Research On Video Compression Method Based On 3-D SDCT

With the development of digital technology, video information has been widely applied in people's life. But video data is very enormous, the high transfer rate and huge storage capacity will be the biggest obstacle to widely apply digital video if the video data will not be compressed. So digital video compression technology is the desirable means to solve this obstacle, and has become one of the key technologies of multimedia communication. At present, video compression technology commonly adopts interframe motion compensation and 2-D DCT, which has become the core part of many video compression standards, such as MPEG and H.26x. But this method involves motion estimation and motion compensation, the required calculation is considerably large.With the development of information technology, the 3-D DCT is used more and more widely, and has become a most potential video compression method. But 3-D DCT is popularly operated by three times 1-D DCT to rows, columns, and pages of 3-D signal continuously. This 3-D DCT operation can not reflect the overall space properties of 3-D transform very well, the physical properties of transform coefficient are difficult to understand, and 3-D DCT has not uniform matrix expression form. In addition, the quantization and scanning order of 3-D DCT transform coefficient have much effect on video compression result, and the 3-D DCT video coding still has not a highly effective quantization table. Therefore, this paper gives up the traditional 3-D DCT operation, and proposes 3-D SDCT (Submatrix Discrete Cosine Transform) video compression method.First, the basic theory of 3-D SDCT is derived and proofed in detail. Several new 3-D matrix operation methods are defined, transform basic matrix used to 3-D SDCT is constructed, the operation process of 3-D SDCT is described, and the properties of 3-D SDCT are analyzed. The theoretical analysis shows that the 3-D SDCT operation method makes 3-D DCT concise to express, easy to understand, convenient to operate.Second, the theoretical framework of 3-D SDCT video compression method is introduced systematically. The fundamental basis of 3-D SDCT video compression is discussed. Especially, the popular method of key steps about 3-D DCT video compression are described, such as quantization principle of 3-D DCT coefficients, quantization methods, scanning order of the quantized coefficients and entropy coding. Third, the basic principle of 3-D integer SDCT is derived in detail. The operation method of 3-D integer SDCT is discussed base on the operation analysis of 1-D integer DCT, which main ideal is token the 3-D video data as a whole, and replace DCT floating-point transform matrix with integer transform matrix. The performance of 3-D integer SDCT is analyzed, and the consistency of video compression effect between 3-D integer SDCT and 3-D DCT is checked by experiment.Fourth, the quantization and scanning order of 3-D SDCT are studied. The relations between 3-D SDCT and scanning order is analyzed, a novel method about non-diagonal quantization and non-diagonal scanning is designed according to the characteristics of video data. On this basis, five quantization and scanning methods are proposed in the paper. The comparable experimental investigation on the compression effect of the five proposed methods and traditional compression methods was carried out. Comparative analysis of the experimental results shows that the compression effect of the proposed methods is improved markedly.