一种改进的视频超分辨率重建方法

An Improved Method for Video Super-resolution Reconstruction

摘要:

针对目前无人机摄像、照相数据的特点,提出一种基于低分辨率视频重建高分辨率视频的改进方法；采用高斯金字塔光流算法对低分辨率视频及高分辨率图像进行运动估计利用小波频带分解方法提取图像的高频细节,并对视频进行运动补偿,采用凸集投影方法对补偿后视频进行迭代优化；并通过MATLAB仿真实现了以上算法,实验结果证明本文算法重建质量更好、处理速度更快。

Abstract:

For the characteristics of the UAV camera and camera data presently, a new improved method is proposed based on reconstructing the low-resolution video into high-resolution video. Primarily, adopting the Gaussian Pyramid Optical Flow algorithm, motion is estimated for low-resolution video and high-resolution image. Then, making use of the method of wavelet frequency band decomposition, the high-frequency details of image is extracted and then motion is compensated. Finally, solving the optimization of video compensated by proposed iteration steps with using the POCS method and realizing the above algorithm by MATLAB simulation.