一种基于几何约束和直线特征的目标跟踪算法

An Object Tracking Algorithm Based on Multiple Geometric Constraints and Line Features

平面目标跟踪在于如何对各种复杂环境下的运动物体进行有效迅速准确的跟踪。为提高移动视点视频中目标跟踪的稳健性,提出一种基于多重几何约束和直线特征的目标跟踪算法。该算法充分利用平面目标的边缘几何特性,结合候选直线周围像素构建特征向量,将相似度匹配最高的作为跟踪结果,再采用单应性变换实现边缘的校正。实验结果表明该算法较Ferns在遮挡和部分可见场景中更具有鲁棒性。

Planar object tracking aims to effectively,quickly and accurately track moving objects in various complex environments.In order to im⁃prove the robustness of object tracking in moving viewpoint videos,an object tracking algorithm based on multiple geometric constraints and line features is proposed.The algorithm uses the geometric features of the edge of the planar object,and combines with the pixels in the neighborhood of the candidate line to construct the feature vector.The line with the best similarity is regarded as the final result,then uses the homography transformation to realize the edge correction.The experimental results show that the proposed method is more robust than the algorithm called ferns in occlusion and partially visible scenes.