基于冗余字典的多特征压缩感知目标跟踪算法  

Multi-feature Compressive Sensing Target Tracking Algorithm Based on Redundant Dictionary

作　　者：朱甦[1,2] 薄煜明[1] 何亮[1] 

机构地区：[1]南京理工大学自动化学院,江苏南京210094 [2]南京理工大学紫金学院电子信息与光电技术学院,江苏南京210046

出　　处：《兵工学报》2017年第6期1140-1146,共7页Acta Armamentarii

基　　金：国家自然科学基金项目(U1330133)

摘　　要：针对多特征压缩感知算法中,要求信号稀疏表示的基是一个正交矩阵的问题,提出了提取红外与可见光的多特征目标构造冗余字典子空间下的稀疏表示,分析了压缩感知算法中感知矩阵的选择和稀疏信号的重构。根据对信号稀疏表示的重构,提出粒子滤波框架下基于冗余字典的多特征压缩感知跟踪方法,能够自动检测复杂场景中出现的动态目标。实验结果表明,与其他经典算法相比,该算法在光照变化、相似外形的干扰目标遮挡等复杂场景中具有更好的鲁棒性及实时性。In consideration that the basis of signal sparse representation is an orthogonal matrix in the multi-feature compressed sensing algorithm, the multi-features of infrared and visible images are extracted to construct a sparse representation in a subspace of redundant dictionary, and the selection of sensing matrix and the reconstruction of sparse signal in the algorithm are analyzed. A redundant dictionary-based target tracking algorithm of multi-feature compressed sensing in the framework of particle filter is proposed by reconstructing the signal sparse representation, which can automatically detect dynamic targets in com- plex environment. Experimental results show that, compared with other classical algorithms, the proposed algorithm has better robustness and real-time in complex environment like illumination change and interference object occlusion.

关 键 词：信息处理技术 冗余字典 压缩感知 粒子滤波 目标跟踪 

分 类 号：TP391.413[自动化与计算机技术—计算机应用技术]