在基于Mean-Shift的目标跟踪算法中，尺度自适应机制是算法研究的一个重要方向。一种典型的方法采

用Lindeberg的尺度空间理论以获取目标尺度信息。但现有算法中将尺度由2．D矢量压缩为1一D量，未能精细地

刻画目标仿射变换时的尺度变化，从而限制了算法的适用范围。为此，该文将尺度维1．D滤波推广至2一D，构造得

到了相应的垂D尺度空间，并利用窄间维和尺度维的Mean-Shift交替迭代，实现了同时在空间位置和尺度方向对

目标的有效跟踪，提高了算法在目标尺度变化时的自适应性，并扩大了算法的适用范围。

关键-Scale adaptive mechanism in the target tracking algorithm based on Mean-Shift algorithm is an important direction of research. A typical method using Lindeberg' s scale space theory to obtain the target scale. However, the existing algorithms will scale from 2. D vector compression for 1-D volume, not fine to portray the target affine transformation of scale changes, thus limiting the scope of application of the algorithm. To this end, the paper will scale dimension. D filter to the 2-D structure has been the vertical D scale space and the use of narrow inter-dimension and scale dimension of the Mean-Shift alternating iteration, effective tracking of the target position in space and scale direction to improve the self-adaptive algorithm when the target scale changes, and to expand the scope of application of the algorithm. Key