KR20060120787A - Method of ultra-fine bit count control of video frame - Google Patents

Abstract

A method for precisely controlling a bit count of a moving picture frame is provided to adaptively select the positions and the number of macro-blocks and control a quantization parameter of the macro-blocks to rapidly and correctly make a generated bit count of a compressed frame correspond to a target bit count. A generated bit count of a compressed frame is compared to a target bit count. When the generated bit count is greater than the target bit count(S100), the quantization parameter of predetermined macro-blocks is increased to reduce the generated bit count such that the generated bit count approximates the target bit count(S200). When the generated bit count and the target bit count have not been changed even though the generated bit count is reduced, the number of macro-blocks is increased and the quantization parameter of the macro-blocks is increased to decrease the generated bit count such that the generated bit count approximates the target bit count. When the generated bit count is smaller than the target bit count, the quantization parameter of predetermined macro-blocks is reduced to increase the generated bit count such that the generated bit count approximates the target bit count(S300). When the generated bit count and the target bit count have not been changed even though the generated bit count is increased, the number of macro-blocks is increased and their quantization parameter is reduced to increase the generated bit count such that the generated bit count further approximates the target bit count.

Description

High precision bit rate control method of video frame {METHOD OF ULTRA-FINE BIT COUNT CONTROL OF VIDEO FRAME}

1 is a block diagram of a video encoding system including a general bit amount control apparatus.

2 is a conceptual diagram of a method for controlling a high precision bit rate of a video frame according to the present invention;

3 is a detailed flowchart of a bit amount reduction process of FIG.

4 is an explanatory diagram of a sequence of macro blocks to be adjusted in the present invention.

5 is a detailed flowchart of the bit amount control termination determination process of FIG.

6 is a detailed flowchart of the bit amount increasing process of FIG.

7 is a detailed flowchart of the bit amount stabilization process of FIG.

<Description of the symbols for the main parts of the drawings>

110: subtractor 120: DCT unit

130: quantizer 140: encoder

150: inverse quantization unit 160: inverse DCT unit

170: adder 180: frame memory

190: motion estimation unit 200: bit amount control unit

The present invention relates to a method for controlling a bit amount of a video frame, and more particularly, to a method for controlling an ultra-precise bit amount of a video frame so that the generated bit amount of each compressed frame can be exactly matched with a target bit amount during video compression.

Digital video requires image compression technology because of the large amount of expression information.Moving Picture Experts Group (MPEG), the international standard for moving picture compression for recent storage media systems, and integrated communication and storage media, or H, the video coding standard for low-speed communications, The overflow or underflow of the buffer used to transmit the variable data generation rate of coded images at the fixed bit rate in the Differential Pulse Coded Modulation (DPCM) / Discrete Cosine Transform (DCT) complex coder The bit rate control to obtain stable image quality without underflow is a very important factor in the image encoder.

FIG. 1 illustrates a bit amount control apparatus of a conventional video encoding system, including a subtractor 110, a discrete cosine transform (DCT) unit 120, a quantization (Q) unit 130, an encoding unit 140, Inverse Quantization (IQ) unit 150, Inverse Discrete Cosine Transform (DCT) unit 160, adder 170, frame memory 180, Motion Estimation (ME) unit 190, and It is configured to include a bit amount control unit 200.

The subtractor 110 outputs the difference between the input image and the motion-compensated image by the motion estimator 190, and the DCT unit 120 performs a DCT operation on an image bit stream input in units of 8 × 8 pixel blocks. The quantization unit 130 performs high efficiency loss compression by performing quantization on the DCT coefficients obtained by the DCT unit 120 using quantization parameters and expressing them as some representative values. The encoder 140 performs variable length coding on the input bit stream as a variable length coding (VLC) unit and transmits the encoded bit stream to the buffer.

The inverse quantization unit 150 inverse quantizes the image data quantized by the quantization unit 130, and the inverse DCT unit 160 performs inverse DCT transformation on the inverse quantized image data in the inverse quantization unit 150. The output of the inverse DCT unit 150 is added to the motion-compensated signal through the motion estimator 190 in the adder 170, which is stored in the frame memory 180 in units of frames.

The motion estimator 190 predicts the motion of the next image to be encoded, which is input from the frame memory 180. The motion estimator 190 compares the current image with the image to predict motion and obtains a motion vector (MV). The motion-compensated image is output to the subtractor 110 and the adder 170 according to the vector value.

The bit amount control unit 200 controls the bit amount by adjusting the quantization parameter of the quantization unit 130 according to the output bit amount of the encoder 140.

As described above, the bit amount is controlled by adjusting the quantization parameter by predicting the relationship between the quantization parameter and the bit amount in various ways in order to match the bit amount of each frame to the target bit amount.

However, because the characteristics of the input image are diverse and difficult to predict, prediction errors always occur and the amount of generated bits of each frame cannot be exactly 100% matched to the target amount of bits.

In addition, Korean Patent Application No. 1998-15859 discloses a video signal encoding method and apparatus according to the amount of target codes.

The code amount of the video data encoded by the initial quantization step size is compared with the target code amount to determine and encode the quantization step size so as to be close to the target code amount.

For example, when the target bit amount of each frame is 1000 bits, if the first bit is encoded and the generation bit amount is 900 bits, the target bit amount of the next frame is adjusted upward to 1100 bits. When the second frame is encoded and the generation bit amount is 1170 bits, the target bit amount of the third frame is adjusted downward by 930 bits. Therefore, on average, the generation bit amount of each frame is maintained at 1000 bits. But not exactly 1000 bits.

In other words, the coded code amount is compared with the target code amount so as to be close to the target code amount, and it is difficult to exactly match the coded code amount to the target code amount.

In addition, "An accurate bit-rate control for real-time MPEG video encoder (Tae Yong Kim, Signal Processing Image Communication, March 2003)" is a bit rate by mathematically modeling the relationship between the codeword length and the bit amount To control. However, due to modeling errors, the amount of generated bits cannot be controlled accurately.

The present invention has been made in view of the above, and an object of the present invention is to adaptively select the position and number of macro blocks and adjust the quantization parameter thereof to quickly and accurately match the generated bit amount of each compressed frame with the target bit amount. To provide an ultra-precise bit rate control method of a video frame.

Another object of the present invention is to provide a method for controlling an ultra-precise bit amount of a video frame regardless of various characteristics of an input image.

According to an aspect of the present invention, there is provided a method for controlling an ultra-precise bit amount of a video frame, the method for controlling the bit amount of a compressed video frame, the method comprising: comparing the generated bit amount of each compressed frame with a target bit amount; 1 course; As a result of the comparison, if the generation bit amount is larger than the target bit amount, the generation bit amount is decreased by increasing the quantization parameter of the predetermined macroblocks to bring the generation bit amount closer to the target bit amount, and the generation bit amount and the target bit even if the generation bit amount is decreased. A second step of increasing the number of macroblocks to be adjusted and increasing its quantization parameter to reduce the amount of generated bits if the magnitude relationship of the amount does not change to bring the generated bit amount closer to the target bit amount; And if the generation bit amount is smaller than the target bit amount as a result of the comparison, the generation bit amount is increased by decreasing the quantization parameter of the predetermined macroblocks to bring the generation bit amount closer to the target bit amount, and the generation bit amount is increased even if the generation bit amount is increased. And a third step of increasing the number of generated macroblocks by decreasing the number of macroblocks to be adjusted and decreasing the quantization parameter thereof to increase the amount of generated bits, thereby bringing the amount of generated bits closer to the target amount of bits if the magnitude of the target bit quantity does not change. It is done.

A first step of initializing a number value of a macroblock to be adjusted at the same time with a quantization parameter when a comparison result generation bit amount of the first process is larger than a target bit amount; Increasing a quantization parameter of the number of macroblocks; A third step of quantizing and variable length coding the macroblocks having the quantization parameter increased; A fourth step of comparing a generation bit amount and a target bit amount of the encoded frame; And if the comparison bit amount and the target bit amount are the same, the operation is terminated. If the generation bit amount is larger than the target bit amount, the second to fourth steps are performed until the generation bit amount is equal to the target bit amount or the generation bit amount is smaller than the target bit amount. And a fifth step of repeating the steps.

In addition, the present invention re-adjusts the number of macroblocks to be adjusted simultaneously with the quantization parameter when the amount of generated bits is smaller than the target bit amount or when the amount of generated bits is smaller than the target bit amount in the process of reducing the amount of generated bits. The first step to do; Reducing a quantization parameter of the number of macroblocks not included in the non-reducible set; A third step of quantizing and variable length coding the macroblocks of the reduced quantization parameter; A fourth step of comparing a generation bit amount and a target bit amount of the encoded frame; And if the generation bit amount is equal to the target bit amount as a result of the comparison, and if the generation bit amount is less than the target bit amount, the second to fourth steps until the generation bit amount is equal to the target bit amount or the generation bit amount are larger than the target bit amount. A fifth step of repeating the steps may be performed.

In addition, the present invention compares the minimum value with the number of macroblocks to be adjusted simultaneously with the quantization parameter when the amount of generated bits is smaller than the target bit amount to increase the amount of bits and then the amount of generated bits is larger than the target bit amount. First step; As a result of the comparison, if the number of macroblocks to be adjusted simultaneously with the quantization parameter is less than or equal to the minimum threshold, the macroblocks of the previously reduced quantization parameters are added to the non-reduced set, and the previously reduced quantization parameters are transferred. A second step of returning to a state and then returning to the second step to increase the bit amount again; A third step of returning the previously reduced quantization parameters to a previous state if the number of macroblocks to be adjusted simultaneously with the quantization parameter as a result of the comparison is greater than the minimum threshold; A fourth step of reducing the number of macroblocks to be adjusted simultaneously with the quantization parameter; A fifth step of reducing the quantization parameters of the predetermined number of macroblocks not included in the non-reduced set and quantizing and variable length coding the macroblocks of the reduced quantization parameter; A sixth step of comparing a generation bit amount and a target bit amount of the encoded frame; And when the generation bit amount and the target bit amount are the same as the result of the comparison, return to the first step of comparing the minimum value with the number value of the macroblock to be adjusted simultaneously with the quantization parameter if the generation bit amount is larger than the target bit amount. If the generation bit amount is smaller than the target bit amount, a seventh step of repeating the fourth step or less is performed to reduce the number of macro blocks to be adjusted simultaneously with the quantization parameter until the generation bit amount is equal to the target bit amount. It is characterized by.

Hereinafter, the present invention will be described in more detail with reference to the accompanying drawings. However, the following examples are merely to illustrate the present invention is not limited to the contents of the present invention.

The present invention targets the bit amount of a compressed frame by adaptively selecting the position and number of macro blocks in the bit precision controller of the conventional video encoding system as shown in FIG. Implement to match the bit amount exactly.

The ultra-precision bit rate control algorithm of the bit rate control unit 200 performs quantization and variable length encoding again by increasing existing quantization parameters of some macroblocks when the bit rate of the compressed frame is larger than the target bit rate, and then, the encoder ( 140) Substituting the re-encoded macroblocks among the data stored in the buffer of the next stage, if the bit amount of the compressed frame is smaller than the target bit amount, the existing quantization parameters are reduced to perform quantization and variable length encoding again, and then The encoder 140 replaces the re-encoded macroblocks among the data stored in the buffer at the end of the encoder 140. Even if the quantization parameter is adjusted in this process, if the magnitude relationship between the bit amount of the compressed frame and the target bit amount is the same as before, Increasing the number of quantization parameters, on the contrary, the magnitude of the frame and the target bit Re-encoding by reducing the number of quantized parameters to control when compared to before reversing and is repeated until matching the bit rate of the compressed frame, the process and the target bit rate.

The ultra-precision bit amount control algorithm of the bit amount control unit 200 according to the present invention will be described in more detail with reference to FIGS. 2 to 7.

First, the ultra-precision bit amount control process by the bit amount control unit 200 is performed as a post-processing process in which primary video compression is performed. As shown in the flowchart of FIG. 2, an initial generation bit amount and a target bit amount are compared. (S100) If the generated bit amount is greater than the target bit amount, go to the bit amount reduction process (S200) through the DEC_START link to reduce the bit amount, and if the generated bit amount is smaller than the target bit amount, increase the bit amount through the INC_START link ( In step S300, the bit amount is increased.

Bit reduction process (S200), as shown in the flow chart of Figure 3, after initializing the number of macro blocks (hereinafter referred to as N) to be adjusted simultaneously with the quantization parameter (S210), the generation bit amount of the compressed frame The opposite of the macroblock compression order to reduce

In step S220, the quantization parameter of the predetermined number of macroblocks is increased. Here, the order of the macro blocks to be adjusted is the reverse of the macro block compression order, as shown in FIG.

In general, the value of N is repeatedly increased and decreased during the process of matching the target bit quantity with the generated bit quantity, and this value should be determined at the beginning of the process. Normally, the process starts with the N value as the minimum threshold, but it is not necessary to set the N value to 1 depending on the purpose of implementation, and may be set to any integer value that is greater than or equal to the total number of macroblocks in one frame above the minimum threshold.

The amount of increase in the quantization parameter is preferably equally small for each macroblock, but may vary depending on the purpose of implementation.

After quantization is performed on the macroblocks of the increased quantization parameters, variable length coding is performed on the macroblocks and other macroblocks having encoding dependencies (S230 and S240).

Briefly, the encoding dependence of the present invention is that a macroblock currently encoded in order to increase the compression rate in encoding encodes only a difference value from a previously encoded macroblock, for example, the quantization parameter of the current macroblock is 10 and the quantization parameter of the previous block. Is 12, the quantization parameter of the current macroblock is encoded as -2. Therefore, if the quantization parameter of the previous block is changed from 12 to 15, even if the quantization parameter of the current macroblock is still 10, it should be changed from -2 to -5 when encoded by the encoding dependency. In addition to the quantization parameter, one macro block has many other values, which also have encoding dependencies.

After the step S240, the bit amount control termination determination process (S250) to be described later is performed, and if the generated bit amount and the target bit amount are the same in the bit amount control termination determination process (S250), the entire process is terminated. If the amount of generated bits is smaller than the target bit amount, the process proceeds to bit increase process S300 through the DEC_TO_INC link. If the amount of generated bits is larger than the target bit amount, the number of bits is greater than the number of macroblocks having the re-coded quantization parameter. The quantization parameters of the macroblock having the quantization parameters in the following order are simultaneously increased and recoded (S260, S270, and S220).

This process is repeated and when all macro blocks are re-encoded, they return to the first macro block. If the generated bit amount is equal to the target bit amount after the process is repeated, the bit amount control is terminated. If the generated bit amount is smaller than the target bit amount, the process proceeds to the bit amount increasing process S300 through the DEC_TO_INC link.

5 is a detailed flowchart of the bit amount control termination determination process (S250). When the generated bit amount and the target bit amount are the same or all the macroblocks are included in the non-reduced set, the entire process is terminated (S251). Exit the amount control end determination process (S250) and proceeds to the next step (S260).

6 is a detailed flowchart of the bit amount increasing process S300.

The bit increase process S300 first initializes the N value when the process starts from the INC_START link (S310), and decreases the N value set in the bit decrease process S200 when the process starts from the DEC_TO_INC link ( S320), if the process starts from the STB_T0_INC link, N value is not changed. However, of course, the N value can be changed to any desired value according to the implementation purpose.

Then, in order to increase the amount of generated bits, the quantization parameter of the predetermined number of macroblocks not included in the non-reduced set in the opposite direction of the macroblock compression order is decreased (S330), and then the macroblocks of the quantization parameters are reduced. A quantization process is performed at step S340. Here, the non-reducible set means a set of macro blocks to which the quantization parameter should not be further reduced.

These quantization parameter non-decreasable macroblocks are determined in the bit amount stabilization process (S400) to be described later and added as elements to the non-decreasable set.

After the quantization, after performing variable length coding on the macroblocks of the reduced quantization parameters and other macroblocks having coding dependence thereof (S350), the bit rate control termination decision process (S250) of FIG. When the end is determined in the same bit amount control end determination process (S360), the entire process ends. If not, the generation bit amount is compared with the target bit amount (S370). If the generation bit amount is large, the process proceeds to the bit rate stabilization process (S400) through the INC_TO_STB link. If the generation bit amount is small, the next larger number of quantization parameters Reducing the quantization parameters of the macro block with (S380, S330), and repeats this process until the generated bit amount and the target bit amount is the same or exit the bit amount increase process (S300) through the INC_TO_STB link.

7 is a detailed flowchart of the bit amount stabilization process (S400).

The bit rate stabilization process S400 firstly adds macroblocks of the quantization parameters that have been previously reduced to the non-decrementable set if N is less than or equal to a predetermined minimum threshold (S420a), and immediately reduces the quantization parameter that has been previously reduced. After returning them to the previous state, the process moves to the bit amount increasing process S300 through the STB_TO_INC link (S420b).

If N is greater than the minimum threshold, the quantization parameters and the macroblock order that have been reduced just before are not returned to the previous state without changing the non-decrementable set (S430). Here, the minimum threshold value (the minimum value of N or the number of minimum macro blocks to be adjusted at the same time) may vary depending on the implementation purpose.

After reducing N, the quantization parameters of the predetermined number of macroblocks not included in the non-decrement set are reduced and quantization is performed on the macroblocks of the reduced quantization parameter (S440-S460). Variable length coding is performed on macroblocks of the quantization parameters and macroblocks having encoding dependencies thereof (S470).

Here, the decrease of the N value is different from the N value increase (S270) in the bit amount reduction process (S200) and the N value increase (S380) in the bit amount increase process (S300). Do.

The increase of the N value in steps S270 and S380 is for the purpose of bringing the generated bit amount to the target bit amount as quickly as possible (the higher the N value, the larger the change amount of the generated bit amount is). This is to adjust the generation bit amount more and more finely to exactly match the generation bit amount already approached to the target bit amount to the target bit amount.

Therefore, the N value is gradually reduced (the smaller the value of N, the smaller the amount of change in the generation bit amount), and accordingly, the process of FIG.

On the other hand, after the encoding through the step (S470), if the end is determined in the bit amount control end determination process (S480) performing the same process as the bit amount control end determination process (S250) of FIG. If not, the target bit amount is compared with the generation bit amount (S490). If the generation bit amount is smaller than the target bit amount, the process returns to step S440.

If the generated bit amount is larger than the target bit amount, the process returns to the beginning of the bit amount stabilization process S400 and repeats until the generated bit amount and the target bit amount become the same or exit the bit amount stabilization process S400 through the STB_TO_INC link. do.

In the present invention as described above, for example, when the target bit amount of each frame is 1000 bits, if the first bit is encoded and the generation bit amount is 900 bits, the present invention (Korean Patent Application No. 1998-15859) Instead of moving to the next frame as described above, the quantization parameter is adjusted again and re-encoding is performed until 1000 bits are obtained. Therefore, the amount of calculation is inevitably increased compared to the conventional methods, but an effort to minimize the increase of the amount of calculation is a process of the present invention.

As described above, although described with reference to a preferred embodiment of the present invention, those skilled in the art various modifications of the present invention without departing from the spirit and scope of the invention described in the claims below Or it can be changed.

As described above, in the method for controlling the ultra-precision bit rate of a video frame according to the present invention, an input image is obtained by adaptively selecting a location and number of macro blocks to exactly match the generated bit amount of each compressed frame with a target bit amount. It can be applied regardless of various characteristics of, and since the relationship between the codeword length and the bit amount is not mathematically modeled and the bit amount is not controlled, the generation bit amount is precisely matched to the target bit amount because no modeling error occurs. Matching can be done.

In addition, the present invention can accurately match the generated bit amount to the target bit amount without an error of 1 byte, so that the bit amount of the compressed frame and the bit of the RAID storage server or compressed frame synchronized with the stripe size It becomes available for applications that need to precisely control the amount of generated bits, such as forward error correction coding where the amount must be constant.

Claims (11)

In the method for controlling the bit rate of a compressed video frame, A first step of comparing a generated bit amount and a target bit amount of each compressed frame; As a result of the comparison, if the generation bit amount is larger than the target bit amount, the generation bit amount is increased by increasing the quantization parameter of the predetermined macroblocks to bring the generation bit amount closer to the target bit amount. A second step of increasing the number of macroblocks to be adjusted and increasing its quantization parameter to reduce the amount of generated bits if the magnitude relationship of the amount does not change to bring the generated bit amount closer to the target bit amount; And As a result of the comparison, if the amount of generated bits is smaller than the target bit amount, the amount of generated bits is increased by decreasing the quantization parameter of the predetermined macroblocks to bring the generated bit amount closer to the target bit amount. A third step of increasing the number of generated macroblocks by decreasing the number of macroblocks to be adjusted and decreasing the quantization parameter thereof to increase the amount of generated bits to bring the generated bit amount closer to the target bit amount if the magnitude of the bit amount does not change; Ultra-precision bit rate control method for a video frame comprising a. The method of claim 1, wherein the comparison result generation bit amount of the first process is larger than a target bit amount. A first step of initializing a number value of a macroblock to be adjusted simultaneously with the quantization parameter; Increasing a quantization parameter of the macro block according to the initialization; A third step of quantizing and variable length coding the macroblocks with the quantization parameter increased; A fourth step of comparing a generation bit amount and a target bit amount of the encoded frame; And As a result of the comparison, if the generation bit amount and the target bit amount are the same, the operation is terminated. Repeating a fifth step; Ultra precision bit amount control method of a video frame, characterized in that to perform. The method of claim 2, wherein the order of the macroblocks to increase the quantization parameter is in a direction opposite to the order of the macroblocks to be compressed during video compression. The method according to claim 2, wherein when the resultant bit amount is smaller than the target bit amount as a result of the comparison in the fourth step, the generated bit amount is increased by decreasing the quantization parameter of the predetermined macroblocks to match the generated bit amount with the target bit amount. Ultra precision bit rate control method for a video frame, characterized in that. 3. The macroblock of claim 2, wherein when the amount of generated bits is greater than a target bit amount in the fifth step, the macroblock has a greater number of next order quantization parameters than the number of macroblocks in which the quantization parameter is increased in the second step. Ultra-fine bit rate control method for a video frame, characterized in that to increase the quantization parameters of the video at the same time. The method of claim 1, wherein the comparison result generation bit amount of the first process is smaller than a target bit amount. A first step of readjusting the number of macroblocks to be adjusted simultaneously with the quantization parameter; Reducing a quantization parameter of the number of macroblocks not included in the non-reducible set; A third step of quantizing and variable length coding the macroblocks of the reduced quantization parameter; A fourth step of comparing a generation bit amount and a target bit amount of the encoded frame; And As a result of the comparison, if the generation bit amount and the target bit amount are the same, the operation is terminated. Repeating a fifth step; Ultra precision bit amount control method of a video frame, characterized in that to perform. 7. The method of claim 6, wherein the order of the macroblocks to reduce the quantization parameter is in the opposite direction to the order of the macroblocks to be compressed during video compression. 7. The macroblock according to claim 6, wherein when the amount of generated bits is less than a target bit amount in the fifth step, the macroblock has a next number of quantization parameters in a higher order than the number of macroblocks in which the quantization parameter is reduced in the second step. A method of controlling high bit rate of a video frame, characterized in that to simultaneously reduce the quantization parameter of the video frame. 7. The method of claim 2 or 6, wherein the amount of increase or decrease of the quantization parameter is variable. The method according to claim 6, wherein when the generated bit amount is smaller than the target bit amount and the bit amount is increased, the generated bit amount is larger than the target bit amount. A first step of comparing the minimum threshold with the number of macroblocks of the quantization parameter that was previously reduced; As a result of the comparison, if the number of the macroblocks of the quantization parameter reduced immediately before is less than the minimum threshold, the macroblocks of the quantization parameters immediately reduced are added to the non-decrement set, and the previously reduced quantization parameters are transferred. A second step of returning to a state and then returning to the second step to increase the bit amount again; A third step of returning the previously reduced quantization parameters to the previous state if the number of macroblocks of the quantization parameters reduced immediately before the comparison result is greater than the minimum threshold; Reducing a number of macroblocks to be adjusted simultaneously with the quantization parameter; A fifth step of reducing quantization parameters of the number of macroblocks not included in the non-reduced set and quantizing and variable length encoding on the macroblocks of the reduced quantization parameter; A sixth step of comparing a generation bit amount and a target bit amount of the encoded frame; And If the comparison bit amount is equal to the target bit amount, the operation is terminated. If the generation bit amount is larger than the target bit amount, the process returns to the first step of comparing the minimum value with the number of macroblocks of the quantization parameter immediately reduced. A seventh step of repeating, if the amount of generated bits is less than the target bit amount, a fourth step of reducing the number of macroblocks to be adjusted simultaneously with the quantization parameter until the amount of generated bits is equal to the target bit amount; Ultra-precision bit rate control method of a video frame comprising a. The ultra-precision of a video frame according to claim 2, 6 or 10, wherein the variable length coding is performed on another macro block having encoding dependency on the macro block of the increased or decreased quantization parameter. Bit quantity control method.

Images