基于图像局部特征的红外小目标检测与跟踪算法  

An Novel Algorithm for Infrared Small Targets Detection and Tracking Based on Local Image Features

作　　者：成兰[1] 王志杰[1] 姬琇荔 

机构地区：[1]安阳工学院科研处,河南安阳455000

出　　处：《红外技术》2008年第6期324-326,334,共4页Infrared Technology

摘　　要：基于红外序列图像中小目标的典型特征,提出了一种新的红外小目标检测与跟踪算法。利用图像小目标的微分几何特性,计算图像的最小法向曲率,并以此为阈值,获得小目标的候选对象。之后利用多幅图像之间目标的相关性,在时间尺度上基于Kalman滤波算法,实现对目标噪声的过滤并实现目标的跟踪。最后对算法的有效性进行了验证。Based on the typical feature of small target in sequence infrared images, a new algorithm is proposed for detecting and tracking small targets in sequence images. By using differential geometry property of small object of infrared image, computing minimal normal curvature which is used as the threshold to obtain candidate objects for small target. By using the relationship of objects between different images, blurred objects are filtered by Kalman filter. Finally, the effectiveness of this algorithm is verified.

关 键 词：目标检测 序列图像 法向曲率 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]