基于SIFT和卡尔曼滤波的电子稳像技术研究

摘要

Using electronic image stabilization method to remove jitter video noise is one of the developing directions in image stabilization technology. This article first introduces the electronic image stabilization process. Considering the presence of scaling in video, it uses SIFT to extract feature points. After motion consistency estimation calculates the motion parameters, it uses Kalman filter to get the correction moving vector of the frame. The results imply that the method can effectively remove the random component of movement while maintaining the intention movement component to obtain stable video.%利用电子稳像方法去除抖动视频中的噪声是目前稳像技术的发展方向之一.文中首先介绍电子稳像的基本流程,随后使用SIFT提取图像中的特征点,以解决图像中存在缩放的问题.在经过运动一致性估计后计算运动参数,并使用Kalman滤波得到各帧的修正矢量.实验结果表明该方法能有效地去除随机运动分量而保持意向运动分量,得到稳定的视频.