基于自适应分层结构的压缩分布场跟踪算法
发布时间：2019-04-20 19:27
【摘要】：为了提高分布场跟踪算法的运算效率,增强其在复杂背景下的鲁棒性,提出基于自适应分层结构的压缩分布场跟踪算法.该方法充分考虑目标区域像素值分布情况,引入k-means算法对首帧标记的目标区域进行聚类分析,根据聚类结果自适应的产生分布场结构.针对分布场模型维数较高的缺点,融合压缩感知方法对分布场进行压缩,降低模型维数,提高算法效率.此外,改变原始分布场跟踪算法采用的局部搜索跟踪策略,利用随机抽样的方式来提高算法跟踪精度.实验结果表明,提出的算法与当前流行的跟踪算法相比,具有更好的表现.
[Abstract]:In order to improve the computational efficiency of distributed field tracking algorithm and enhance its robustness in complex background, a compressed distributed field tracking algorithm based on adaptive hierarchical structure is proposed. In this method, the pixel value distribution in the target region is fully considered, and the k-means algorithm is introduced to cluster the target region marked by the first frame. According to the clustering results, the distribution field structure is generated adaptively. In view of the high dimension of distributed field model, the compression sensing method is used to compress the distributed field, reduce the dimension of the model and improve the efficiency of the algorithm. In addition, the local search and tracking strategy used in the original distributed field tracking algorithm is changed and random sampling is used to improve the tracking accuracy of the algorithm. Experimental results show that the proposed algorithm has better performance than the current popular tracking algorithm.
【作者单位】： 国家数字交换系统工程技术研究中心;
【基金】：国家自然科学基金(No.61379151,No.61521003) 国家科技支撑计划(No.2014BAH30B01) 河南省杰出青年基金(No.144100510001)
【分类号】：TP391.41
本文编号：2461875
[Abstract]:In order to improve the computational efficiency of distributed field tracking algorithm and enhance its robustness in complex background, a compressed distributed field tracking algorithm based on adaptive hierarchical structure is proposed. In this method, the pixel value distribution in the target region is fully considered, and the k-means algorithm is introduced to cluster the target region marked by the first frame. According to the clustering results, the distribution field structure is generated adaptively. In view of the high dimension of distributed field model, the compression sensing method is used to compress the distributed field, reduce the dimension of the model and improve the efficiency of the algorithm. In addition, the local search and tracking strategy used in the original distributed field tracking algorithm is changed and random sampling is used to improve the tracking accuracy of the algorithm. Experimental results show that the proposed algorithm has better performance than the current popular tracking algorithm.
【作者单位】： 国家数字交换系统工程技术研究中心;
【基金】：国家自然科学基金(No.61379151,No.61521003) 国家科技支撑计划(No.2014BAH30B01) 河南省杰出青年基金(No.144100510001)
【分类号】：TP391.41
【相似文献】
相关硕士学位论文 前1条
1 叱干鹏飞;基于分布场模型的目标跟踪方法研究[D];西北农林科技大学;2014年
，本文编号：2461875
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2461875.html
