摘要
针对背景感知算法未与目标的时空域特性建立联系,以及无法准确处理遮挡、形变等异常跟踪情况的问题,提出了能够动态感知时空异常的目标跟踪算法。在相关滤波器训练过程中引入动态空间正则项,使其与样本的时空域特性建立联系;结合响应图的峰值唯一性和锐利信息,提出异常感知方法;利用历史滤波器具有不同置信度的特点以及目标在时域中的连续性,通过异常感知方法自适应选择高置信度的历史滤波器作为时间正则化的参考模板,降低滤波器退化的风险。在OTB50、OTB100和TC128测试基准上进行仿真实验,该算法能够适应外观变化、画面杂乱等复杂条件下的跟踪任务,具有较强的鲁棒性和实用性。
In view of the fact that the background perception algorithm does not establish a relationship with the spatio-temporal domain characteristics of the target,and cannot accurately deal with the occlusion,deformation and other abnormal tracking,a object tracking algorithm which can adaptively perceive the spatio-temporal anomalies is proposed.In the training stage of correlation filter,the adaptive spatial regularization term is introduced to establish a relationship with the spatio-temporal characteristics of sample.The abnormal perception method is proposed according to the peak value of response map.Taking advantage of the different confidence of historical filter and the continuity of target in the time domain,the historical filter with high confidence is adaptively selected as the reference template of time regularization through the abnormal perception method,which reduces the risk of filter degradation.Simulation experiments carried out on OTB50,OTB100 and TC128 test benchmarks show that the algorithm can adapt to the tracking tasks under complex scenarios such as appearance changes and messy pictures,and has strong robustness and practicability.
作者
邱云飞
卜祥蕊
张博强
Qiu Yunfei;Bu Xiangrui;Zhang Boqiang(College of Software,Liaoning Technical University,Huludao 125105,China)
出处
《系统仿真学报》
CAS
CSCD
北大核心
2024年第2期338-351,共14页
Journal of System Simulation
基金
国家自然科学基金(6217022980,61404069)。
关键词
目标跟踪
相关滤波器
异常感知
滤波器退化
动态时空正则化
object tracking
correlation filter
abnormal perception
filter degradation
adaptive spatiotemporal regularization
