Abstract: Usually template matching algorithm is utilized to deal with the floating and orientation of the picture．But it easily creates strenuous disturbance，when using it in the war to anti-tremble，because of panzer video tracking system is influenced by the land form and the ground object.It is inevitably to generate matching mistakes.This paper founds the image moving model firstly，then improves the template matching algorithm according to the gray character of the image.It uses the characteristic points to correction and remove the pseudo-characteristic points.This method can cancel the wrong-matching and meet the need of the quality of the war-image.

Key words: anti-tremble, template matching, characteristic points

摘要： 在图像消旋消抖处理中通常运用到模板匹配算法，但在战场目标抗抖动的运用中，由于装甲战车车载图像跟踪系统受到地形地物的影响，会产生剧烈扰动，会产生匹配错误。建立了图像运动模型，根据图像灰度特征，对模板匹配算法进行改进。根据几何距离约束特征点的性质，运用特征点校正去除伪特征点，较好地消除误匹配，达到战场图像消旋消抖的要求。

关键词: 抗抖动, 模板匹配, 特征点