Two-dimensional discrete cosine transformation (DCT)/inverse DCT circuit and method

Abstract

The invention relates to a two-dimensional discrete cosine transformation (DCT) / inverse DCT circuit and method. The circuit comprises an input memory, a first multiplexer, a register list, a demultiplexer, a DCT / inverse DCT unit, a second multiplexer and an output memory, wherein the input memory, the first multiplexer, the register list, the demultiplexer, the DCT / inverse DCT unit, the second multiplexer and the output memory are sequentially connected, the DCT / inverse DCT unit comprises a register, a first sum-and-difference operation unit and a second sum-and-difference operation unit, the first sum-and-difference operation unit is used for carrying out sum-and-difference operation on the data in the register for once, the second sum-and-difference operation unit is used for carrying out two sum-and-difference operations on the shifted data in the register, the input ends and output ends of the first sum-and-difference operation unit and the second sum-and-difference operation unit are respectively connected with the register, the register is also connected with the demultiplexer and the second multiplexer. The circuit and the method have the advantage that an adder-subtractor and two shifters are used for replacing a multiplier, the particular coefficients are selected, and then the multiplier with more wasted resources and lower speed is not used in the hardware circuit, so a calculation unit of the hardware can be repetitively used.

Description

technical field [0001] The present invention relates to the FPGA hardware structure and transformation method of discrete cosine transform (DCT transform) and inverse discrete cosine transform (IDCT transform) used in image and video compression technology, and particularly relates to two-dimensional DCT transform for image data of 8×8 blocks Or a two-dimensional IDCT transform circuit and method thereof. Background technique [0002] Discrete cosine transform is one of the important transformations in the field of digital image processing and compression. It adopts the method of non-uniform quantization to reduce the high-frequency components in the image, eliminate the spatial redundant information in the image, and convert the image data into normalized , quantized and variable-length coded frequency-domain coefficients. After image information is transformed by DCT, most of the energy is concentrated on the DC component and a few low-frequency components, which are wide...

AI Technical Summary

Problems solved by technology

[0012] When the binDCT algorithm implements DCT transformation, although only adders and shifters are needed, the algorithm is divided into four stages, and each stage is executed sequentially. The structure of the transform circuit is still relatively complicated, and more adders and shifters are required. Bit register, the conversion speed and accuracy are also limited

Images