero-crossing rate method. Finally ,by the use of the zero-crossing threshold, a switch interval threshold is obtained, which combines the EMD-IT and EMD-DT to de-noise the noise signals. The comparisons of different thresholds and the de-noise experiment of lidar shows that noise can be removed effectively by the proposed method, meantime the mode mixing is restrained. Compared with EMD-DT and EMD-IT methods, EMD-SIT method has more advantages Therefore ,the demonstrated method has a promising future.