基于相关滤波扩展记忆系统模型的智能目标跟踪

将目标跟踪过程看作一个多重记忆系统模型,提 出了基于相关滤波的扩展记忆系统模型,实现了基 于记忆系统模型的智能目标跟踪。首先,通过提取跟踪目标特征学习目标信息,生成短时相 关滤波器,产 生短时记忆；然后利用每一帧短期记忆的不断重复与更新,产生长时记忆,生成长时相关滤 波器。短时与 长时记忆构成相关滤波记忆系统模型,完成目标跟踪。在此模型基础上,分析与挖掘模型中 的相关滤波数 据,加入四种智能化控制信息,构建扩展记忆系统模型,实现智能化的目标跟踪。基于相关 滤波的扩展记 忆系统模型利用生物记忆的原理使目标跟踪更加自动化、智能化,增强目标跟踪的准确性。 实验结果表明, 与当前流行的相关滤波跟踪算法相比,本文算法提高了目标跟踪的抗干扰性、抗遮挡性与抗 形变能力,同时保证了在尺度跟踪的有效性。

An extended memory system model based intelligent tracker with correlation filte ring

The whole tracker is regarded as a multiple memory system model.An ex tended memory system model based intelligent tracker with correlation filtering is presented in this paper. Firstly,short-term filter is generated by extracting the target features,and short-term memory is generated.Secondly,t he repeated and updated short-term memory of each frame produces long-term memory,and long-term filter is genera ted.Short-term and long-term memory constitute the memory system model with correlation filtering.By analyzi ng and mining the filter data in the modal,four intelligent control information is added into the model,which e xtends the model to achieve intelligent and efficient target tracking.The extended memory system model base d tracker used the principle of biological memory to make target tracking more automatic and intelligent,which improved the accuracy of target tracking.Experimental results show that the proposed algorithm improves the ant i-interference,anti-occlusion and anti-deformation of target tracking,and ensures the effectiveness of scale tra cking.