摘要
在基于无线传感器网络多目标跟踪数据融合系统研究的基础上,提出了改进的模糊聚类平均算法,并给出了相应的集中式数据融合整体方案.算法将每一批观测数据按照航迹估计位置的关联门限进行划分,然后分别对航迹和关联门限内的采集信息进行模糊关联,再把获得的最大关联度数据分配给各条航迹作为目标的实际位置.数据融合的思路是,删除所有关联门限内的数据,将剩余数据再进行航迹起始模块处理.模拟实验表明,所提算法可解决多目标跟踪的误跟、漏跟和振荡问题,数据融合方案既能保存有用信息,又能去除冗余数据,进而有效避免了漏跟和重复跟踪的问题.
According to the study on the data fusion problem of multi-target tracking in wireless sensor networks, an improved fuzzy cluster mean (FCM) algorithm is proposed and the corresponding centralized fusion scheme is provided. Firstly in the algorithm, each batch of sensed data is divided according to association threshold of the estimated position of the tracks. Then the fuzzy association is carried out for the collected data in the track and the corresponding threshold respectively, and the data that has the highest associated degree is assigned to each track and regarded as the practical position. The idea of the data fusion scheme is to delete all data within the association threshold and the track starting module is processed again for the rest of data. Simulation experiments show that the proposed scheme can solve the mistracking and surge problem of multi-target tracking, furthermore, it effectively avoids the problem of missing targets and re- peated tracking targets, since the scheme can not only save the usable data but also delete redundant data.
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2006年第10期1043-1046,1051,共5页
Journal of Xi'an Jiaotong University