论文部分内容阅读
基于DCT域内系数量化的压缩算法 ,如MPEG ,JPEG等 ,随着码率的降低 ,量化也越来越粗糙 ,使重建后的图象出现“块效应” ,振铃效应 ,视觉模糊等缺陷 ,本文提出一种混合算法去除这些缺陷来有效地提高主观质量。首先采用自适应边界滤波 ,再将DCT块分为平坦区域和边缘区域 ,并针对二区域的不同特性采用非线性空间移变滤波来去除图象缺陷
Compression algorithms based on DCT coefficient quantization, such as MPEG, JPEG and so on, with the decrease of the code rate, the quantization is more and more rough, which makes the reconstructed image appear the defects of “block effect”, ringing effect and visual obfuscation, This paper presents a hybrid algorithm to remove these defects to effectively improve subjective quality. Firstly, the adaptive boundary filter is adopted, and then the DCT block is divided into a flat area and an edge area. Non-linear spatial transform filtering is adopted to remove image defects according to different characteristics of the two areas