基于分块速度域改进迭代运动目标检测算法的红外弱小目标检测  

A block-based improved recursive moving-target-indication algorithm

作　　者：侯旺[1] 于起峰[1] 雷志辉[1] 刘晓春[1] 

机构地区：[1]航天科学与工程学院国防科技大学,长沙410073

出　　处：《物理学报》2014年第7期149-161,共13页Acta Physica Sinica

基　　金：国家重点基础研究发展计划(批准号:2013CB733100)资助的课题~~

摘　　要：提出一种基于分块速度域的迭代红外运动目标检测算法来解决传统算法计算量巨大这一难题.首先,采用二维最小均方差滤波器对红外序列图像进行滤波,获得包含弱小目标以及残差的红外序列图像.然后,通过在序列图像块的速度域上应用改进的迭代运动目标检测算法进行能量累积,从而将弱小目标的运动速度在速度域进行累积增强,达到检测弱小运动目标的目的.最后在解算出的速度值附近进行搜索,得到弱小目标运动的精确速度.利用此速度进行空域能量累积,得到叠加图像,在此图上进行目标检测.与传统方法相比较,几组实验结果显示,本文提出的方法大大缩短了检测的时间,而且本文方法的检测效果也较好.A new block-based recursive moving-target indication algorithm in velocity domain is proposed to solve the problem which is the rapid detection of dim and small target for infrared search and tracking system. Firstly, the two-dimensional least mean square filter is adopted to filter the infrared image sequence, which extracts small targets and residual errors of image sequence. Then, block-based recursive moving-target indication algorithm is adopted to accumulate small target in image block sequence for the enhancement of small target velocity in velocity domain. Finally, resulting image is obtained by using classical recursive moving-target indication algorithm and target velocity for small target detection. Compared with classical method, the proposed method requires less running time, and can be used to detect dim small target image effectively as demonstrated by several groups of experimental results.

关 键 词：目标检测 能量累积 速度域 分块 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]