一种补偿平移与旋转运动的快速电子稳像算法

A Fast Electronic Image Stabilization Algorithm for Translational and Rotational Motion Compensation

针对视频图像序列的非稳特性,研究了电子稳像算法中灰度投影算法和块匹配算法各自的不足之处,提出了一种快速补偿视频图像序列间平移及旋转运动的稳定成像算法.该算法先采用灰度投影算法估计并补偿视频图像序列间的平移运动,再利用拉普拉斯变换在靠近图像的边缘区域选取几个具有明显特征的小块,运用块匹配算法进行匹配,计算并补偿其旋转运动量,以得到稳定的视频图像序列.通过理论分析和实验验证,表明这种稳像算法具有速度快、准确度高的特点.

In consideration of the motion between image sequences, a research on some disadvantages of the gray projection algorithm and the block matching algorithm in electronic image stabilization is carried out. And a new method of image stabilization is presented in order to quickly solve the problem of the translational and rotational motion between image sequences. First, the translational motion is calculated and compensated based on the gray projection algorithm. Second,near the side of the image,several blocks with distinguishing characteristic is selected by Laplacian Transform, then, the rotational motion is also calculated and compensated by block matching algorithm so as to get stable image sequences. Through theoretical analysis and experimental results, this method is proved to be rapid and accurate.