基于动态Snake模型的机械手运动轨迹视觉跟踪

针对机械手运动在图像序列空间的轨迹分布，提出一种基于时空轨迹线的动态Snake跟踪模型.定义相应的能量函数，可使其在机械手轨迹分布上取得极小，通过Snake能量的轨迹收敛实现对机械手运动点的跟踪定位.利用轨迹能量系数的动态调节，可避免Snake搜索过程陷入局部极小．使用平方轨迹最小二乘预测器对轨迹点位置进行预测，可提高Snake搜索的实时性和准确性．微装配机械手运动实验证明了该模型及跟踪算法的有效性.

Visual Tracking of Robotic Manipulator Motion Trajectory Using Dynamic Active Contour Model

A dynamic active contour model for spatio-temporal trajectory is proposed to track robotic manipulator motion in image sequence space. The model employs a snake function with minimum energy at the trajactory distribution. The manipulator is localized through searching the converged energy point set. Energy weights in the function are real-time adjusted to avoid local minima during convergence. To improve the snake searching efficiency, a quadratic-trajectory least square estimator is developed to predict motion position before tracking. Experimental results of tracking micromanipulator motion demonstrate the effectiveness of the proposed model.