基于关键点匹配的多策略尺度自适应跟踪算法

Multi-strategy scale adaptive tracking algorithm via keypoint matching

针对目标尺度变化问题,提出一种结合关键点匹配算法的多策略尺度自适应算法。提出一种中心—边缘区域比较法判断目标尺度变化情况,依据不同的尺度变化情况采取不同的尺度调整策略;引入SURF关键点匹配算法,对目标与模板进行特征匹配,构建仿射矩阵,自适应调整窗口尺度;提出一种基于关键点匹配的目标重定位方法,克服窗口漂移问题。实验结果表明,该算法具有较好的实时性和尺度自适应性。

As the main challenge for target tracking is target scale change,a multi-strategy scale adaptive method was we presented to solve the problem.A center-edge region comparison method was proposed to determine the change of target scale,and different adjustment was made according to different scale change.SURF algorithm was introduced to construct affine matrix to adjust window scale adaptively.Furthermore,a target relocation method based on keypoint matching was proposed to prevent from losing target.Experiments demonstrate favorable performance of the proposed tracking algorithm.