Method and device for processing discrete cosine transform

Abstract

The invention discloses a method and a device for processing discrete cosine transform. The method for processing discrete cosine transform provides a device which corresponds to the method on the basis of the formula modification of a discrete cosine transform algorithm. The device forms a flow line structure by a preprocessing module, a main processing module and a post processing module. A main processing unit adopts a systolic array structure comprising a delay unit and an operation unit. The whole structure realizes higher parallelism and lower area-time complexity by the improvement of the algorithm and an idea of a systolic array. The invention can be applied in the fields of image signal processing, communication, and the like, particularly the internationally public coding standard, such as JPEG, MPEG, H26x, and the like.

Description

technical field [0001] The present invention relates to discrete cosine transform in the fields of image signal processing, communication and the like, in particular to a discrete cosine transform processing method and device thereof. Background technique [0002] Discrete cosine transform is an orthogonal transform method proposed by N.Ahmed et al. in 1974. It is often considered as the best method for transforming speech and image signals, and it is a quasi-optimal transform whose performance is close to K-L transform. DCT is an important part of H.261, JPEG, MPEG and other international public coding standards, and occupies an important position in the current audio and video coding. In view of the importance of DCT, its VLSI realization is very necessary, and the development of integrated circuit technology also provides the possibility for this. [0003] There are many definitions of DCT, and the DCT usually referred to is the second type of discrete cosine transform,...

AI Technical Summary

Problems solved by technology

Obviously, the above algorithm and structure are not ideal in terms of area time complexity. When we pay more attention to real-time performance, we need a structure with higher parallelism to reduce time complexity, so as to obtain a smaller Aτ or Aτ 2

[0067] At the same time, the above-mentioned algorithm and structure do not reflect the characteristics of pulsation structure rules, control and communication as a whole, and cannot take advantage of the pulsation structure itself.

Images