条纹管激光成像雷达条纹图像噪音分析与处理  

Noise Analyzing and Processing of Streak Image for Streak Tube Imaging Lidar

作　　者：臧俊恒[1] 李思宁[1] 王骐[1] 刘金波[1] 广宇昊[1] 

机构地区：[1]哈尔滨工业大学光电子技术研究所,哈尔滨150080

出　　处：《光子学报》2008年第8期1533-1538,共6页Acta Photonica Sinica

基　　金：哈尔滨工业大学优秀团队支持计划资助

摘　　要：从条纹管激光成像雷达的结构和成像原理出发,讨论了各种噪音来源、噪音特点、影响因素、对最终图像的贡献大小和抑制方法.通过条纹管激光成像雷达阶梯目标扫描成像实验结果分析,噪音源理论分析得到了验证.结合条纹图像处理的特殊目的,通过对比几种边界保持类平滑滤波算法,得出K近邻平滑均值滤波器具有更低的时间复杂度和空间复杂度和更好的滤噪效果,当取N=7,K=25时,可以在允许的处理时间内极大地提高条纹图像信噪比,然后,利用阈值算法有效滤除了背景噪音,最终成功地从复杂的噪音中提取到了条纹数据.这项工作为后续的目标像重构奠定了基础,并指出了下一步工作的方向和重点.The first step of streak image processing is noise suppressing, but noise suppressing algorithms for images of traditional scanning lidar are not applicable for streak image. Through the configuration and imaging principle of streak tube imaging lidar, noise sources, noise characteristics, influencing factors, contribution to streak image and suppression methods are discussed in details. Noise source analyzing result gets verified through streak tube imaging experiment for staircase tagets. For the special processing purpose of streak image, K Nearest Neighbor Mean Filter indicates lower time and space complexity and better noise filtering performance among classical edge preserving filters. As N = 7, K = 25, K Nearest Neighbor Mean Filter can improve the signal-to-noise of streak image in a certain extent in tolerance time. Then threshold algorithm is used to filter background noise and streak data is obtained finally among complex noise. This work lays a foundation for the following target image reconstruction and points out the direction and keystone of the next step.

关 键 词：激光成像雷达 噪音抑制 K近邻平滑均值滤波 条纹管 噪音分析 图像处理 

分 类 号：TN249[电子电信—物理电子学]