一种低比特率JPEG编码图像质量改进算法

JPEG图像编码系统当比特率较低时,常会产生“方块效应”,即重建图像的子块边缘不连续.用空间的低通滤波器对重建图像中每个子块的边界像素进行滤波,也就是通过块间的平滑处理可以消除“方块效应”,但滤去了图像边缘的高频分量而使边缘变模糊.通过对未经量化的离散余弦变换(DCT)系数进行预测,提供了一个后处理系统来减轻“方块效应”的影响且减少了边缘的模糊度.实验表明,该算法与低通滤波相比,在减轻“方块效应”的同时,保留了高频分量,改善了信噪比,图像质量明显改善.

An Improved JPEG Algorithm at Low Bit Rate for Image Coding

At low bit rate,a JPEG image coding system frequently produces block artifact It means that the edges of the sub blocks of the reconstructed image are not continuous.However, lowpass filter often drowns the high frequency, and the edges of the image become vague.In this paper we calculated the unquantified discrete cosine transform coefficients, and provided a subsequent processing system to eliminate the block artifact without blurring the edges The experiment results show that,compared with the lowpass filter method,the algorithm proposed here lessens the block artifact,preserves high frequency edges, improves the signal noise rate and the object effect.