基于多阶段假设的目标实时检测  

Real-time Target Detection Based on Multistage Hypothesis

作　　者：蒲旭敏[1] 艾斯卡尔.艾木都拉[1] 向桂英[1] 

机构地区：[1]新疆大学信息科学与工程学院,乌鲁木齐830046

出　　处：《计算机工程》2011年第6期175-177,共3页Computer Engineering

基　　金：国家自然科学基金资助项目(60507005)

摘　　要：研究一种在三维时空中快速检测红外图像序列弱点状运动目标技术。利用目标的速度大小和方向初步删除噪声轨迹,从而减少计算量并提高检测的实时性。采用一种基于改进的多阶段假设的逐帧检测、逐帧删除噪声轨迹技术,获得弱点状运动目标运动的真实轨迹。实验结果表明,该方法计算量小,具有良好的实时性。A new technology of real-time detection on infrared dim moving point targets in image sequences is studied.Through speed and direction of small target,useless trajectories is cut,which reduces calculation amount and calculation time greatly.The technology based on multistage hypothesis test which detects the dim moving point targets and delete the trajectories of noises is used.It gets the targets more accurate.Experimental results show that the technology requires less calculation and has good real-time performance.

关 键 词：三维时空 实时性 目标运动信息 点目标检测 

分 类 号：TN911.73[电子电信—通信与信息系统]