为了提高彩色视频序列的压缩比，根据四维矩阵及四维矩阵离散余弦变换(4D-MDCT)理论，提出了一种基于内容统计的编码方法，即先对彩色视频序列所构成的四维矩阵进行子阵划分和四维矩阵离散余弦变换，然后对变换系数进行四维矩阵量化。为了能够充分利用彩色视频序列的帧间相关性和帧内相关性，又通过对量化后的四维子阵系数进行统计来找出其中完全相同的四维子矩阵。对于系数完全相同的子矩阵，只对第1个子矩阵进行编码，而不对其他的子矩阵进行编码，而且在对第1个四维子矩阵进行编码时，采用了传统的游程编码和熵编码。实验结果表明，在保持高信噪比的情况下，此方法可以大幅度地提高压缩比。

This paper presents a new coding method based on content statistics according to four-dimensional matrix and four-dimensional matrix DCT (4D-MDCT) theories in order to increase the compression ratio of color video sequence. First, four-dimensional matrix, which is constructed by color video sequence, is divided into submatrices and 4D-MDCT is performed to the submatrices. Then four-dimensional matrix quantization is applied to the transformed coefficients. In order to use the correlation between the inter color frames and intra color frames, statistics on quantized four-dimensional submatrix coefficients is performed to search four-dimensional submatrices which have identical quantized coefficients. Only the first one of these identical quantized four-dimensional submatrices is coded. When the first quantized four-dimension submatrix is coded, conventional runlength coding and entropy coding are performed. Experimental results proved that the compression ratio can be increased greatly and the high PSNR is preserved. Because conventional runlength coding and entropy coding are simple, and easy to be realized by using hardware, the method in this paper has a prosperous future.