基于速度估计的双Hough变换运动轨迹检测算法  

Velocity estimation-based combinational Hough transform algorithm for moving point target detection

作　　者：胡静[1] 张天序[1] 

机构地区：[1]华中科技大学多谱信息处理技术国家级重点实验室,湖北武汉430074

出　　处：《华中科技大学学报（自然科学版）》2013年第1期106-110,共5页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：武器装备预研基金资助项目(9140A01060110JW0515);华中科技大学自主创新基金资助项目(2011QN073)

摘　　要：提出一种速度估计约束下的双Hough变换算法,并用于序列图像运动点轨迹检测.对于像面投影近似直线的匀速运动点轨迹,首先进行速度预估计,然后在此基础上进行连续2次不同空间的Hough变换.第1次Hough变换将疑似目标点由图像空间X-Y根据极坐标方程转换到ρ-θ空间,在初步筛选疑似二维轨迹后,引入时间维信息,将疑似目标点投影到T-L空间.第2次Hough变换由T-L空间根据斜截式方程转换到V-S空间,再利用最初的速度预估计值限制投影范围,进一步剔除虚警.仿真结果表明:所提算法对目标运动速度的相关先验知识要求较少,在噪声较强的条件下,依然能取得较好的检测结果,且算法开销更少.A combinational Hough transform with velocity estimation was proposed to detect the uniform linear moving targe.After pre-estimating the moving point velocity,two combinational Hough transforms（HT）were implemented in different space.In the first HT,the points in the image space were transformed to the ρ-θ space according to the polar coordinate equation.Then these points were projected to T-Lspace with they temporal information after a preliminary choosing for trajectories in the space X-Y.In the second HT,points in the T-Lspace were transformed to the V-Sspace according to the slope-intercept equation.Finally the projection of these points were limited with their estimated velocity,and the rate of false alarm was further reduced.Experimental results show that the algorithm needs less prior information of the moving point target and can operate in suppressing noise points with less computational cost.

关 键 词：图像处理 运动点目标 轨迹检测 目标检测 HOUGH变换 匀速直线运动 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]