Compressing, coding and quantizing method

Abstract

The invention relates to a compressing, coding and quantizing method, comprising the following steps of: (a) setting a quantization macro block; (b) obtaining the prediction residual and the first quantization parameter of each pixel in the macro block; (c) respectively calculating SAD1 and SAD2 according to the prediction residual and the first quantization parameter; (d) comparing size of the SAD1 corresponding to the first quantization method and the SAD2, if SAD1<=SAD2, completing the coding, compressing and quantizing according to the SAD1, and otherwise, completing the coding, compressing and quantizing according to the second quantization method corresponding to the SAD2. According to the embodiment of the compressing, coding and quantizing method, the most suitable quantizing method is selected in two quantizing methods, so that the image loss in the bandwidth compressing and quantizing process is reduced, and the data transmission quantity is reduced.

Description

technical field [0001] The invention belongs to the field of compression coding, and in particular relates to a compression coding quantization method. Background technique [0002] Bandwidth compression belongs to low-complexity and low-magnification compression, so spatial domain quantization is more commonly used in it. In order to save bits for transmitting quantization parameters, the same quantization parameter is often used for all pixels in one MB. This has the following disadvantages: 1. For MBs that cross texture boundaries, due to the subjective characteristics of the human eye, using the same QP for different textures will not only cause subjective losses, but also cannot save the number of coding bits; 2. In MBs, the size of the prediction residual is different First, adopting the same QP for all pixels will enlarge the difference loss between original pixels and reconstructed pixels. If pixel-level quantization is used, although the above problems can be avoi...

AI Technical Summary

Problems solved by technology

This has the following disadvantages: 1. For MBs that cross texture boundaries, due to the subjective characteristics of the human eye, using the same QP for different textures will not only cause subjective losses, but also cannot save the number of coding bits; 2. In MBs, the size of the prediction residual is different First, using the same QP for all pixels will expand the difference loss between the original pixel and the reconstructed pixel

If pixel-level quantization is used, although the above problems can be avoided, since each pixel must transmit a quantization parameter, it will result in a large number of additional bits, which will not achieve the effect of compression

If the prediction residual is directly quantized, the loss generated by quantization is not further processed according to the position of the loss and the size distribution of the loss, and the quantization loss cannot be further reduced in the end.

Images