一种从低分辨率图像序列获取高分辨率图像的算法

在某些应用领域,常常需要得到目标的高分辨率图像,考虑到在这些应用场合中,往往可以获得对同一景物或目标的多帧图像,本文提出了一种基于亚像素级图像配准与类似于中值滤波插值的从多幅低分辨率(LR)图像中获取一幅高分辨率(HR)图像的算法。算法首先采用梯度方法计算出LR图像之间的位移量,经过图像配准后,每个HR像素点被赋予其作用域内所有LR像点值的中值。仿真结果表明,该算法简单有效,既能提高图像分辨率,又能较好地去除非线性噪声。

An algorithm for extracting high resolution image from low resolution frames

In some applications, high resolution images of the targets are often required. Since in such applications we can obtain an image sequence of the same target, an algorithm based on registration with sub-pixel accuracy and interpolation similar to median filtering is represented to extract a High Resolution (HR) from a set of Low Resolution (LR) frames. The unknown displacements are computed from the LR image sequence using gradient method. After image registration, each HR pixel is assigned a value which is the median of all LR pixel values in its effective region. Computer simulations demonstrate that this algorithm is simple but effective. It can not only enhance resolution but also remove non-linear noise.