红外序列图像中缓动点目标的流水线检测算法  

A Pipeline Algorithm for Detecting Slowly Moving Targets in Infrared Sequence Images

作　　者：韩客松[1] 王永成[1] 

机构地区：[1]上海交通大学图像处理与模式识别研究所,200030

出　　处：《系统工程与电子技术》2000年第8期66-67,84,共3页Systems Engineering and Electronics

基　　金：国家"8 63" - 4 0 9基金资助课题

摘　　要：主要讨论了红外序列图像中缓动点目标的流水线检测算法。把序列图像看成一个三维空间 ,设计了两个流水线 ,即分割流水线和检测流水线。前者通过求差值图像的门限分割 ,获得侯选目标点 ;后者对剩下的目标点进行窗口幅值累加和门限处理 ,获得目标点。实验显示 ,可以用 15帧图像较高效地检测出目标。最后对算法的性能和实验结果作了分析。This paper focuses on a pipeline algorithm for detecting slowly moving targets in infrared sequence images with complex background. Taking the sequence images as a 3-dimension space, we design two pipelines, one is named segmentation pipeline, and the other, detecting pipeline. The former computes the difference between two images and uses threshold to get candidate targets, while the latter accumulates the values of the points in the window and gets the targets. We can satisfactorily detect the targets within 15 frames of images. At the end of the paper, the algorithm's performances and experiment results are analyzed respectively.\;

关 键 词：红外图像 目标识别 流水线检测算法 

分 类 号：O434.32[机械工程—光学工程]