| 题名 | 基于CT原理的激光成像雷达关键技术研究 |
| 作者 | 金晓峰 |
| 学位类别 | 博士 |
| 答辩日期 | 2012 |
| 授予单位 | 中国科学院上海光学精密机械研究所 |
| 导师 | 刘立人 |
| 关键词 | CT原理 反射层析激光成像雷达 聚束模式非相干合成孔径激光成像雷达 投影图对准 相位恢复 成像分辨率 |
| 其他题名 | Research on Key Technologies of Imaging Ladar Based on CT Principles |
| 中文摘要 | 合成孔径激光成像雷达（SAIL）的原理来自于微波合成孔径雷达，是目前报道能够在远距离获得厘米量级分辨率的唯一光学成像观察手段。由于合成孔径激光成像雷达属于时间-空间全相干处理，对光频信号的振幅、偏振、频率、时间相位和空间相位都有严格的要求，实施上有非常高的难度。同时涉及到空间域衍射光学、时间域干涉光学、统计光学和图像处理四个领域基础科学问题，对激光、光机械、光电子、电子、光学和测量等单元和器件提出了更高的要求。针对合成孔径激光成像雷达结构复杂、实现难度高的问题，本文开展了基于Computed Tomography（CT）原理的激光成像雷达的研究，其优点在于成像中没有相位匹配的过程，不需要考虑光频的时间相位和空间相位同步问题，而是通过层析滤波反投影算法或傅里叶切片算法实现目标图像重建，从而降低了实施技术难度，增加了远距离高分辨激光成像雷达的应用范围。本文研究的基于CT原理的激光成像雷达主要分为反射层析激光成像雷达和聚束模式非相干合成孔径激光成像雷达两种，不同点在于反射层析激光成像雷达采用正视剖面扫描体系结构得到目标横截面轮廓像，聚束模式非相干合成孔径激光成像雷达采用侧视剖面扫描体系结构得到目标表面像。 文章首先给出了反射层析激光成像雷达的基本成像原理和数学分析，并进行了基于距离分辨的反射层析激光成像雷达实验，采用滤波反投影方法、傅里叶切片方法和迭代方法实现目标横截面轮廓像的重建。在此基础上，从以下几个方面对反射层析激光成像雷达进行了研究： 1、适合于反射层析激光成像雷达的傅里叶切片算法。对透射层析傅里叶切片算法进行了分析，将反射层析相隔180度的两个反射投影转换成一个对应的透射投影，从而避免了在同一个切片方向上引入不同的频域估计，使其适用于反射层析激光成像雷达情况。 2、投影图角度受限下的反射层析激光成像雷达成像分析。给出了全角度下的反射层析激光成像雷达系统点扩散函数，导出相应的理论成像分辨率，将其与距离分辨率建立比例关系。计算机模拟仿真投影图角度受限下点目标的水平成像分辨率，与理论成像分辨率比较发现，投影图角度受限引起的成像分辨率退化相当于对频域在相应角度方向进行了低通滤波，且成像分辨率在投影图角度受限范围小于40度下急剧恶化，此时增加角度采样率并不能使成像分辨率得到质的提高。 3、基于相位恢复和特征点跟踪的投影图中心对准技术。为获取目标精确图像重建，反射层析激光成像雷达的反射投影图数据要实现中心对准。探测目标旋转轴未知和目标存在随机平移的情况将导致投影图中心不在一条线上，基于未校准的投影图数据得到的重建图像将会发生错位模糊，严重影响成像质量。文章引入特征点跟踪和相位恢复两种方法改善图像质量，特征点跟踪方法基于点目标绕轴旋转在多个角度反射投影中的位置呈现正弦轨迹，相位恢复方法基于单个角度投影在时间（距离）上的延迟只在傅里叶频域引入线性相位，而不影响其频域模值分布。 4、设计了一种基于多普勒分辨的接发同轴反射层析激光成像雷达系统，给出了基本成像原理和数学表达，相应分析了目标截面横向距离分辨率和单个角度采样时间的关系。并在实验室平台上模拟远场传输，获取探测目标的多普勒反射投影图，采用滤波反投影算法实现了目标横截面图像重建。 文章第六章给出了聚束模式非相干合成孔径激光成像雷达的基本原理和数学分析，与反射层析激光成像雷达相比，其成像结果为目标二维平面像，不再是目标横截面轮廓像。进行了基于多普勒分辨的聚束模式非相干合成孔径激光成像雷达实验，由于存在局域角度信息量不足的问题，目标重建图像存在模糊和畸变，尚需进一步的研究。 |
| 英文摘要 | The technique of synthetic aperture imaging ladar (SAIL) was developed from the synthetic aperture radar in radio frequencies. It can provide the centimeter-class resolution with an aperture size no larger than a few meters on an observation range of thousand kilometers or more theoretically. The SAIL technique belongs fully coherent processing in both the time domain and space domain, accordingly there are the rigorous constraints on the amplitude, polarization, frequency, temporal phase and spatial phase of optical signals in time domain and space domain during the signal collection and image reconstruction. The SAIL involves spatial diffractive optics, temporal interference technology, statistical optics and image processing, which put forward higher requirements for laser, mechanism design, photoelectron, electronic equipment, optics and measurement devices. According to complex signal processing in SAIL, key technologies of imaging Ladar based on CT principles were studied in this paper. The main advantages of this imaging ladar are as follows: signal phrase processing and spatial phrase processing including phrase matched process are not necessary. The image is reconstructed by virtue of typical CT algorithm such as filtered back-projection algorithm or the Fourier-Slice algorithm. This technique reduces the difficulties in signal collection and data processing, and has a great potential for applications in extensive imaging Ladar fields. Imaging Ladar based on CT principles studied in our lab can be divided into two types: reflective tomography Imaing Ladar and spot-mode incoherently synthetic aperture imging Ladar. The cross section of the target was reconstructed by front-looking profiles scanning architecture in reflective tomography imaging Ladar, while the 2-D plane of the target was reconstructed by side-looking profiles scanning architecture in spot-mode incoherent synthetic aperture imaging Ladar. Firstly, the mathematical analysis and imaging process of reflective tomography imaging ladar were described in detail. The experimental setup for range resolved reflective tomography imaging Ladar was implemented. Filtered back-projection algorithm, Radon-Fourier algorithm and iterative algorithm were used to reconstruct the target. Based on the experimental data, the following parts about reflective tomography imaging Ladar are discussed: 1. Fourier-Slice algorithm for reflective tomography imaging ladar. In reflective tomography imaging Ladar, measurements over full 360°are necessary to reconstruct the target exactly. Projections of sampling angle interval 180o usually carry different information of the target. It will lead different Fourier estimates along the same line through the origin based on the standard Radon-Fourier theorem. Here, using the functional similarity between transmissive tomography and reflective tomography, we transformed reflective projections to corresponding transmissive projections. The target could be reconstructed by the same operation in transmissive tomography. 2. Imaging analysi in limit-view projections in reflective tomography imaging Ladar. The theoretical point spread function (PSF) of imaging system over full views for reflective tomography imaging Ladar was presented. The theoretical imaging resolution was derived approximately from the range resolution resolved by the laser pulse. The simulated horizontal imaging resolutions of an ideal diffused single point target in different limited-views were compared with the theoretical value. It is found that limited number of views degrades the imaging resolution by low pass filtering the Fourier domain of the target in certain directions. Also the imaging resolution does not obtain substantial improvement in limited views less than 40 degree, even with smaller sampling angle interval. 3. Projections registration based on phase retrieval method and feature tracking method. To obtain the image reconstruction accurately, the projections data in reflective tomography imaging Ladar must be registered in the same line. The unknown location of the rotation axis and random translation of the target will bring image blur and geometric distortion in the image reconstructed. Feature tracking method and phase retrieval method were introduced to improve imaging quality. 4. A system of Doppler resolved reflective tomography imaging Ladar and its algorithm were given. The issue between transverse range resolution and the sampling time of single angle was solved. We designed the condition of far-field diffraction in the laboratory, and collected the Doppler resolved reflective projections of the target. Filtered back-projection algorithm was used to reconstruct the cross section of the target. A concept of spot-mode incoherently synthetic aperture imaging Ladar and its algorithm were proposed in Charter 6. Compared with the reflective tomography imaging Ladar, the imaging result of spot-mode incoherently synthetic aperture imaging Ladar was 2-D plane of the target, not the cross section of the target.The expreimental setups of Doppler resolved spot-mode incoherently synthetic aperture imaging Ladar were explemented. Because of the incomplete projections data in limited views, blur and geometric distortion exist in the images reconstructed, and it needs further study. |
| 语种 | 中文 |
| 内容类型 | 学位论文 |
| 源URL | [http://ir.siom.ac.cn/handle/181231/15699]  |
| 专题 | 上海光学精密机械研究所_学位论文 |
| 推荐引用方式 GB/T 7714 | 金晓峰. 基于CT原理的激光成像雷达关键技术研究[D]. 中国科学院上海光学精密机械研究所. 2012. |
| 个性服务 |
| 查看访问统计 |
| 相关权益政策 |
| 暂无数据 |
| 收藏/分享 |
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。
修改评论