The following information represents the comprehensive notes from the ‘’Video Communication Course 2022 (Part4)’’. These notes provide the fundamental concepts, principles, and practical applications covered in the course. They serve as a valuable resource for students and professionals seeking to enhance their knowledge and skills in the field of video communication.
Part4 contains the following sections
- PSNR (Peak Signal-to-Noise Ratio) 峰值訊噪比
- Coding Delay
- Rate-Distortion of Lossy Algorithms
- Redundancy
- Intra Coding
- Rate Control
PSNR (Peak Signal-to-Noise Ratio) 峰值訊噪比
The higher the PSNR, the closer to the original image, and the lower the PSNR, the more distortion the image has.
What’s the difference between (peak value)/(error value) and (peak value)/(error value)²? One is divided by constant value, another is divided by square value.
Square will enlarge the the error, and make the error value has more influence on the PSNR value. e.g. peak value = 255, error value = 5
255/5 = 51
255/⁵² = 10.2
For PSNR higher value means less loss, if only divide a constant value can not notice the error very well.
Coding Delay
- Delay from compression, compression needs time.
- Some encoding algorithm doesn’t follow the sequence of the frame.
For example:
- original frame sequence: 1>2>3>4>5>6>7>8>9
- algorithm frame sequence: 1>5>9>2>3>4>6>7>8 it process 1, 5, 9 frame first, then 2, 3, 4 frame , then, 6, 7, 8 frame. First order frame: 1, 5, 9 Second order frame: 2, 3, 4, 6, 7, 8
Rate-Distortion of Lossy Algorithms
Distortion v.s Bit-rate
Distortion & PSNR are contrary (相反的)
Here shows some example condition.
- Fig. A
A better than B.
At the same bit-rate, A has lower distortion than B.
- Fig. B
B better than A.
At the same bit-rate, B has higher PSNR than A.
- Fig. C
A has higher complexity than B.
At the same PSNR value, A can use lower bit-rate than B to achieve the same image quality. Which means that A’s algorithm is more complex than B’s.
Redundancy
Spatial redundancy 空間上的冗餘資訊
In the same frame, there is usually a strong correlation between adjacent pixels, so The relevance of is spatially redundant information.
在同一張幀之中，相鄰的像素之間通常有很強的關連性，這樣的關連性即為空間上的冗餘資訊。
- DPCM (differential pulse code modulation) 誤差訊號編碼
- DCT (discrete cosine transform) 離散餘弦變換
- Subband (sub-band coding) 次頻帶編碼
- Wavelet (wavelet analysis) 小波分析
Temporal redundancy 空間上的冗餘資訊
In video data, there is usually a strong correlation between adjacent frames, and such correlation is temporal redundant information.
在視訊資料中，相鄰的幀（frame）與幀之間通常有很強的關連性，這樣的關連性即為時間上的冗餘資訊。
- DPCM (differential pulse code modulation) 誤差訊號編碼
- MC/ME (Motion Estimation/Compensation)
Statistical redundancy 統計上的冗餘資訊
Statistically redundant information means that the probability distribution of symbols to be encoded is non-uniform.
統計上的冗餘資訊指的是欲編碼的符號（symbol）的機率分布是不均勻（non-uniform）的。
- RLC (run-length coding) 遊程編碼
- VLC (Variable-length code) 可變長度程式碼
Perceptual redundancy 感知上的冗餘資訊
Perceptual redundant information refers to information that cannot be detected by human eyes when people watch video.
感知上的冗餘資訊是指在人在觀看視訊時，人眼無法察覺的資訊。
- SQ (Scale Quantization)
- VQ (Vector Quantization)
- fractal (Fractal compression) 分形壓縮
Intra Coding
Transform Coding
Transform: use different way to perform same signal (e.g. linear algebra 線性代數)
Discrete Cosine Transform (DCT) 離散餘弦轉換
It can be seen from the formula that there is a close relationship between the two 8*8 tables of pixel value (left) and coefficient (right)
As long as the value of one of the squares changes, it will affect the value of the entire (or part of) the other table.
After DCT, A transfer is better than B. A is more centralize (集中化). Which means from 0 to certain frequency, A has more data than B.
Quantization Table
Y Component Quantization Table is a experimental results 實驗得出的結果, only the up-left corner “DC coefficient value” 16 is define on purpose.
Rate Control 位元控制率
The bit rate refers to the data flow per unit time. The higher the bit rate per unit time, the smaller the compression ratio of the file and the better the picture quality; conversely, the lower the bit rate, The larger the compression ratio, although the file size can be saved, but the quality of the video will be sacrificed. Therefore, the bit rate is directly proportional to the video quality, but the size of the video file will also increase as the bit rate increases.
The higher the bit rate is not always better, because the higher image quality is difficult for the human eye to detect small differences, but it will increase the size of the video file, so the encoding focuses on how to use the lowest bit rate to achieve the least distortion. Adopting an appropriate bit rate can be said to strike a balance between “quality first” or “economic efficiency”.
If bandwidth is limit,
- Bigger data use high DCT compression
- Smaller data use low DCT compression
MPEG-1 use fixed marcoblock 16*16. Newer technique such as H.264…etc, they can use variable marcoblock. Variable marcoblock: 128*128, …, 16*16)
Bigger marcoblock can save bit-usage.
- Complex region use small marcoblock (16x16)
- Simple region use big marcoblock (128x128)
VBR — Variable Bitrate 可變位元速率
It means that there is no fixed bit rate, the encoder can analyze the picture, provide more bit rate in the more complex part of the picture, and reduce the bit rate where the image is easy to compress, so as to ensure that the control file is a reasonable size Under the circumstances, high-quality images can still be taken into account. Variable bit rate is suitable for most productions, especially if there are many randomly moving particles in the picture, such as a lot of raindrops or heavy snow, a higher bit rate can ensure that the picture will not be overly compressed and reduce the image quality.
CBR — Constant Bitrate 固定位元速率
It means that the video uses a fixed bit rate to compress the picture. This is a compression method based on the premise of a fixed file size. It is suitable for static main videos with simple backgrounds and few changes, such as live videos of Internet celebrities. Using a fixed bit rate ensures a steady stream of video streaming without stuttering or buffering.
Thank you for taking the time to read this article, and I sincerely hope that the information provided proves to be valuable to you. Whether you are a student, professional, or simply someone interested in video communication, it is my utmost wish that these notes enhance your understanding and contribute to your success in this field. Thank you once again, and best of luck on your journey in the world of video communication!
