基于卡尔曼滤波的序列图像超分辨率重建算法

针对给定一序列低分辨率图像重建出一序列高分辨率图像的问题,应用卡尔曼滤波理论,将时间轴加入重建算法,把利用多幅低分辨率图像生成高分辨率图像的模型转化为利用当前低分辨率图像和上一幅高分辨率图像来构建当前高分辨率图像.通过引入信息对的概念对算法运算复杂度进行改进.实验证明与已有算法相比该方法有较好的时效性和实用性.

A IMPROVED SUPER-RESOLUTION RECONSTRUCTION ALGORITHM OF IMAGE SEQUENCE BASED ON KALMAN FILTER

This paper presents a method for the problem of super-resolution reconstruction of continuous image sequence when given a low-resolution image sequence.In the reconstruction algorithm,we bring in the time axis,so that use the Kalman filter to turn the model of reconstruction by low-resolution images into the one of reconstruction by the low-resolution image at present and the previous high-resolution image.And improve the algorithm by the information pair which reduce the complexity.Experiment results show the way is time limited effectively and practically.