Method of controlling encoding rate, method of transmitting video data, encoding rate controller for video encoder, and video data transmission system using the encoding rate controller

• US 20050036698A1
• Filed: 06/16/2004
• Published: 02/17/2005
• Est. Priority Date: 08/12/2003
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: Beom, Jae-Young
• Granted Patent: US 7,426,309 B2
• Time in Patent Office: Days
• US Class Current: 382/239
• CPC Class Codes: H04N 19/126 Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
• CPC Class Codes: H04N 19/149 by estimating the code amount by means of a model, e.g. mathematical model or statistical model
• CPC Class Codes: H04N 19/152 by measuring the fullness of the transmission buffer
• CPC Class Codes: H04N 19/159 Prediction type, e.g. intra-frame, inter-frame or bidirectional frame prediction
• CPC Class Codes: H04N 19/172 the region being a picture, frame or field
• CPC Class Codes: H04N 19/176 the region being a block, e.g. a macroblock
• CPC Class Codes: H04N 19/177 the unit being a group of pictures [GOP]
• CPC Class Codes: H04N 19/196 being specially adapted for the computation of encoding parameters, e.g. by averaging previously computed encoding parameters processing of motion vectors H04N19/513
• CPC Class Codes: H04N 19/197 including determination of the initial value of an encoding parameter H04N19/56 takes precedence
• CPC Class Codes: H04N 19/198 including smoothing of a sequence of encoding parameters, e.g. by averaging, by choice of the maximum, minimum or median value
• CPC Class Codes: H04N 19/61 in combination with predictive coding

Abstract

A method of controlling an encoding (bit) rate and a method of transmitting video data, and an encoding (bit) rate controller for a video encoder and a video data transmission system employing the methods, wherein the number of bits is allocated to a current picture on the basis of previous encoding results without defining a relation between an encoding rate and the distortion, the limited number of bits is not forced when features of the current picture are different from those of the previous pictures, and a quantizer scale is set adaptively to various features of the current picture without using an additional number of bits corresponding to variation of the quantizer scale, so that it is possible to improve the picture quality of video data to be displayed on a monitor by using such real time variable bit-rate control.

Claims

1. A method of controlling an encoding rate of a video encoder, the method comprising the steps of:

(a) calculating a target number of bits-per-picture of a current picture type by using the average number of encoding bits-per-picture and the average target number of bits-per-picture of each picture type I, P, and B of a previous group of pictures (GOP) and a buffer balance value;

(b) calculating a quantizer scale value corresponding to a first macroblock-row of the current picture type by using the average quantizer scale value of a previous picture of the same type as the current picture and using the target number of bits;

(c) calculating a target number of bits per unit row and a number of encoding bits per unit row in a previous row for each macroblock-row of the current picture type, calculating a quantizer scale value corresponding to the current row by using functions of the target number of bits-per-picture and the number of encoding bits per unit row, and calculating a total number of encoding bits-per-picture of the current picture type;

(d) storing the quantizer scale value calculated in step (c), and outputting the quantizer scale values of the previous macroblock-rows and of the previous picture types; and

(e) calculating the buffer balance value by using the total number of encoding bits-per-picture and the average target number of bits-per-picture of the current picture type.

2. The method according to claim 1, wherein the target number of bits-per-picture is calculated by using the following equations:

$\begin{array}{c}{\mathrm{Tavg}}_{i,p,b}=\frac{B_{i,p,b}·<br><br>\left(N_i+N_p+N_b\right)}{B_i·<br><br>N_i+B_p·<br><br>N_p+B_b·<br><br>N_b}·<br><br>\frac{\mathrm{bit\_rate}}{\mathrm{frame\_rate}}\\ T={\mathrm{Tavg}}_{i,p,b}-\frac{B_{i,p,b}·<br><br>\left(N_i+N_p+N_b\right)}{B_i·<br><br>N_i+B_p·<br><br>N_p+B_b·<br><br>N_b}·<br><br>\frac{d}{N_{\mathrm{dis}}}\end{array}$

where T is the target number of bits-per-picture;

Tavg_i,p,b is the average target number of bits-per-picture for each picture type;

B_i,p,b is the average number of encoding bits-per-picture for each picture type;

N_i, N_p and N_b are the number of I, P and B pictures in a group of pictures, respectively;

bit_rate is a predetermined transmission rate;

frame_rate is a frame rate;

d is the buffer valance value; and

N_dis is the number of dispersed pictures.

3. The method according to claim 2, wherein the buffer balance value is calculated by using the following equation:

d+=S_i,p,b−

Tavg_i,p,b

where d+ is the incremental change of the buffer balance value d, and S_i,p,b is the total number of encoding bits of one picture for each picture type.

4. The method according to claim 2, wherein the average number of encoding bits-per-picture B_i,p,b for each picture type is updated by using the following conditional expression:

If (B_p<

1.2·

B_b) B_p=1.2·

B_b;


If (B_i<

1.2·

B_p) B_i=1.2·

B_p;


If (B_i>

3.5·

B_p) B_i=3.5·

B_p;


If (B_b<

0.6·

B_p) B_b=0.6·

B_b.

5. The method according to claim 2, wherein the quantizer scale value corresponding to the first macroblock-row of the current picture type is calculated by using the following equations:

$\begin{array}{c}{Q}_{\mathrm{init}}=Q_{i,p,b}+\frac{\mathrm{\gamma <br><br>}}{\mathrm{\delta <br><br>}}\\ \mathrm{\gamma <br><br>}=\frac{{\mathrm{Tavg}}_{i,p,b}-T}{{\mathrm{Tavg}}_{i,p,b}}\\ Q_{i,p,b}=\frac{1}{N_{\mathrm{row}}}·<br><br>\underset{j=1}{\overset{N_{\mathrm{row}}}{\sum <br><br>}}<br><br>{\mathrm{mquant}}_j\end{array}$

where Q_i,p,b is the average quantizer scale value;

, mquant_j is a quantizer scale value corresponding to a j-th macroblock-row;

N_row is the total number of rows in a picture;

Tavg_i,p,b is the average target number of bits-per-picture for each picture type;

T is the target number of bits, Q_init is a quantizer scale value corresponding to a first macroblock-row; and

δ

is a constant.

6. The method according to claim 5, wherein the quantizer scale value corresponding to each macroblock-row of the current picture type is calculated by using the following equation associated with the functions:

If (f(j,S_j,S<sub>j−

1</sub>, T_row)>

β

_up_j), then mquant_j+1=mquant_up(mquant_j,q_scale_type);


else if (g(j,S_j,S<sub>j−

1</sub>,T_row)>

β

_—

dn_j), then mquant_j+1=mquant_—

dn(mquant_j,q_scale_type);


else mquant_j+1=mquant_j.

where S_j is the number of encoding bits from the first macroblock-row to a j-th macroblock-row in a picture;

T_row is a target number of bits per unit row;

mquant_up(mquant_j, q_scale_type) is a quantizer scale value corresponding to a code larger than a previous k-th code corresponding to mquant_j in any one of a plurality of q_scale_types;

mquant_dn(mquant_j, q_scale_type) is a quantizer scale value corresponding to a code smaller than the previous k-th code corresponding to mquant_j in any one of a plurality of q_scale_types; and

β

_up_j and β

_dn_j are first and second critical constants, respectively.

7. The method according to claim 6, wherein the functions of f and g include an expression representing the number of encoding bits in the previous row.

8. The method according to claim 6, wherein the mquant_up(mquant_j, q_scale_type) and the mquant_dn(mquant_j, q_scale_type) are a quantizer scale value corresponding to a (k+1)-th code larger by one than a previous k-th code corresponding to mquant_j in any one of a plurality of q_scale_types, and a quantizer scale value corresponding to a (k−

1)-th code smaller by one than the previous k-th code corresponding to mquant_j in any one of a plurality of q_scale_types, respectively.

9. A method of transmitting video data, the method comprising steps of:

(a) receiving a quantizer scale value corresponding to a current macroblock-row of a current picture type, and encoding video data in accordance with the quantizer scale value;

(b) calculating the quantizer scale value corresponding to the current macroblock-row by using a target number of bits per unit row and the number of encoding bits per unit row in the previous macroblock-row for the encoded video data; and

(c) outputting the encoded video data in accordance with a transmission rate.

10. The method according to claim 9, wherein the step of (b) comprises sub-steps of:

(b-1) calculating the target number of bits-per-picture of the current picture type by using an average number of encoding bits-per-picture and an average target number of bits-per-picture of each picture type I, P and B of the previous group of pictures for the encoded video data and a buffer balance value;

(b-2) calculating the quantizer scale value corresponding to the first macroblock-row of the current picture type by using the average quantizer scale value of the last picture out of the previous pictures of the current picture'"'"'s type, and the target number of bits-per-picture;

(b-3) calculating the target number of bits per unit row and the number of encoding bits per unit row in the previous row for each macroblock-row of the current picture type, calculating the quantizer scale value corresponding to a current row by using functions including the target number of bits per unit row and the number of encoding bits per unit row, and calculating a total number of encoding bits-per-picture of the current picture type;

(b-4) storing the quantizer scale value output in the sub-step of (b-3), and outputting the quantizer scale values of the previous macroblock-rows and the previous picture types; and

(b-5) calculating the buffer balance value by using the total number of encoding bits and the average target number of bits-per-picture of the current picture type.

11. An encoding rate controller for a video encoder comprising:

a bit allocation unit which calculates a target number of bits-per-picture of a current picture type by using an average number of encoding bits-per-picture and an average target number of bits-per-picture of each picture type I, P and B of a previous group of pictures (GOP) and a buffer balance value;

a rate control unit which calculates a quantizer scale value corresponding to a first macroblock-row of the current picture type by using an average quantizer scale value of the last picture out of the previous pictures of the current picture'"'"'s type, and the target number of bits per unit row;

an adaptive quantization unit which calculates a target number of bits per unit row and a number of encoding bits per unit row in a previous row for each macroblock-row of the current picture type, calculates a quantizer scale value corresponding to the current row by using functions including the target number of bits per unit row and the number of encoding bits per unit row, and calculates a total number of encoding bits of the current picture type;

a memory which stores the quantizer scale value calculated by the adaptive quantization unit, and outputs the quantizer scale values of the previous macroblock-rows and the previous picture types; and

a buffer rate calculating unit which calculates the buffer balance value by using the total number of encoding bits-per-picture and the average target number of bits-per-picture of the current picture type.

12. The encoding rate controller according to claim 11, wherein the target number of bits-per-picture is calculated by using the following equations:

$\begin{array}{c}{\mathrm{Tavg}}_{i,p,b}=\frac{B_{i,p,b}·<br><br>\left(N_i+N_p+N_b\right)}{B_i·<br><br>N_i+B_p·<br><br>N_p+B_b·<br><br>N_b}·<br><br>\frac{\mathrm{bit\_rate}}{\mathrm{frame\_rate}}\\ T={\mathrm{Tavg}}_{i,p,b}-\frac{B_{i,p,b}·<br><br>\left(N_i+N_p+N_b\right)}{B_i·<br><br>N_i+B_p·<br><br>N_p+B_b·<br><br>N_b}·<br><br>\frac{d}{N_{\mathrm{dis}}}\end{array}$

where T is the target number of bits-per-picture;

Tavg_i,p,b is the average target number of bits-per-picture for each picture type;

B_i,p,b is the average number of encoding bits-per-picture for each picture type;

N_i, N_p and N_b are the number of I, P and B pictures in a group of pictures, respectively;

bit_rate is a predetermined transmission rate;

frame_rate is a frame rate;

d is the buffer valance value; and

N_dis is the number of dispersed pictures.

13. The encoding rate controller according to claim 12, wherein the buffer balance value is calculated by using the following equation:

d+=S_i,p,b−

Tavg_i,p,b

where d+ is the incremental change of the buffer balance value d, and S_i,p,b is the total number of encoding bits of one picture for each picture type.

14. The encoding rate controller according to claim 12, wherein the average number of encoding bits-per-picture B_i,p,b for each picture type is filtered by using the following conditional expression:

If (B_p<

1.2·

B_b) B_p=1.2·

B_b;


If (B_i<

1.2·

B_p) B_i=1.2·

B_p;


If (B_i>

3.5·

B_p) B_i=3.5·

B_p;


If (B_b<

0.6·

B_p) B_b=0.6·

B_b.

15. The encoding rate controller according to claim 12, wherein the quantizer scale value corresponding to the first macroblock-rows of the current picture type is calculated by using the following equations:

$\begin{array}{c}{Q}_{\mathrm{init}}=Q_{i,p,b}+\frac{\mathrm{\gamma <br><br>}}{\mathrm{\delta <br><br>}}\\ \mathrm{\gamma <br><br>}=\frac{{\mathrm{Tavg}}_{i,p,b}-T}{{\mathrm{Tavg}}_{i,p,b}}\\ Q_{i,p,b}=\frac{1}{N_{\mathrm{row}}}·<br><br>\underset{j=1}{\overset{N_{\mathrm{row}}}{\sum <br><br>}}<br><br>{\mathrm{mquant}}_j\end{array}$

where Q_i,p,b is the average quantizer scale value;

mquant_j is a quantizer scale value corresponding to a j-th macroblock-row;

N_row is the number of total rows in a picture;

Tavg_i,p,b is the average target number of bits-per-picture for each picture type;

T is the target number of bits-per-picture;

Q_init is a quantizer scale value corresponding to a first macroblock-row; and

δ

is a constant.

16. The encoding rate controller according to claim 15, wherein the quantizer scale value mquant_j+1 corresponding to each macroblock-row of the current picture type is calculated by using the following equation associated with the functions:

If (f(j,S_j,S<sub>j−

1</sub>, T_row)>

β

_up_j), then mquant_j+1=mquant_up(mquant_j,q_scale_type);


else if (g(j,S_j,S<sub>j−

1</sub>,T_row)>

β

_—

dn_j), then mquant_j+1=mquant_—

dn(mquant_j,q_scale_type);


else mquant_j+1=mquant_j.

where S_j is the number of encoding bits from the first macroblock-row to a j-th macroblock-row in a picture;

T_row is a target number of bits per unit row;

mquant_up(mquant_j, q_scale_type) is a quantizer scale value corresponding to a code larger than a previous k-th code corresponding to mquant_j in any one of a plurality of q_scale_types;

mquant_dn(mquant_j, q_scale_type) is a quantizer scale value corresponding to a code smaller than the previous k-th code corresponding to mquant_j in any one of a plurality of q_scale_types, and β

_up_j and β

_dn_j are first and second critical constants, respectively.

17. The encoding rate controller according to claim 16, wherein the functions of f and g include an expression representing the number of encoding bits in the previous row.

18. The encoding rate controller according to claim 16, wherein mquant_up(mquant_j, q_scale_type) is a quantizer scale value corresponding to a (k+1)-th code larger by one than a previous k-th code corresponding to mquant_j in any one of a plurality of q_scale_types, and mquant_dn(mquant_j, q_scale_type) is a quantizer scale value corresponding to a (k−

1)-th code smaller by one than the previous k-th code corresponding to mquant_j in any one of a plurality of q_scale_types.

19. A video data transmission system comprising:

a video encoder that receives a quantizer scale value corresponding to a current macroblock-row of a current picture type, encodes video input data in accordance with the quantizer scale value, and outputs the encoded video data;

an encoding rate controller that calculates the quantizer scale value corresponding to the current macroblock-row by using a target number of bits per unit row and the number of encoding bits per unit row in the previous macroblock-row for the encoded video data; and

an encoder buffer that outputs the encoded video data in accordance with a transmission rate.

20. The video data transmission system according to claim 19, wherein the encoding rate controller comprises:

a bit allocation unit that calculates a target number of bits-per-picture of a current picture type by using an average number of encoding bits-per-picture and an average target number of bits-per-picture of each picture type I, P and B of the previous group of pictures for the encoded video data, and a buffer balance value;

a rate control unit that calculates a quantizer scale value corresponding to the first macroblock-row of the current picture type by using an average quantizer scale value of a prior picture of the same type as the current picture, and the target number of bits;

an adaptive quantization unit that calculates a target number of bits per unit row and a number of encoding bits per unit row in the previous row for each macroblock-row of the current picture type, calculates a quantizer scale value corresponding to the current row by using functions including the target number of bits per unit row and the number of encoding bits per unit row, and calculates a total number of encoding bits of the current picture type;

a memory that stores the quantizer scale value calculated by the adaptive quantization unit, and outputs the quantizer scale values of the previous macroblock-rows and the previous picture types; and

a buffer rate calculating unit that calculates the buffer balance value by using the total number of encoding bits-per-picture and the average target number of bits-per-picture of the current picture type.