CAJ | 학술논문

针对摄像机运动情况下的多目标跟踪问题,提出了基于粒子滤波的跟踪算法.在粒子滤波算法基础上,将二阶自回归过程作为系统状态转移模型,HSV颜色直方图作为观测模型,对视频中多个目标的位置、大小进行跟踪.实验结果表明,该算法能实时正确地跟踪多个目标,并对局部遮挡有较好的鲁棒性,也能在目标短暂消失导致跟踪失败后,在目标重新出现后及时捕获并继续进行跟踪.
침대섭상궤운동정황하적다목표근종문제,제출료기우입자려파적근종산법.재입자려파산법기출상,장이계자회귀과정작위계통상태전이모형,HSV안색직방도작위관측모형,대시빈중다개목표적위치、대소진행근종.실험결과표명,해산법능실시정학지근종다개목표,병대국부차당유교호적로봉성,야능재목표단잠소실도치근종실패후,재목표중신출현후급시포획병계속진행근종.
A tracking algorithm based on particle filter is proposed to deal with multi-target tracking in a camera motion situation. In the proposed algorithm, the position and scale of targets in the video are tracked by using a second-order auto-regressive process as the transition model and an HSV histogram as the observation model. Experimental results show that the proposed algorithm can track multi-target correctly in real-time and is robust to the partial occlusion. In addition, it can recover from tracking failure due to the target out of camera for a while.