一种旋转运动估计稳像算法的研究与实现

Research and Implementation of Image Stabilization in a Rotary Motion Estimation Algorithm

针对摄像机在拍摄视频图像时存在抖动问题,包括平移运动和旋转运动,提出一种旋转运动估计稳像算法.该旋转运动估计稳像算法主要以灰度投影算法为基础,并对灰度投影算法进行了一定的改进,用于估计旋转角度矢量,解决了视频图像序列的较小角度旋转运动;同时提出采用分块灰度投影算法,解决了由于存在局部运动物体所造成的稳像精度问题.实验表明,该方法能够有效解决视频图像序列的旋转问题,通过FPGA实现的电子稳像系统,能够满足实时性要求,且能够输出稳定流畅的实时视频,从而提高视频的质量.

In this paper, the presence of jitter problems when shooting video image for the cameras, including translational and rotational movement, proposed a rotary motion estimation algorithm for image stabilization. The rotary motion estimation image stabilization algorithm mainly gray projection algorithm based on gray projection algorithm and some improvements, is used to estimate the angle of rotation vector to solve the small-angle video image sequences rotary motion; also proposed the use of block gray projection algorithm to solve the problem of image stabilization accuracy due to the presence of moving objects caused localized. Experimental results show that this method can effectively solve the problem of video image sequences rotate through FPGA implementation of electronic image stabilization system that can meet the real-time requirements, and can output stable and smooth real-time video, thereby improving the quality of the video.