一种快速红外人体轮廓跟踪算法

Fast contour tracking algorithm for human body in infrared images

为实现红外图像序列中人体轮廓的精确跟踪,提出了一种基于快速水平集的新算法.首先,在目标区域及其邻近背景区域带上,而不是在整个图像平面上,采用模式分类中的最近邻决策思想来构建快速水平集算法的速度函数;然后,采用基于动态邻近区域的快速水平集来演化目标边界曲线以实现目标的轮廓跟踪.实验结果表明,该算法能适应目标尺度的变化、目标的分裂或合并,并获得人体的精确轮廓.

A novel algorithm based on the level set is proposed for tracking the precise contour of the human body in infrared videos. The algorithm first adopts the nearest neighbor decision method to construct the velocity expression for the fast level set only in the target region and its neighbor background region but not in the whole image plane, then contour tracking is realized by evolving the zero level set curve using the dynamic neighbor region fast level set algorithm proposed in this paper. Experiments show that this algorithm can adapt to the scale change of the target and the split or mergenee of the target, and lead to the exact contour of the human body.