Huaguang Yuneng analyzes the 4:4:4 in image compression and what is going on

In recent years, friends in the industry have often seen or heard concepts such as 4:4:4, but many people are confused and confused about the inner meaning. Let’s talk about 4:4:4 today.

Visually, a video signal is broken down into two signal elements, luma and chrominance. The color sampling method refers to the allocation method of how many pixels are used to record brightness and color difference signal information from a 4-pixel area (2 vertical pixels and 2 horizontal pixels) . That is to say, various color sampling methods are described by the ratio between the luminance value and the chrominance value transmitted when a line is scanned, and then described in the form of 4:X:Y, X and Y are in each two chrominance channels The relative number of values ​​of .

For example, the following picture:

The color sampling method is detailed in the table below:

The format with the highest resolution of the chrominance signal is 4:4:4, that is to say, for every 4 consecutive sampling points Y sampling on each scanning line, there are corresponding 4 points Pb/Cb and 4 points Pr/ Cr samples. In other words, in this format, the resolution of the chrominance signal is the same as that of the luma signal. This format is mainly used in video processing equipment to avoid image quality degradation during processing.

Therefore, 4:4:4 is the highest sampling recording and output method with full bandwidth and full chroma. Its bandwidth and color gamut are superior to other methods such as 4:2:2/4:2:0.

Although the 4:4:4 sampling method can guarantee the highest image quality, it takes up huge bandwidth. The price of 4:4:4 transmission equipment and playback equipment will be higher than other equipment.

And 4:2:2 is to perform 2:1 compression processing on the actual chroma bandwidth 4:4:4.

And 4:2:0 is to read B—Y (Pb/Cb) in odd rows of pixels, and read R—Y (Pr/Cr) color sampling method in even rows. Although the pixel used to read B—Y (Pb/Cb) and R—Y (Pr/Cr) information in 4 pixels is only 1 pixel, but in order to distinguish it from 4:1:1, it is marked as 4: 2 : 0.

The most obvious difference between 4:2:0 and 4:2:2 is that there are gaps in the color transition area, and the color is not as delicate and full as 4:2:2.

Color sampling is a method of selectively recording color information by ignoring the color difference signal of some pixels.

In the last decade, video engineers have discovered that the human eye is less sensitive to chroma than to luminance, and the eye distinguishes light from dark more finely than hue and color saturation. It is by using the characteristic of human vision to compress the signal amount by reducing the method of reading color-difference signal pixels, which can save signal bandwidth and power.

By selecting an appropriate color model, the bandwidth occupied by the two color-difference signals can be significantly lower than the bandwidth of Y without significantly affecting viewing of the reproduced color image.

Since 2014, Hangalaxy has been focusing on the optical communication industry, long-distance transmission of optical fiber lines, and professional optical fiber transmission lines in the audio and video industry at home and abroad. It has more than 10,000 large, medium and small engineering cases, with excellent quality standards and The accumulated efforts have condensed into the trust of the market and the reputation of the brand. The word "reliable" is the word of mouth recognized by the industry for the Hangalaxy brand.

                          

 

Guess you like

Origin blog.csdn.net/hgynHangalaxy/article/details/123821475