轨迹预测在红外多目标跟踪中的应用  

Application of trajectory prediction on infrared multi-targets tracking

作　　者：罗寰[1] 于雷[1] 廖俊[1] 来强[1] 

机构地区：[1]空军工程大学工程学院,陕西西安710038

出　　处：《红外与激光工程》2009年第3期397-401,共5页Infrared and Laser Engineering

基　　金：军队重点科研项目基金(KJ06090)

摘　　要：针对红外多目标跟踪过程中的噪声、轨迹预测和轨迹交联问题,讨论了轨迹预测算法在红外多目标跟踪中的应用。首先,利用邻域轨迹预测算法剔除目标检测算法中误检的噪声点,然后运用卡尔曼滤波和最小二乘滤波的轨迹预测方法解决在跟踪过程中目标丢失的问题,及多目标轨迹出现交联时如何辨识出各个目标轨迹的问题。算法充分运用了目标运动的连续性和方向性,避免了噪声的干扰,实现了目标轨迹的预测和辨析。通过物理实验,验证了各个算法的有效性,并比较了工程应用中的优劣。实验证明:基于目标轨迹预测的算法较好地解决了红外多目标跟踪过程中的噪声、轨迹预测和交联问题。Aiming at the problems of noise disturbing,trajectory prediction and crosslinking in the process of infrared multi-target tracking, the application of the trajectory prediction algorithmic in infrared multi-target tracking was discussed.First, adjacent-regions trajectory prediction algorithm was used to eliminate the noise points generated by error detection. Then the trajectory prediction based on Kalman f＇llter and LS filter was adopted to solve the problem of target missing in the tracking process,and identify the different target trajectory while crosslinking occured.This algorithm avoided the interference of the noise, realized prediction and discrimination of target trajectory by using the continuity and directivity. Experiment verifies the effectiveness of each algorithmic.The advantage and disadvantage on engineering application are compared.The results illustrate that it perfectly solves the problems of noise disturbing, trajectory prediction and crosslinking in the process of infrared multi-target tracking.

关 键 词：多目标跟踪 预测算法 邻域卡尔曼滤波 最小二乘 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]