针对视觉和激光耦合simultaneous localization and mapping（SLAM）中存在的视觉特征丢失、雷达闭环轨迹矢量漂移和高程位姿偏差问题，提出一种通过扫描上下文回环检测的紧密耦合视觉和激光雷达SLAM方法。采用基于SIFT、ORB特征点检测器的视觉里程计解决特征点丢失和匹配失败问题；通过雷达里程计融合视觉里程计帧间估计消除雷达点云畸变和大幅度漂移；通过扫描上下文进行回环检测，引入因子图优化里程计矢量漂移，消除回环检测失败问题。在多个KITTI数据集中对所提算法进行验证，并与经典算法进行对比，实验结果表明，该算法具有高稳定性、较强鲁棒性、低漂移和高精度。

In this paper, a foveated imaging system using a reflective liquid crystal spatial light modulator (SLM) was designed. To demonstrate the concept of foveated imaging, we simulated with software Code V and established a laboratory prototype. The result of the experiment shows that an SLM can be used to correct the aberration of region of interest (ROI) while the resolution of other area was still very low. The varyresolution system was relative simple compared to the traditional high resolution system and obviously can reduce the amount of data transmission. Such systems will have wide application prospect in various fields

为了研制精确的星点定位误差补偿算法，得到星像轮廓面的形貌信息，研究了检测星敏感器点扩散函数（PSF）的方法。在高精度测量平台上进行光电采样，获取星像像素灰度值分布，根据扩展Nijboer-Zernike理论进行PSF拟合重建。噪声仿真实验证明，该算法能够在星像信噪比较低的情况下准确地得到Zernike系数，从而快速地重建甚高精度星敏感器的空域PSF。基本满足自动检测甚高精度星敏感器星像轮廓面的要求。