基于踪片Tracklet关联的视觉目标跟踪:现状与展望

发布时间：2018-11-03 12:24

【摘要】：近年来,由于计算机视觉技术的发展和计算机硬件性能的提高,基于视觉的目标跟踪方法得到了飞速的发展.其中,基于踪片(Tracklet)关联的目标跟踪方法因为具有对目标遮挡的强鲁棒性、算法运行的快速性等优点得到了广泛关注,本文对这类方法的最新研究进展进行了综述.首先,简明地介绍了视觉目标跟踪的基本知识、研究意义和研究现状.然后,通过感兴趣目标检测、跟踪特征提取、踪片生成、踪片关联与补全四个步骤,系统详尽地介绍了基于踪片关联的目标跟踪方法,分析了近年来提出的一些踪片关联方法的优缺点.最后,本文指出了该研究问题的发展方向,一方面要提出更先进的目标跟踪模型,另一方面要采用平行视觉方法进行虚实互动的模型学习与评估.
[Abstract]:In recent years, due to the development of computer vision technology and the improvement of computer hardware performance, visual based target tracking method has been rapidly developed. Among them, the tracking method based on tracer (Tracklet) association has been paid more and more attention because of its strong robustness to target occlusion and fast running of the algorithm. In this paper, the latest research progress of this kind of methods is reviewed. Firstly, the basic knowledge, significance and present situation of visual target tracking are briefly introduced. Then, through the four steps of object detection of interest, feature extraction of tracking, generation of tracer, correlation and complement of tracer, the method of target tracking based on tracer correlation is introduced in detail. The advantages and disadvantages of some tracer correlation methods proposed in recent years are analyzed. Finally, this paper points out the development direction of this research problem. On the one hand, we should put forward a more advanced target tracking model, on the other hand, we should use parallel visual method to study and evaluate the model of virtual reality interaction.
【作者单位】：中国科学院自动化研究所复杂系统管理与控制国家重点实验室;中国科学院大学;青岛智能产业技术研究院;国防科技大学军事计算实验与平行系统技术研究中心;
【基金】：国家自然科学基金(61533019,71232006,91520301)资助~~
【分类号】：TP391.41

【相似文献】