遮挡目标检测与识别技术研究  

作　　者：蔡星艳 赵和鹏[2] 邱鹏[3] 朱长仁[3] 

机构地区：[1]91635部队 [2]海装电子部 [3]国防科技大学ATR国防科技重点实验室

出　　处：《数字技术与应用》2013年第9期73-75,共3页Digital Technology & Application

摘　　要：遮挡问题在现实世界中经常发生,如何在遮挡情况下准确检测和识别感兴趣的目标是计算机视觉领域的一个重要课题。由于遮挡目标的信息不全,所以遮挡目标的检测和识别是一个具体挑战的课题,其研究具有重要意义。本文对基于图像的遮挡目标检测与识别方法按照三大类进行分类总结分析,在分析各类方法基本思路和优缺点基础上,给出了遮挡目标检测识别方法的发展趋势。Object occlusion often appears in real world, and it is an important problem how to detect and recognize interesting objects under occlusion in computer vision research field. Due to some information loss in occluded object, it＇s a challenging task to detect and recognize occluded object. First, this paper classifies the occluded object detection and recognition methods into three categories. Then, it summarizes and analyses the fundamentals, merits and drawbacks of the methods in details. Finally, it gives out the outlook of these methods.

关 键 词：遮挡目标 目标检测 目标识别 局部特征 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]