Image encoding/decoding method and apparatus therefor

Abstract

An image encoding method includes generating a predictive signal and encoding mode information according to each encoding mode from a macroblock signal corresponding to each macroblock, selecting a quantization code table corresponding to each macroblock, generating a predictive error signal for each encoding mode based on the macroblock signal and the predictive signal, subjecting the predictive error signal to orthogonal transformation, quantizing the orthogonal-transformed predictive error signal while changing a quantization parameter for every plural sub-pixel-blocks, using the quantization code table corresponding to the macroblock, encoding quantization transformation coefficient, calculating an encoding cost, selecting one encoding mode based on the encoding cost, selecting one quantization code table based on the encoding cost, and encoding information of an index indicating the selected quantization code table for every frame of the input image signal or every region of the frame.

Description

technical field [0001] The present invention relates to an image encoding / decoding method and device for encoding / decoding video or still images. Background technique [0002] In recent years, by combining ITU-T and ISO-IEC, a video coding method (ISO / IEC 14496-10 : 2003 information technology-Coding of audio-visual objects-Part 10: Advanced Video Coding) (called H.264). Conventional coding systems, such as ISO / IEC MPEG-1, 2, 4, ITU-TH.261, H.263, utilize 8*8 blocks of two-dimensional discrete cosine transform (DCT) for compression. On the contrary, since the H.264 coding system uses a two-dimensional integer orthogonal transformation of 4*4 blocks, it does not need to consider IDCT mismatch, and can use 16-bit registers to realize the operation. [0003] Moreover, the H.264 coding system can realize orthogonal transformation processing and quantization processing through addition and bit shifting. Similar to the orthogonal transform and quantization processing described ...

AI Technical Summary

Problems solved by technology

Then, there is a problem that the transform coefficient in the small pixel block of the macroblock cannot be finely controlled

[0022] Further, when using a given quantization matrix to determine and dequantize quantization parameters for each macroblock through the method specified in the H.264 high profile, there is a problem that the The prediction block shape and the prediction method to control the coefficient of variation

Images