基于单目视觉的移动光斑跟踪定位方法

Moving spot tracking and positioning mothed based on monocular vision

针对移动光斑跟踪定位中目标快速运动、严重遮挡导致光斑位置难以长时间跟踪的问题,提出了一种基于光斑跟踪模块、光斑中心定位模块和光斑轨迹预测模块三部分组成的单目视觉的移动光斑跟踪定位方法。其中,光斑跟踪模块是通过图像分割出目标光斑区域并以该区域初始化相关滤波器CN算法;光斑中心定位模块在跟踪框内通过灰度质心法求取光斑中心;光斑轨迹预测模块是通过自适应无迹卡尔曼滤波器(UKF)结合轨迹线性拟合来估计目标位置,在受干扰情况下校正CN跟踪器的搜索区域,提高系统抗干扰能力;利用实际移动光斑视频图像序列进行实验研究,对比试验结果表明所提方法的有效性。

In order to solve the problem that the fast movement and serious occlusion of the target lead to the difficulty in tracking the spot for a long time,a monocular vision tracking and positioning method is presented based on three parts:the tracking module of spot,the positioning module of spot centre,and the prediction module of spot track.The tracking module of spot realizes dynamic tracking of the spot using the colour features and geometric shape features of the spot to identify the target spot area,which initializes the correlation filter CN algorithm with this area.The positioning module of spot center divides the target in the tracking frame,and obtains the spot center by grayscale centroid algorithm.The prediction module of spot track adopts Unscented Kalman Filter(UKF)to estimate target position and corrects searching area of CN tracker in the case of interference,which improves the anti-interference ability of the system.Experimental study by the image sequences of moving spot,the results show the effectiveness of the proposed method.