Photon counting lidar signals generally require smoothing to suppress random noise. While the process of reducing the resolution of the profile reduces random errors, it can also create systematic errors due to the smearing of high gradient signals. The balance between random and systematic errors is generally scene dependent and difficult to find, because errors caused by blurring are generally not analytically quantified. In this work, we introduce the use of Poisson thinning, which allows optimal selection of filter parameters for a particular scene based on quantitative evaluation criteria. Implementation of the optimization step is relatively simple and computationally inexpensive for most photon counting lidar processing.

中文翻译：

---

使用泊松细化技术优化光子计数激光雷达中的线性信号处理。

光子计数激光雷达信号通常需要平滑以抑制随机噪声。虽然降低轮廓分辨率的过程减少了随机误差，但由于高梯度信号的拖尾，它还会产生系统误差。随机错误和系统错误之间的平衡通常取决于场景，并且很难找到，因为模糊造成的错误通常无法进行分析量化。在这项工作中，我们介绍了泊松细化的使用，它可以基于定量评估标准为特定场景优化选择滤波器参数。对于大多数光子计数激光雷达处理而言，优化步骤的实现相对简单且计算上不昂贵。