DCT域中任意比例的图像上下采样算法

在许多实际应用中, 为了满足传输信道和终端显示设备的要求, 需要通过上采样和下采样来改变图像的尺寸. 压缩域中的图像上下采样可以在空域中进行, 然而, 直接在压缩域中实现将更为快速. 本文根据空域中块与子块的相互关系以及酉变换对矩阵乘法的分配率, 提出了 DCT 域内任意比例图像上下采样算法. 与现存的算法相比, 本算法具有较高的信噪比, 较低的运算复杂度, 并适用于帧内与帧间编码的不同情况, 可应用于不同视频编码转码的实时处理.

An Arbitrary Resizing Algorithm of Image in the DCT Domain

In many applications, image resizing is needed to change the scale of digital images. Resizing of a compressed image can be performed in a spatial domain. However, the performance will be much faster when in a compressed domain. In this work, an algorithm for arbitrarily resizing images in DCT domain based on the distributive property of unitary transformation to matrix multiplication is advanced and the relationship between blocks and sub-blocks is addressed. This algorithm has higher PSNR and lower computation costs than the conventional ones and is applicable to both intracoding frames and intercoding frames. When combined with motion compensation algorithm in DCT domain, it can achieve real time video transcoding.