CAJ | 학술논문

仅利用单一特征的目标跟踪难以克服光照和目标形变等外部条件变化，给出一种基于协方差区域描述子的粒子滤波，协方差描述子可融合目标区域内的多种特征处理复杂背景下的目标跟踪问题，提高跟踪的鲁棒性；针对粒子滤波计算量大的问题，引入积分协方差矩阵计算，提高跟踪的实时性，对比实验表明新的跟踪算法比仅用单一特征跟踪鲁棒性更高，处理速度更快。
부이용단일특정적목표근종난이극복광조화목표형변등외부조건변화，급출일충기우협방차구역묘술자적입자려파，협방차묘술자가융합목표구역내적다충특정처리복잡배경하적목표근종문제，제고근종적로봉성；침대입자려파계산량대적문제，인입적분협방차구진계산，제고근종적실시성，대비실험표명신적근종산법비부용단일특정근종로봉성경고，처리속도경쾌。
The target tracking only using single feature is difficult to overcome the influence of external conditions such as the illumination and target deformation etc.,a particle filter target tracking algorithm based on the covariance region descriptor is proposed.The covariance descriptor can fuse different features of the target region to handle target track-ing under complex background.And then,the tracking robustness is improved.Moreover,aiming at the problem that calculation of particle filter is large,the integral covariance matrix computation is introduced to Bayesian tracking framework,which makes the tracking process realtime.The comparative experiments show that the proposed algorithm is more robust and faster than the single feature tracking.