CAJ | 학술논문

基于视觉图像的飞行器目标跟踪中，一般要面临目标严重遮挡和目标大尺度形变等问题。为了实现这种情况下的准确目标定位和目标跟踪，提出采用融合卡尔曼滤波器的改进SOAMST算法：一方面利用卡尔曼滤波的递推与估计能力来解决目标运动过程中目标被遮挡的问题；另一方面使用基于大津阈值的改进SOAMST算法来处理运动目标的大尺度形变以实现目标的准确定位。实验结果表明，该算法能够处理运动目标的严重遮挡，实现对运动目标进行准确的定位和跟踪。
기우시각도상적비행기목표근종중，일반요면림목표엄중차당화목표대척도형변등문제。위료실현저충정황하적준학목표정위화목표근종，제출채용융합잡이만려파기적개진SOAMST산법：일방면이용잡이만려파적체추여고계능력래해결목표운동과정중목표피차당적문제；령일방면사용기우대진역치적개진SOAMST산법래처리운동목표적대척도형변이실현목표적준학정위。실험결과표명，해산법능구처리운동목표적엄중차당，실현대운동목표진행준학적정위화근종。
In aircraft tracking based on visual images, serious target occlusion and large-scale deformation usually occur. In order to get precise position and target tracking in this case, an improved SOAMST algorithm adopted Kalman filter is put forward: on one hand, to solve the block problem of moving targets by recurrence and estimation ability of Kalman filter;on the other hand,to deal with large-scale deformation of moving targets by the improved SOAMST algorithm based on ostu threshold to achieve the goal of precise po-sitioning. Experimental results show that the proposed method can handle severe occlusion of moving targets, realizing the accurate po-sitioning and tracking.