Abstract: In this paper a novel tracking feature selection method is presented. It assumes the features that have best distinctiveness between object and background are also best for tracking the object. A two-class variance ratio is employed to measure the distinctiveness. Particle swarm optimization algorithm is used to optimize the different features combination to generate the best tracking feature. To demonstrate the proposed method, selected feature is combined with kernel-based tracking method. Experimental results show that the proposed method can robustly track moving object in low discriminately background scenario.

Key words: target tracking, tracking features selection, particle swarm optimization, kernel-based tracking

摘要： 针对复杂背景下的运动目标跟踪特征选择问题，提出了一种基于粒子群优化的目标跟踪特征选择算法。假设具有目标与背景间最好可分离性的特征为最好的跟踪特征。通过构建目标与背景的图像特征分布方差的比值函数作为衡量目标与背景间的可分离性判据。使用粒子群优化算法优化不同的特征组合实时获取最优的目标跟踪特征。为验证该算法的有效性，将选择的最优特征与一种基于核的跟踪算法相结合进行跟踪实验。实验结果表明，算法能有效提高传统基于核的跟踪算法对于复杂场景下的运动目标跟踪的鲁棒性与准确性。

关键词: 目标跟踪, 跟踪特征选择, 粒子群优化, 基于核的跟踪算法