一种新的形状自适应DCT算法

ＭＰＥＧ－４协议里的形状自适应ＤＣＴ算法计算简便，易与现存的混合编码标准相容，但存在一些不足．提出了一种新的对任意形状图像块作形状自适应ＤＣＴ算法．该算法利用少量低频ＤＣＴ系数近似表示任意形状图像块．与ＭＰＥＧ－４协议中的标准形状自适应ＤＣＴ算法相比较，在复原图像质量和图像数据压缩比等方面均有改进．理论分析与实验结果表明，该算法可以成为视频对象边界处理的方法之一

New Shape-Adaptive DCT Algorithm

Object based representation is one of the key technologies of MPEG 4, so shape adaptive coding becomes an important component of MPEG 4. The shape adaptive DCT algorithm in MPEG 4 has low complexity and can be easily incorporated into the existing hybrid coding standards, but it has some disadvantages. A new shape adaptive DCT algorithm suitable for coding pixels in arbitrarily shaped image segments was presented, which was based on representing arbitrarily shaped image blocks by some low frequency coefficients. Compared with the SADCT algorithm in MPEG 4, it achieves some improvement in the quality and compression performance of reconstructed images. It is shown by the theoretical analysis and experiment results that this algorithm can be used for the coding of visual objects' boundaries.