DCT域视频图像转码的下采样算法

DCT Domain Video Transcoding Downscaling Algorithms

研究了DCT域视频图像转码的下采样算法,包括DCT域帧内编码帧I帧的两种下采样算法———线性内插法和DCT系数截断法,以及DCT域帧间编码帧P帧的下采样算法———DCT域反运动补偿和DCT域运动补偿算法。对stefan视频序列第一帧(I帧)分别采用两种下采样算法进行转码下采样,通过峰值信噪比(PSNR)值的比较说明线性内插法转码下采样的效果要优于DCT系数截断法。采用DCT域反运动补偿算法实现在DCT域重构stefan视频序列第四帧(P帧)的帧内数据。

DCT domain video transeoding downsealing algorithms are investigated, including intra flame （Ⅰ frame） downsealing-linear interpolation and DCT coefficient decimation and inter frame （such as P frame） downsealing-inverse motion compensation and motion compensation on the DCT domain . Linear interpolation and DCT coetticient decimation are used for downsealing the first frame of Stefan video sequence （ Ⅰ frame）. Linear interpolation is better than DCT coetticient decimation comparing with their Peak Signal -to -Noise Ratio （PSNR）. Rebuilding the fourth frame of Stefan video sequence （P frame） is realized by using inverse motion compensation.