基于POCS算法的块效应消除方法研究

Research of Reducing Blocking Artifacts Method Based on POCS Algorithm

基于分块的离散余弦变换(BDCT)在众多的国际图像、视频压缩标准中是一种主要的压缩算法,它的一个主要不足就是在低比特率下会出现明显的块效应。而凸集投影法(POCS)作为一种经典的图像恢复算法,在去除块效应的效果上并不是很理想。本文针对一点,对其进行了改进,设计了基于图像边缘的点扩散函数(PSF),提出了基于像素的方向性和边缘灰度级相结合的选择过程。实验结果表明,该算法能够有效地消除低码率图像中的块效应,同对图像中的边缘细节有一定的恢复作用。

The block -based Discrete Cosine Transform （BDCT） has been a main compression technique in mass of international standards for images and video compression. One of the major drawbacks in BDCT is visible blocking artifacts at low bit rates. However, Projections onto Convex Sets （POCS） , as a classical algorithm of image restoration, can not availably reduce the blocking artifacts. In this paper, aiming at mending it, Point Spread Function based on the edge of images has been designed, in addition, and a choice process based on the direction of pixels, combining the gray levels of image edges, has been presented. The experiment results indicate that this algorithm can availably reduce the blocking artifacts in the images at low bit rate, at the same time, the edge details in the images can be restored at a certain extent.