管道内壁视频图像防抖动算法研究

为了减小抖动对管道内壁视频图像的影响，提出一种在抖动比较小的条件下将横向抖动与旋转分开处理的算法。首先通过选取合适的阈值将灰度化的全景图像二值化为一个圆，利用重心法求出圆心位置，根据与基准帧圆心位置的差别平移全景图像，从而减小横向抖动的影响。之后进行展开，通过求取连续两帧展开图像上重合区域灰度加权重心的差别，对当前帧展开图像平移后再进行拼接，从而减小旋转的影响。实验表明该算法处理速度非常快，可以比较好地减小管道内壁视频图像抖动的影响。

Research on Video Image Stabilization Algorithm for Pipeline Inner Wall

To eliminate the influence of camera shaking upon pipeline inner wall,a new stabilization algorithm is proposed,which processes crosswise shift and rotation separately under conditions that camera shaking is not too large.Firstly,each panoramic gray image is translated into a binary image of circle by selecting an appropriate threshold,the center of the circle is obtained by the center-of-gravity-method.Then the panoramic image is shifted by comparing the position to that of the base frame,and thus the influence of crosswise shift is decreased.Secondly,the processed panoramic image is unfolded,the gray weighted centers of the overlapping region of every two continuous frames are calculated,then the unfolded image is shifted according to the difference between the centers,and thus the influence of rotating is decreased.Experiment shows that the algorithm can eliminate the influence of camera shaking quickly and perfectly.