短程线主动轮廓跟踪算法的研究——在复杂背景和非刚性运动目标跟踪中的应用

Research on Tracking Multiple Nonrigid Moving Objects in Clutter Background with Geodesic Active Contours

在复杂背景下对多个非刚性目标进行跟踪是计算机视觉中的一个难点 在短程线主动轮廓模型的基础上 ,利用力场正则化方法 ,并加入运动边缘信息 ,提出了一种在复杂背景下多个非刚性目标进行跟踪的方法 该方法由运动检测和跟踪两部分组成 :运动检测利用运动边缘信息对运动目标的运动做出检测 ,让轮廓曲线运动到目标轮廓附近 ;跟踪利用当前帧中的静态边缘信息对运动检测的结果加以修正 而跟踪这一步引入的偏差将在下一帧的运动检测中得到修正

Tracking nonrigid moving objects in clutter background is a difficult problem in computer vision On the basis of geodesic active contour models, incorporating with the motion edge information, and using the force field regularization method, a new framework for tracking multi nonrigid moving objects in clutter background is presented in this paper This framework consists of two stages: motion detection and tracking Using motion edge information, the contour is evolved to the region near the moving object boundary at the motion detection stage Then a tracking stage, only using the static edge information in current input frame, is considered to correct the motion detection deviation Furthermore, the tracking deviation will be corrected at the motion detection stage of the next frame Experiments demonstrate that the proposed approach is valid for tracking multiple nonrigid moving objects in clutter background