基于多特征Mean-Shift的灵巧弹药末制导跟踪算法  

A Target Tracking Algorithm of Smart Munition Based on Multiple Features

作　　者：聂振钢[1,2] 孙协昌[3,2] 梁银川 

机构地区：[1]北京理工大学信息与电子学院,北京100081 [2]天地一体化信息技术国家重点实验室(筹),北京100086 [3]航天恒星科技有限公司,北京100086

出　　处：《微电子学与计算机》2016年第5期89-91,96,共4页Microelectronics & Computer

基　　金：天地一体化信息技术国家重点实验室(筹)开放基金(2014_CXJJ-YG_11);航空基金资助项目(2012ZC51035)

摘　　要：针对灵巧弹药在复杂背景下跟踪目标的问题,将跟踪目标颜色信息和边缘信息进行融合,提出了一种基于多特征的Mean-Shift跟踪算法,从而解决了复杂背景下跟踪目标的鲁棒性问题.应用色差空间表示跟踪目标的颜色信息,同时引入边缘信息表示跟踪目标的纹理信息,从而可以有效解决复杂背景下Mean Shift跟踪鲁棒性问题.应用色差空间代替色调空间,并用反映射方法求解加权值,从而节省了运算复杂度,满足实时性要求.实验结果表明,改进算法跟踪性能较经典Mean Shift法具有明显提高,满足工程应用需求.For the accurate target tracking of smart munitions＇ under complex background, a improved Mean Shift tracking algorithm based on color features and edge feature is proposed to improve the mbusmess of the Mean Shift algorithnx Due to a color-difference space is introduced to represent color features and edge inforrmfion is introduced to represent texture feature, the mbusmess of Mean Shift algorithm is significantly enhanced. Using color-difference space instead of hue space and using back-projection method to resolve weighted coefficients, the complexity of algorithm is reduced and the real-time dermnd is meet. Experimental results show that compared with the state-art Mean Shift algorithm, the proposed multiple feature Mean Nlift algorithm has improved the tracking performance and suited to engineering implementation, autonomous landing.

关 键 词：灵巧弹药 Mean SHIFT算法 目标跟踪 鲁棒性 

分 类 号：V556.5[航空宇航科学与技术—人机与环境工程]