Research Of Fast Walsh Transform Algorithm Based On Multi-dimensional Vector Matrix

Multimedia communication is the mainstream communication now and in the future years, the development prospect of video information exchange in which is broader.Theoretically, multimedia communication reflects the information processing and communication theory progress; Technically,multimedia communication and information reflects the expansion of business and value-added services of broadband communication network. For example, relative to the second of communication services, almost all of the value-added services are multimedia services in the third generation mobile communication services. Image and video as an important component of multimedia services, occupy an increasingly important position in today's communication. From the arrival of 3G era and the upcoming 4G era, the traditional voice communication is already a very weak function, and video calls and voice mail and other new business is the mainstream, strong visual impact, quick and direct video call be more pervasive and rapid development. However, the amount data of video and image is huge, which need to be more efficient compression to meet the storage and transmission requirements.The traditional color video compression method process the Y frame, U frame and V frame, and the time frames separately which does not fully take into account the intra and the correlation between frames, the compression coding efficiency is not high. In order to improve the rate of color image and video compression, we use the following two methods:first, use the concept of multi-dimensional vector matrix took by our laboratory, express the color video in a four-dimensional model, which make the time frame in a unified model, fully eliminate the time redundancy between different time frames. Secondly, a combination of multi-dimensional vector matrix theory and fast Walsh transform, expand fast Walsh transform to the multi-dimensional, and then compress video. This is not only combines the advantages of multi-dimensional vector matrix theory, but also combines the advantages of fast Walsh transform, not only compress effectiveness is good, speed also have been significantly improved. There is still a significant speed improvement which can be widely applied to the video in real-time applications. In addition, Feng Hua's matrix theory based on multi-dimensional vector orthogonal Walsh transform vector quantization has been improved, using a scalar quantization method to overcome the disadvantage of computational complexity of vector quantization, the compression speed has been more substantialy increased.Finally, C is as a programming language in Visual C++6.0 environment, the programming based on multi-dimensional vector matrix of the fast Walsh transform algorithm research is achieved. In the experiment, select the standard video library of video, size is 352×288, YUV ratio 4:2:0,CIF format color video. Thus, counting 30 frames per second frame rate, the compression process of a frame must be controled within the 33ms, in order to meet the requirements of real-time video. We used 4×4×4,8×8×8 and 16×16×16 sub-division method for array experiments for this video, after the data block, we carry the block data fast Walsh transform based on multi-dimensional vector, and analyze the experimental results. The results from the three groups can be seen that the proposed method on the one hand to ensure the quality of reconstructed video, on the other hand the video compression speed has been more substantially increased. Therefore, the proposed method is feasible. In quantitative terms, the paper also take the experiment of 8×8×8 block method, Feng Hua used for vector quantization has been improved, this article uses a scalar quantization, vector quantization to fully overcome the disadvantage that computational complexity and consuming a long time. Finally, giving the total time including block, fast Walsh transform, scalar quantization, and anti-scalar quantization, inverse fast Walsh transform, and anti-block. The data indicators show that the method used in this paper meet the requirements of the video in real time. Also summarizes the advantages and disadvantages of the algorithm and some research areas which can be improved in the further.