融合时域像特征的空间可见光小目标检测技术  

Spacial Small Target Detection Based on Optical Temporal Profile Characteriscs

作　　者：杨述强[1] 崔琦[2] 李飚[1] 

机构地区：[1]国防科技大学ATR重点实验室,长沙410073 [2]洛阳理工学院机电工程系,河南洛阳471023

出　　处：《电光与控制》2010年第6期36-39,共4页Electronics Optics & Control

摘　　要：在对空间目标进行远距离可见光成像观测过程中,由于目标信号较弱而且平台不稳,目标的信号因在焦平面上产生亚像素抖动而大大削弱,增加了远距离目标检测的难度。针对远距离目标检测难题,提出一种融合像素点时域像均值与方差两种特征的远距离弱信号目标检测算法。经实测图像序列分析,存在弱目标的像素点与纯背景像素点相比,时域像方差因平台抖动而较大,利用这一特征结合广义似然比准则可以有效检测出图像中的弱目标。最后通过计算机仿真验证了检测算法的有效性。In spacial dim target detection by optical image sequence,it is difficult to detect the target since the target signal is weak,the platform is unstable,and the target signal is weakened greatly due to subpixel dithering.To solve the problem,we proposed a novel detection method combining mean value with variance of optical temporal profile.Analysis to actual optical image sequence showed that： compared with the pixels of the background,the pixels of the target have larger variance due to platform shaking.By use of this character and the General Likelihood Ratio Test（GLRT）,we can figure out the target pixels in the image sequence.Simulation was made to verify the effectiveness of the detection method.

关 键 词：可见光图像序列 时域像 广义似然比检测 抖动噪声 

分 类 号：V271.4[航空宇航科学与技术—飞行器设计] TN959.17[电子电信—信号与信息处理]