压缩感知LIDAR三维成像原理与方法研究

项目名称： 压缩感知LIDAR三维成像原理与方法研究

项目编号： No.61307022

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 无线电电子学、电信技术

项目作者： 柯钧

作者单位： 北京理工大学

项目金额： 24万元

中文摘要： 本项目将利用压缩感知技术以及多次成像与数据处理相结合的超分辨方法实现高分辨、单元探测、无需扫描的LIDAR三维成像（Compressive LIDAR 3D imaging - C-LIDAR3D）。 C-LIDAR3D系统直接采集多目标点深度信息的线性组合，而非对每一目标点逐点扫描。为了实现高分辨率，我们首先将分析影响系统分辨率的因素，确定系统工作条件。在光路横截面方向上，将使用机械移位装置改变空间光调制器(SLM)及其上目标图像间相对位置，形成SLM亚像素大小的移位距离。通过对多次移位后采集所得数据进行综合处理，突破SLM像素对分辨率的限制。在纵深方向上，同样地利用多次数据采集超分辨原理，将直接采集不同延时后回波信号的线性组合。通过对多组测量值综合处理，获得高于采样器件采样速度限制的测距分辨率。此外,将开展测量矩阵设计与恢复算法研究，搭建试验验证系统。

中文关键词： 压缩感知；激光雷达；超分辨；测量矩阵设计；并行计算

英文摘要： Compressive sensing and multi-image-acquisition superresolution method, will be used in this project to achieve high resolution compressive LIDAR 3D imaging (C-LIDAR3D), with a single detector and without a scan. Using compressive sensing, a C-LIDAR3D system directly acquires linear combinations of multiple pixel depth information, instead of scanning through all pixels. To achieve high resolution, we will study the important parameters for resolution, find the working conditions for C-LIDAR3D first. For the transversal direction, a mechanical device will be used to modify the relative position between SLM and the image focused onto it. The shift distance is in SLM sub-pixel scale. The acquired data after each shift will be processed together. By using this method, we wish to improve the resolution restricted by SLM pixel. For the axial direction, using the same multi-data-acquisition superresolution idea, we will directly measure linear combinations of delayed object reflected light signals. By processing multiple sets of such measurements, we wish to obtain a ranging resolution smaller than the value determined by an A/D device sampling step. Additionally, we will study measurement matrixes and object reconstruction methods for C-LIDAR3D, and set up an experiment for demonstration.

英文关键词： Compressive sensing；LIDAR；superresoltuion；sensing matrix design；prallel computing