CAJ | 학술논문

针对模型不准确时，传统Kalman滤波目标跟踪算法精度有限甚至发散的问题，研究了基于新息协方差在线匹配技术的自适应Kalman滤波算法，提高跟踪精度；并以Kalman滤波估计的目标位置为基础，利用一步Kalman预测得到下一时刻目标可能的位置范围，避免对整幅后帧图像进行遍历搜索，减小了计算量；为了避免存在干扰时异常量测对目标跟踪的影响，研究了量测信息异常检测算法，以Kalman预测的量测代替异常量测，增强抗干扰能力。实验证明，所提算法能够有效提高目标跟踪的精度和鲁棒性。
침대모형불준학시，전통Kalman려파목표근종산법정도유한심지발산적문제，연구료기우신식협방차재선필배기술적자괄응Kalman려파산법，제고근종정도；병이Kalman려파고계적목표위치위기출，이용일보Kalman예측득도하일시각목표가능적위치범위，피면대정폭후정도상진행편력수색，감소료계산량；위료피면존재간우시이상량측대목표근종적영향，연구료량측신식이상검측산법，이Kalman예측적량측대체이상량측，증강항간우능력。실험증명，소제산법능구유효제고목표근종적정도화로봉성。
An adaptive Kalman filter, based on the online matching of innovation covariance, is presented to overcome the problem of accuracy degrade or even divergence when there exists tremendous modeling errors and to improve the accuracy of target tracking. The area where the target may appear at the next epoch is predicted by one-step Kalman predictor,based the position of the target estimated by Kalman filter at present to avoide searching the whole image to find the target and to reduce the calculation burden. Abnormal measurement detection is also studied and the abnormal measurements are replaced by the Kaman predicted measurement,to avoid the disturbance caused by the abnormal measurement and to increase the anti-interference ability. Experimental results show that the accuracy and robustness of target tracking can be improved by the algorithm presented here.