摘要
在分析Mean shift实质的基础上,针对灰度直方图目标模型描述能力较弱的问题,引入基于灰度-边缘特征融合的目标模型,将kalman滤波器用于目标运动估计。通过研究互信息量与目标尺寸变化的关系,建立了基于kalman-Mean shift的特征融合自适应跟踪算法模型。经过试验证明该算法有利于增强直方图目标描述能力,能够适应目标尺寸的变化,提高Mean shift算法的鲁棒性。
In this paper,because the ability of using the gray level histogram to describe target model was weak,though the analysis of the Mean shift essence,target model based on gray-edge features fusion was introduced.Kalman filter was used to estimate target motion,through the study of mutual information and the relationships of target size changes,adaptive kalman-Mean shift tracking algorithm model based on feature fusion was established.Through the experiment,it showed that this algorithm was beneficial to strengthen histogram goal description ability,was able to adapt to the change of the target size,and improve the robustness of Mean shift algorithm.
作者
罗航
尹逊青
刘晨
许彩
LUO Hang;YIN Xunqing;LIU Chen;XU Cai(Navy Representative Office of Wuhan,Department of the Navel Equipment,Wuhan 430000,China;The 5th Military Representative Office in Wuhan,Navy Representative Office of Wuhan,Wuhan 430000,China;The 3rd Military Representative Office in Wuhan,Navy Representative Office of Wuhan,Wuhan 430205,China)
出处
《直升机技术》
2020年第4期6-11,15,共7页
Helicopter Technique
