改进的KCF红外空中目标跟踪方法  

Improved KCF infrared aerial target tracking method

作　　者：郑武兴 王春平[1] 付强[1] 

机构地区：[1]军械工程学院电子与光学工程系,河北石家庄050003

出　　处：《激光与红外》2017年第12期1553-1558,共6页Laser & Infrared

摘　　要：针对红外空中目标跟踪中遮挡导致的跟丢问题,结合核相关滤波跟踪算法(KCF)无法跟踪尺度的缺点,提出一种基于KCF的红外空中目标跟踪方法。该方法借助KCF估计的目标位置,增加Sobel算子提取目标扩展区域的边缘信息,从而获取目标尺度并进行二次定位。然后,判断目标是否受遮挡或跟丢,采用帧差法重新检测目标,确保目标的持续跟踪。在7个视频序列上对所提的算法进行了实验,结果显示,所提算法能够有效计算尺度,验证了多种环境下跟踪的有效性。平均跟踪速度达到44 f/s,能够满足实时性要求。因此,本文提出的方法对于红外空中目标的跟踪具有一定的实用意义。Aiming at the problem of target lost caused by occlusion in infrared aerial target tracking, combined with the shortcoming that kernel correlation filter（KCF） tracking algorithm is unable to estimate object scale, a method of in- frared aerial target tracking method based on KCF is proposed. In this method, the Sobel operator is used to extract the edge information of the target extended region by using the target position estimated by KCF, so as to obtain target scale and perform secondary positioning. Then, to judge whether the target is occluded or lost, frame difference method is used to detect the target to ensure the continuous tracking. The proposed algorithm is tested for 7 video sequences, and the results show that the proposed algorithm can effectively calculate the scale and verify the effectiveness of tracking in various environments. The average tracking speed is up to 44 frames per second, which can meet the real - time requirements. The proposed method has certain practical significance for infrared aerial target tracking.

关 键 词：红外目标跟踪 核相关滤波 边缘检测 帧差法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]