基于JPEG序列的图像重建

该文提出了一种基于JPEG序列的图像重建方法。该方法在已有的单帧图像复原技术的基础之上，依据超分辨率重建的思想，将凸集投影(POCS)理论与迭代反投影(IBP)算法相结合，在频域内降低量化误差，修复离散余弦系数。此外，它还利用了最大后验概率(MAP)估计以及相应优化算法的特点，在去除高斯噪声的同时，保护边缘和细节信息。实验结果表明，该方法一方面能够抑制高比率压缩所造成的块效应和振铃效应，另一方面能较好地恢复图像的细节部分，有效地提高图像的清晰度。

Recovery of JPEG Compressed Image Sequences

A method for the recovery of JPEG compressed image sequences is proposed in this paper. The method is based on the reduction of compression artifacts, and it is inspired with the thoughts of super-resolution reconstruction. To minimize the quantization errors and restore the discrete cosine transformation components, the method utilizes the theory of Projection Onto Convex Sets (POCS) and Iterative Back Projection (IBP) algorithm, which helps computing the projections. It also applies Maximum A Posteriori (MAP) estimation techniques to remove Gaussian noise without damaging edges or details of the image. Experimental results show that the proposed method produces good Peak Signal to Noise Ratio (PSNR) results, and could not only remove blocking and ringing artifacts, but also recover details of images, especially in low bit-rate conditions.