-
脉冲激光目标精确定位算法的核心在于如何减小波形畸变,因此EMD分解重构算法的降噪性能及峰值点位置偏移情况需要加以仿真验证[6]。另外,开展外场实验可以获取在接近真实应用条件下的数据,对目标精确定位算法的性能加以实验验证。下面实验主要针对单帧回波信号降噪及峰值点位置提取开展仿真实验。
实验1:基于EMD的激光回波降噪
将单帧脉冲激光目标回波信号以一定的信噪比与高斯白噪声进行叠加,得到待处理的含噪回波信号。使用EMD方法对含噪回波信号进行处理,并对选择EMD分解得到的若干IMF项及残差项进行重构,得到降噪后的波形。其中,IMF项及残差项波形如图3所示,脉冲激光回波信号EMD降噪结果如图4所示。图5给出了不同降噪算法的处理结果。
Figure 3. Empirical Mode Decomposition of the pulse laser echo signal. (a) IMF-1; (b) IMF-2; (c) IMF-3; (d) IMF-4; (e) IMF-5; (f) IMF-6; (g) IMF-7; (h) Residual
Figure 4. Results of the echo signal denoised by EMD. (a) Emitting laser waveform; (b) Echo signal; (c) Result of the echo signal denoised by EMD algorithm
Figure 5. Results of the echo signal denoised at SNR=−15 dB by various denoising methods. (a) Emitting signal; (b) Echo signal; (c) Result of the echo signal denoised by SF algorithm; (d) Result of the echo signal denoised by WRD algorithm; (e) Result of the echo signal denoised by EMD algorithm
实验2:针对单帧激光回波信号降噪,几种常用数字滤波算法性能对比
机载多脉冲激光测距机接收的目标回波脉冲波形的宽度为20 ns,依照对数幅度信噪比−15 dB混合高斯白噪声。采用数字平滑滤波(Smoothing Filter,SF)算法、小波降噪(Wavelet Reconstruction Denoising,WRD)算法、EMD降噪(Empirical Mode Decomposition Denoising,EMDD)算法进行100次回波信号降噪处理,统计回波信号的最小可检测信噪比。
表1给出了最小可检测信噪比的统计结果。可以看出,经窗口宽度为10点的平滑滤波,其最小可检测信噪比稍逊于窗口宽度为15点的SF算法。而WRD算法的信噪比改善效果明优于SF算法。而与SF和WRD算法相比,文中提出的EMDD算法的可检测信噪比改善效果更优。
Algorithms SF(M=10) SF(M=15) WRD EMDD SNR/dB −10 −12 −15 −17 Table 1. The least SNR detected by various denoising algorithms
实验3:针对降噪后的单帧激光回波信号波形,几种常用数字处理算法峰值点位置偏移情况对比
采用三种降噪算法滤波后直接检测峰值点的位置并进行对比。表2给出了不同信噪比时不同算法针对单帧回波信号进行目标检测后的峰值点位置与真实值的误差点数,其中统计次数为100次平均,X表示无结果。
Algorithms SF(M=10) SF(M=15) WRD EMDD SNR=−10 dB 15 9 5 2 SNR=−12 dB X 11 5 3 SNR=−15 dB X X 6 3 Table 2. The estimation error of the peak point position by different signal processing algorithms
由表2可以看出,随着回波信号信噪比不断降低,SF、WRD和文中EMDD算法的目标位置估计偏移量不断增大;另外,文中EMDD算法的目标位置估计偏移量在不同回波信号信噪比情况下均小于对比算法(SF和WRD算法),且当回波信号信噪比为−10 dB时,文中EMDD算法的目标位置估计偏移量达到了两个点,进一步改进了对比方法目标位置估计偏移量过大的问题。文中提出的EMDD激光回波降噪数字处理方法不仅有较好信噪比改善性能,且处理后得到的回波波形峰值点位置更接近真实值。
图6为脉冲激光测距机消光比法测距机性能的外场评估结果。采用文中算法研究开发的测距机在中国计量科学研究院外场打靶测试,靶面位置不动(靶面距定位孔27.5 cm),测距机分别放置在不同的测距台上,出光口到定位孔的距离为25 cm。测试前先用校准好的枪瞄瞄准目标,然后加合适倍率的衰减片完成距离测量,用于绝对距离的校准及测距精度的分析。记录完数据后,逐渐增加衰减倍率,直到不能稳定得到回波数据,用于评估测距性能。中国计量科学研究院标定距离分别为:431.027 m (5号靶)、623.049 m (6号靶)、863.040 m (7号靶)、1 175.0611 m (8号靶),测距机实测距离分别为:430.502 m、622.524 m、862.515 m和1 174.535 m。
Target echo denoising algorithm of airborne platform multi pulse laser rangefinder under low SNR
doi: 10.3788/IRLA20210005
- Received Date: 2021-01-05
- Rev Recd Date: 2021-02-09
- Publish Date: 2021-06-30
-
Key words:
- multi pulse laser ranging /
- digital echo denoising /
- Empirical Mode Decomposition(EMD) /
- target positioning accuracy
Abstract: For long-distance aircraft targets, airborne multi pulse laser rangefinders usually adopt digital processing system. The factors that affect the ranging accuracy include the range resolution of emitting laser pulse width, the waveform broadening caused by atmospheric propagation and target reflection characteristics, the nonlinear phase frequency characteristics of echo signal receiving and processing channel, etc. Digital sampling rate, echo signal denoising and target waveform peak point position estimation usually cause large target range positioning error. In this paper, an echo denoising method based on Empirical Mode Decomposition(EMD) and reconstruction was proposed. Firstly, the echo waveform of pulse laser target was modeled, and then the performance of existing digital echo denoising algorithms and its influence on positioning accuracy were analyzed. The experimental results show that the proposed method can accurately extract the target peak point position and improve the target positioning accuracy of laser rangefinder.