JPH1042293A - Encoding controller, encoding device and encoding control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To suppress the deterioration of a picture and to prevent the occurrence of a fault in a reproduction processing in a decoder by judging estimated code quantity for respective blocks in a frame and adjusting the generated code quantity of the blocks in accordance with estimated code quantity. SOLUTION: A code quantity estimation means 1 estimates code quantity for the respective blocks of the frame for this time by executing temporary encoding on the frame for this time and the frame for previous time. A code quantity judgment means 2 judges whether estimated quantity obtained in the code quantity estimation means 1 is less than a prescribed threshold TH or not for the respective blocks. The average value of the estimated code quantity of the respective blocks in the prescribed frame can be used for the threshold. A code quantity adjustment means 3 adjusts the generated code quantity of the block in accordance with the estimated code quantity on the block whose estimated code quantity is less than the prescribed threshold. An encoding means 4 executes an encoding processing for the respective blocks on data in the prescribed frame so that code quantity for the respective blocks becomes code quantity adjusted by the code quantity adjusting means 3.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] The present invention relates to JPEG and MPE.
The present invention relates to an encoding control device, an encoding device, and an encoding control method applicable to G and the like and suitable for a variable transfer rate.

[0002]

2. Description of the Related Art Conventionally, as a compression apparatus for image data used in a recording / reproducing system (digital image encoding / decoding system) using a CD-ROM or the like as a recording medium,
For example, a compression device as disclosed in Japanese Patent Application Laid-Open No. 4-177990 is known.

In this compression apparatus, a hybrid coding method (a typical coding method for data compression) is performed by sequentially combining orthogonal transform such as DCT (discrete cosine transform), quantization and variable length coding. In the case of using, the encoding control is performed so that the variation of the bit amount of the compressed data for each frame due to the variation of the data amount according to the state of the screen is suppressed, and the encoding amount is substantially constant, Decoder
It is intended to reduce (playback) and complexity on the transmission side.

[0004] That is, this compression device is basically
When the output of the orthogonal transform means is quantized under quantization characteristics that can be changed by the quantizing means, and the quantized output is subjected to variable-length encoding by the variable-length encoding means to obtain compressed image data, the code amount control means In this method, the code amount of the compressed image data is detected, and the quantization characteristic of the quantization means is changed and controlled so as to suppress the fluctuation. As a result, in the image frame, the code amount of the compressed image data becomes substantially constant for each coding block in the image frame, and when this compression apparatus is used for encoding by the encoder, the transmission system Processing load can be reduced.

[0005]

However, in the above-described conventional compression apparatus (encoding control method), a constant value is set for each encoded block, irrespective of the spectral component of each encoded block in an image frame and its distribution. Since the code amount is allocated so as to generate the code amount of, the code amount can be stably controlled, but the coding efficiency is likely to be degraded in a block having a high frequency spectrum,
There is a drawback that the S / N of the entire coded block is deteriorated and that the image deterioration is conspicuous.

On the other hand, if each coded block in an image frame is coded with a constant requantized gradation,
Although the unnaturalness of image deterioration can be reduced, it is difficult to control the code amount of the entire image frame to a desired code amount (target code amount). More specifically, if the code amount control is not performed, the transmission information capacity per unit time of the transmission channel will be exceeded, and the conditions such as the reproduction buffer capacity will not be satisfied, and the reproduction processing will be hindered. There is.

As described above, in variable bit rate coding in which coding is performed in accordance with the information amount of an image frame, when coding control is performed so that coding is performed at a constant requantization gradation, decoding is performed, for example. In some cases, there may be a problem in the reproduction processing in the encoder, and in the case where the encoding control is performed so that the encoding is performed with a constant code amount, the image deterioration becomes conspicuous, and contradictory to each other. There was a problem.

According to the present invention, when encoding is performed in accordance with the amount of information of an image frame, image deterioration is suppressed (image deterioration is made inconspicuous), and, for example, there is no trouble in reproduction processing in a decoder. It is an object of the present invention to provide an encoding control device, an encoding device, and an encoding control method capable of performing optimal encoding control.

[0009]

According to one aspect of the present invention, there is provided an encoding control apparatus for performing encoding control on data in a predetermined frame. A code amount determining unit that determines whether the estimated code amount of each block in a predetermined frame is equal to or smaller than a predetermined threshold for each block,
For a block having an estimated code amount equal to or smaller than a predetermined threshold, a code amount adjusting means for adjusting the generated code amount of the block according to the estimated code amount is provided.

According to a second aspect of the present invention, in the encoding control apparatus according to the first aspect, an average value of the estimated code amount of each block in a predetermined frame is used as the predetermined threshold value. I have.

According to a third aspect of the present invention, in the encoding control apparatus of the first aspect, the code amount adjusting means includes a code amount adjusting unit for a block whose estimated code amount is equal to or less than a predetermined threshold value and whose estimated code amount reaches a minimum value. Setting an evaluation parameter value that gives a high priority between the predetermined threshold value and the minimum value of the estimated code amount, and, based on the evaluation parameter value, a group having the largest evaluation parameter value if the generated code adjustment amount is small. If the amount of generated code adjustment is large, the quantization parameter is changed only for blocks that rank low, and the quantization parameter is sequentially changed uniformly to blocks with lower evaluation parameter values to adjust the generated code amount of the block. It is characterized by.

According to a fourth aspect of the present invention, in the encoding control apparatus according to any one of the first to third aspects, the code amount adjusting means is configured to determine that the total code amount of the entire predetermined frame is equal to the total code amount. For a block whose estimated code amount is equal to or smaller than a predetermined threshold value, the generated code amount of the block is adjusted according to the estimated code amount so as to approach the target code amount allocated to the frame. I have.

According to a fifth aspect of the present invention, in the encoding control apparatus according to the fourth aspect, the estimated code amount for each block in the predetermined frame and the target code amount in the predetermined frame are determined based on the current predetermined frame. It is characterized in that it is determined by encoding with a temporary encoding means, or is estimated based on the previous frame.

According to a sixth aspect of the present invention, there is provided an encoding apparatus for encoding data in a predetermined frame for each block, and estimating a code amount for each block for data in a predetermined frame. Code amount estimating means for performing the following, code amount determining means for determining, for each block, whether or not the estimated code amount for each block estimated by the code amount estimating means is equal to or smaller than a predetermined threshold value; Code amount adjusting means for adjusting the generated code amount of the block according to the estimated code amount of the block having a predetermined threshold value or less, and a code amount adjusted by the code amount adjusting means for each block. Encoding means for performing an encoding process on data in a predetermined frame so as to obtain an amount of data.

According to a seventh aspect of the present invention, in the encoding apparatus according to the sixth aspect, the code amount estimating means estimates the code amount by performing provisional encoding on the current frame. Alternatively, it is characterized in that the code amount of the current frame is estimated based on the previous frame.

According to an eighth aspect of the present invention, in the encoding apparatus according to the sixth aspect, the code amount adjusting means is arranged such that the total code amount of the entire predetermined frame approaches the target code amount allocated to the frame. As described above, for a block in which the estimated code amount is equal to or less than the predetermined threshold, the generated code amount of the block is adjusted according to the estimated code amount.

According to a ninth aspect of the present invention, in the encoding apparatus according to the eighth aspect, the code amount adjusting means includes an evaluation parameter value which gives a high priority between a predetermined threshold value and a minimum value of the estimated code amount. Based on the evaluation parameter value, if the generated code adjustment amount is small, the quantization parameter is changed only for the block ranked in the group having the largest evaluation parameter value, and if the generated code adjustment amount is large, the evaluation parameter The present invention is characterized in that the quantization parameter is sequentially changed uniformly to the blocks having the lower values, the requantization gradation is controlled, and a predetermined frame is encoded.

According to a tenth aspect of the present invention, there is provided an encoding control method for performing encoding control on data in a predetermined frame, wherein the estimated code amount for each frame in the predetermined frame is determined for the data in the predetermined frame. Is determined for each block, and the code amount is not adjusted for a block with a large estimated code amount, but the code amount is adjusted for a block with a small estimated code amount to some extent.

The invention according to claim 11 is the first invention.
0, the code amount is adjusted for a block whose estimated code amount is equal to or less than a predetermined threshold so that the total code amount of the entire predetermined frame gradually approaches the target code amount allocated to the frame. , According to the amount of code.

The invention according to claim 12 is the first invention.
In the encoding control method according to 1, the estimated code amount for each block in the predetermined frame and the target code amount in the predetermined frame are determined by coding the current predetermined frame by a temporary coding unit. Alternatively, the estimation is performed based on the previous frame.

According to the first to twelfth aspects of the invention,
For the data in the predetermined frame, the amount of the estimated code amount for each frame in the predetermined frame is determined for each block. Since the code amount is adjusted for a small number of blocks, when coding is performed in accordance with the information amount of the image frame, image deterioration is suppressed (image deterioration is less noticeable) and, for example, reproduction processing by a decoder is performed. Optimal encoding control can be performed so as not to cause trouble. For storage media with a variable transfer rate data channel, etc., while making the most of its advantages and performing high-efficiency image coding and uniform playback image quality coding, The generated code amount can be controlled in accordance with the constraint.

[0022]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows an encoding apparatus according to the present invention.
FIG. 3 is a diagram illustrating a configuration example of an (encoder). Referring to FIG. 1, the encoding apparatus includes a code amount estimating unit 1 for estimating a code amount for each block for data in a predetermined frame, and an estimated code for each block estimated by the code amount estimating unit 1. A code amount determining means 2 for determining, for each block, whether or not the amount is equal to or less than a predetermined threshold value TH, and for a block having an estimated code amount equal to or less than the predetermined threshold value TH, A code amount adjusting means 3 for adjusting the generated code amount of the block; and a code amount adjusting means 3 for adjusting the code amount of each block to the code amount adjusted by the code amount adjusting means 3. And an encoding means 4 for performing the encoding process.

Here, the code amount estimating means 1 estimates the code amount of each block of the current frame by performing provisional encoding on the current frame, or based on the previous frame. The code amount for each block of the current frame is estimated. For example, in order to guarantee reproduction by the decoder, the decoder simulates the state of the data buffer during reproduction, determines the global code amount distribution between each image frame, and determines this as the estimated code amount. be able to.

In the code amount determining means 2, an average value of the estimated code amount of each block in a predetermined frame estimated by the code amount estimating means 1 can be used as the predetermined threshold value TH.

The code amount adjusting means 3 adjusts the estimated code amount to a predetermined threshold TH so that the total code amount of the entire predetermined frame gradually approaches the target code amount allocated to the frame.
For the following blocks, the generated code amount of the block is adjusted according to the estimated code amount.

More specifically, a predetermined threshold value T of the estimated code amount
An evaluation parameter value that gives a high priority between H and the minimum value is set. Based on the evaluation parameter value, if the generated code adjustment amount is small, only the blocks ranked in the group with the highest evaluation parameter value If the quantization parameter is changed, and if the generated code adjustment amount is large, the quantization parameter is sequentially and uniformly changed to a block having a lower evaluation parameter value, and the generated code amount of the block is adjusted.

FIG. 2 is a diagram showing the relationship between the correction coefficient A of the reference quantization parameter _{Qref and} the estimated code amount (local code information amount) d as described later. FIG. 9 is a diagram showing a relationship between an evaluation parameter value u and a parameterization parameter Q _ref and an estimated code amount (local code information amount) d.

As can be seen from the example of FIG. 2, when the estimated code amount (local code information amount) d and the reference quantization parameter _Qref are given for a certain block, the code amount adjusting means 3 The correction coefficient A for the quantization parameter _Qref is set based on the estimated code amount (local code information amount) d. For example, if the reference quantization parameter Q _ref is “3” and the local code information amount d of this block is “0.6” for a certain block, the correction coefficient for the reference quantization parameter Q _ref (= 3) A becomes “+0.25”, so the code amount adjusting means 3 corrects the reference quantization parameter Q _ref (= 3) with the correction coefficient A according to the table of FIG. 2, and obtains Q _ref + A (=
3 + 0.25), which is referred to as a requantization parameter Q ′.
(= Q _ref + A).

In the present invention, the code amount adjusting means 3
Further, using the table of FIG. 3, a correction amount x for the correction coefficient A obtained in FIG. 2 is calculated, the correction coefficient A is corrected by the correction amount x, and finally the requantization parameter Q ′ is
Q _ref + A + x is set.

More specifically, referring to FIG.
For a block in which the (local code information amount) d is equal to or greater than a predetermined threshold value (for example, an average value) TH (TH = 0.5 in the example of FIG. 3), the evaluation parameter value u is set to “0”.
For a block in which the estimated code amount (local code information amount) d is equal to or less than a predetermined threshold value TH (TH = 0.5 in the example of FIG. 3), the amount of the estimated code amount (local code information amount) d depends on the degree of d. An evaluation parameter value u is set. That is, a block whose estimated code amount (local code information amount) d is close to the predetermined threshold value TH (d = 0.5 in the example of FIG. 3) or a block whose estimated code amount (local code information amount) d is close to the minimum value ( In the example of FIG. 3, for example, for a block of d <0.1), a small evaluation parameter value u is set, and the estimated code amount
(Local code information amount) d is a predetermined threshold TH (in the example of FIG. 3,
A large evaluation parameter value u is set for a block (for example, a block with d = 0.3) near the middle between d = 0.5) and the minimum value (d <0.1 in the example of FIG. 3). Have been.

When the estimated code amount (local code information amount) d and the reference quantization parameter _Qref are given for a certain block, the code amount adjusting means 3 determines from the table in FIG.
An evaluation parameter value u corresponding to this is calculated, and a correction amount x for the correction coefficient A is calculated based on the evaluation parameter value u.

That is, the code amount adjusting means 3 sets an evaluation parameter value giving a high priority between the predetermined threshold value TH and the minimum value of the estimated code amount, and based on the evaluation parameter value, the generated code adjustment amount. Is small, the group having the largest evaluation parameter value (for example, d = 0.
The quantization parameter is changed only for the blocks ranked in group No. 3 and, if the amount of generated code adjustment is large, the blocks with lower evaluation parameter values (for example, from d = 0.3 block to d = 0.5, d = 0.1 blocks),
The quantization parameter is sequentially and uniformly changed.

More specifically, when the necessity of code amount control arises, a block group in which the evaluation parameter value u takes the largest value in the table of FIG. 3 (d = d in the example of FIG. 3).
0.3), as a correction amount x for the correction coefficient A of the reference quantization parameter _Qref ,
When controlling in the direction of increasing the frame code amount,
For example, when x = −1 is determined and control is performed in a direction to reduce the frame code amount, x = + 1 is determined, for example. Further, in this case, x = 0 is determined for other blocks (blocks other than the block belonging to d = 0.3 in the example of FIG. 3).

If the code control amount is insufficient even after performing the code amount control, a group of blocks having the second largest evaluation parameter value u (d = 0.4, d =
As for the block belonging to 0.2), the same correction amount (x = + 1 or -1) as above is calculated as the correction amount x for the correction coefficient A of the reference quantization parameter _Qref .

Further, when the code control amount is insufficient,
For each block group, x = ± u as a correction amount x for the correction coefficient A of the reference quantization parameter _Qref.
Is to be determined.

By performing such code amount control, the code amount adjusting means 3 finally obtains the requantization parameter Q '.

That is, the requantization parameter is determined by
The optimal value can be selected by changing the correction amount x stepwise according to the necessity of adjusting the code amount. At this time, the range of x is restricted by the evaluation parameter u (positive integer), and −u <x <u. If the correction (adjustment) amount is insufficient in this range, the reference quantization parameter Q _ref of the entire screen is used.
Is shifted by one step. Therefore, the requantization parameter Q ′ applied to each block is obtained as follows.

[0038]

(Equation 1)

In other words, the code amount adjusting means 3 sets the estimated code amount to the predetermined threshold value T so that the total code amount of the entire predetermined frame gradually approaches the target code amount allocated to the frame.
For blocks below H, for example, using a table TBL as shown in FIG. 3, weighting is performed according to the estimated code amount, and the generated code amount of the block is adjusted.

The coding means 4 performs coding processing for each block on data in a predetermined frame so that the code amount for each block becomes the code amount adjusted by the code amount adjusting means 3. It is supposed to do it.

As described above, according to the present invention, when encoding control is performed on data in a predetermined frame, the amount of estimated code for each frame in the predetermined frame is determined for each block in the data in the predetermined frame. The code amount is not adjusted for a block having a large estimated code amount, but the code amount is adjusted for a block having a small code amount to some extent.

That is, in order to obtain a uniform and natural reproduced image, it is desirable that as many blocks as possible be encoded with the same requantized gradation. However, in this case, when no control is performed on the generated code amount, the transmission information capacity per unit time of the transmission channel is exceeded, and the conditions such as the reproduction buffer capacity are not satisfied, which causes a problem in the reproduction processing and the like. There's a problem. In order to solve such conflicting problems, the present invention controls the code amount.
(Adjustment), the block in which image quality deterioration is conspicuous, that is, the estimated code amount becomes a predetermined threshold value (for example, an average value) TH
= 0.5) or more, the code amount control (adjustment) is not performed, and even if the code amount is controlled (adjusted) to a certain extent, the block in which the image deterioration is not conspicuous, that is, the estimated code amount is a predetermined threshold (for example, average ) Code amount control (adjustment) is performed only for blocks of TH (for example, d = 0.5) or less. As a result, the image quality of the entire frame is not significantly degraded, and the code amount of the entire frame is asymptotic to the target code amount, so that reproduction by, for example, a decoder can be guaranteed.

Even if the estimated code amount is equal to or less than the predetermined threshold value TH (for example, d = 0.5), the estimated code amount is not more than the minimum value.
For blocks close to (for example, d <0.1), if the adjustment amount of the code amount is increased, the deterioration of the image quality becomes more conspicuous. Therefore, the estimated code amount is a minimum value (for example, d <0.1).
For a block close to, the code amount is not adjusted, or even if it is adjusted, the adjustment amount should be small.

Even if the estimated code amount is equal to or less than the predetermined threshold value TH (for example, d = 0.5), for a block whose estimated code amount is close to the predetermined threshold value TH (for example, d = 0.5), In order to ensure continuity of image quality with blocks having a code amount equal to or larger than a predetermined threshold value TH (for example, d = 0.5), the code amount is not adjusted, or even if adjustment is performed, the adjustment amount is small. It is better to

From this, the estimated code amount becomes equal to the predetermined threshold value TH.
Even in the following cases, the adjustment of the code amount is preferably weighted according to the estimated code amount, and the estimated code amount is mainly determined by the predetermined threshold value TH (for example, d = 0.5) and the minimum value (for example, , D <
0.1) (eg, d = 0.3, 0.2,
It is preferable to make a large weighting (with a large adjustment amount) on a block (eg, a block in which image quality deterioration is not particularly conspicuous even if the code amount is adjusted) in the block (eg, 0.4). That is, the adjustment amount is preferably weighted and determined.

FIG. 4 is a diagram showing a more specific configuration example of the encoding apparatus of FIG. Referring to FIG. 4, the encoding apparatus includes an orthogonal transform unit 9 that performs orthogonal transform (for example, discrete cosine transform (DCT)) of image data in a predetermined frame (image data of each block in the predetermined frame) for each block. ,
A storage unit 10 for temporarily storing output data (orthogonal transformation image data for each block) DATA from the orthogonal transformation unit 9;
The output data (orthogonal transformation image data for each block) DATA from the orthogonal transformation unit 9 is provisionally quantized by, for example, standard encoding characteristics (standard quantization parameters, standard variable length encoding). , First encoding unit 1 for provisional encoding
1 and the code amount of each block obtained by the provisional coding by the first coding unit 11 is added to calculate the total code amount of the predetermined frame. It is determined whether the code usage rate and the communication channel buffer usage rate are high or low, and based on the determination result, a frame target code amount OBJ that can be allocated to the frame is generated and output. The frame reference quantization parameter Q _ref is determined based on the quantity OBJ, and further, the first encoding unit 1
A first code amount control unit 12 that generates and outputs a local code information amount d in which the code amount of each block from 1 is normalized by, for example, the total code amount of this frame;
Local code information amount (normalized code amount) for each block from
It is determined whether or not the block requires code amount control (code amount adjustment) based on d, and code amount control (code amount adjustment).
For a block that requires, the evaluation parameter value (rank value) u according to the code amount of the block, using the tables of FIGS. The correction coefficient A of the reference quantization parameter _Qref and the correction amount x for the correction coefficient A
Is generated and output as flag information FLG, and a block determined to perform code amount control (code amount adjustment) based on the flag information FLG from the code amount adjustment suitability determination unit 13 is described below. The reference quantization parameter _Qref from the first code amount control unit 12 is set to the correction coefficient of the flag information FLG so that the total code amount of the frame approaches the target code amount OBJ from the first code amount control unit 12. A, a second code amount control unit 14 that adjusts according to the correction amount x to obtain a requantization parameter Q ′, a requantization parameter Q ′ from the second code amount control unit 14 and a predetermined variable length code length Thus, a second encoding unit 15 that performs quantization (main quantization) and encoding (main encoding) on the orthogonally transformed image data DATA for each block of this frame stored in the storage unit 10, Have

Here, the first encoding unit 11 corresponds to the code amount estimating means 1 of FIG. 1, has a temporary quantization unit 21 and a temporary encoding unit 22, Output data from
(Orthogonal transform image data for each block) For DATA, standard encoding characteristics (standard quantization parameters,
The provisional quantization and provisional encoding using the standard variable-length coding scheme predict and estimate the code amount of each block in the frame.

The second encoding unit 15 corresponds to the encoding means 4 in FIG. 1, and includes a quantization unit 23 and an encoding unit 2
4 and the re-quantization parameter Q ′ from the second code amount controller 14, the weighting matrix W, and a predetermined variable-length code length for each block of the frame stored in the storage unit 10. By subjecting the orthogonally transformed image data D (DATA) to quantization and variable-length encoding, the intended main encoding is executed.

Now, it is assumed that the orthogonal transform data D of the coded block of the frame and the weighting coefficient matrix W corresponding to each spectrum in performing the quantization are as shown in the following equation.

[0050]

(Equation 2)

When the orthogonal transformation data D and the weighting coefficient matrix W are expressed as shown in Expression 2, the requantized result S
Is a value represented by the following equation.

[0052]

(Equation 3)

Here, WD represents an inner product, and the suffix of each vector is an index value corresponding to each spectrum when the two-dimensional image is orthogonally transformed.

The encoding section 24 of the second encoding section 15
Is such that variable-length coding is performed using the same variable-length coding method as the variable-length coding method (standard variable-length coding method) used in the coding unit 22 of the first coding unit 11, for example. Alternatively, the variable-length coding may be performed by a method different from that of the coding unit 22 of the first coding unit 11.

When generating the frame target code amount OBJ that can be allocated to a predetermined frame, the first code amount control unit 12 determines the conversion characteristic between the code amount and the quantization parameter as shown in FIG. 5, for example. The frame reference quantization parameter _Qref is determined based on the conversion table shown.

The code amount adjustment suitability judging section 13 determines the local code information amount d for each block by, for example, a threshold value TH as shown in FIG. 2 (in the example of FIG. 2, the average value of the code amount d of each block). ), It is determined that a block in which the local code information amount d is larger than the threshold value TH is not suitable for code amount control, while a block in which the local code information amount d is smaller than the threshold value TH is suitable for code amount control. For blocks that are determined to be suitable for code amount control,
Further, for example, a correction amount x for the correction coefficient A is calculated based on an evaluation parameter value u set in a table as shown in FIG.

Further, the second code amount control unit 14 uses the code amount adjustment suitability determination unit 13 as the flag information FLG, together with the code amount control suitability determination information and the correction coefficient A and the correction coefficient for the reference quantization parameter value _Qref . When the correction amount x for A is output, the reference quantization parameter Q _ref for the block determined to be suitable for the code amount control
Is the correction coefficient (adjustment amount) calculated for that block.
The correction is performed using A and the correction amount x, and this is provided to the second encoding unit 15 as a requantization parameter Q ′.

The storage unit 10 stores the orthogonal transform unit 9 from when the output data DATA is output from the orthogonal transform unit 9 to when the encoding process by the second encoding unit 15 is performed.
Encoding unit 11 based on output data DATA from
Since there is a time delay (a time delay of one frame) due to the processing performed, the output data DATA from the orthogonal transform unit 9 is held until the encoding process of the second encoding unit 15 starts. It is provided for.

Next, the operation of the coding apparatus having such a configuration will be described. Image frame data to be coded is orthogonally transformed by the orthogonal transformation unit 9 for each block, and becomes orthogonally transformed image data DATA for each block. The converted image data DATA for each block is sent to the first encoding unit 11 in order to estimate the amount of data after encoding and to allocate an appropriate amount of code. Storage unit 10
Is temporarily stored.

The first encoding unit 11 performs, for example, tentative quantization and tentative encoding on the converted image data DATA for each block using standard encoding parameters (encoding characteristics), for example. Predict and estimate the amount (data amount).
More specifically, after performing the tentative quantization, the first encoding unit 11 performs block conversion by code amount conversion that is equivalent to, for example, the variable length encoding used in the main encoding in the second encoding unit 15. The code amount of each is estimated.

The predicted generated code amount for each block from the first coding unit 11 is sent to the first code amount control unit 12, whereby the first code amount control unit 12 outputs the second code amount. At the time of the main encoding in the encoding unit 15, a global code usage amount (frame target code amount) adapted to the average code transfer rate or the maximum code transfer rate as the target code amount that can be allocated to the current frame by the encoder. ) Adjust OBJ and frame reference quantization parameter _Qref . That is, the first code amount control unit 12 determines that the code amount (total code amount) obtained by totaling the code amounts for each block from the first encoding unit 11 for the entire frame is appropriate as the code usage amount in one frame. If the total code amount exceeds the maximum transfer rate of the target communication channel or exceeds the average code rate, the frame target code amount DATA and the frame reference coding parameter _Qref are changed. , Adjust the code amount.
That is, frame code amount allocation is performed.

Specifically, for this frame, the code length data of the variable length code for each block is cumulatively added (integrated).
Then, the amount of frame code generation is calculated, the error with the target code amount is accumulated within a certain period of time, and negative feedback is performed to automatically perform control to approach the target average code transfer rate within a certain period of time. , The first code amount control unit 12 calculates a target code amount (target total code amount of this frame) OBJ for this frame and a frame reference quantization parameter Q
_{Find ref} . In the above processing, the frame reference quantization parameter Q _ref is determined from the target code amount OBJ of the frame by using a conversion table as shown in FIG. 5, for example.

The first code amount control unit 12 normalizes the predicted generated code amount for each block from the first coding unit 11 and calculates the local code information amount (code amount) d for each block. I do. More specifically, in the normalization, the evaluation value corresponding to the code amount of the block is determined by the code amount adjustment appropriateness determination unit 13.
This is performed in order to obtain a relative value without being affected by the total code amount of the entire frame. Specifically, a histogram of the code amount for each block is created, and the frequency is averaged to perform normalization. it can. Thereby, for example, the first
The maximum code amount and the minimum code amount are searched from the code amount for each block of the frame calculated by the encoding unit 11 of FIG. .
The code amount for each block can be normalized so that it becomes zero.

The local code information amount d for each block normalized as described above is sent to the code amount adjustment suitability judging unit 13, and the code amount adjustment suitability judging unit 13 checks the local code information amount d for each block. Is determined as to whether or not each block is suitable for adjusting the code amount, and the local code information amount d is determined for each block determined to be suitable for adjusting the code amount.
Is given as the evaluation parameter value u. Specifically, for example, as shown in FIG. 3, the code amount is determined based on whether the local code information amount d is smaller than the average value of the entire frame (in this case, the normalized average value “0.5”). It is determined whether or not the block is suitable for the adjustment of. If the local code information amount d is smaller than the average value "0.5" and the block is suitable for the adjustment of the code amount, the block An evaluation parameter value u corresponding to the local code information amount d is obtained, and a correction amount x for the correction coefficient A of the reference quantization parameter Q corresponding to the evaluation value is determined.

As described above, the code amount adjustment propriety judging section 1
3 is for each block, a determination result of whether the block is suitable for code amount adjustment, and a correction coefficient (adjustment amount) A and a reference quantization parameter _Qref of the reference quantization parameter _Qref if the block is suitable for code amount adjustment. The correction amount x is generated as flag information FLG and provided to the second code amount control unit 14.

The second code amount controller 14 corrects (adjusts) the correction coefficient A of the quantization parameter _Qref from the first code amount controller 12 based on the flag information FLG, and Find Q '. That is, for a block determined to be unsuitable for code amount adjustment, the requantization parameter Q ′ is obtained as Q ′ = Q + A,
For a block determined to be suitable for code amount adjustment, the requantization parameter Q ′ is obtained from, for example, Equation 1.
That is, the adjustment of the correction coefficient A is performed only for the blocks requiring the code amount adjustment.

In this embodiment, the correction amount x is weighted with the quantization gradation in comparison with the code generation frequency of the block image. That is, in general, blocks having a low code generation frequency often have a small number of fine components, and quantization distortion such as block boundaries tends to be more conspicuous when the quantization gradation is reduced as compared with a block with a large number of fine components. Is given to the correction amount x. Therefore, the adjustment of the requantization parameter Q ′ is performed with weighting taking these factors into consideration.

When the above-described requantization parameter Q ′ is output from the second code amount control unit 14, the second encoding unit 15 Using the parameter Q ′ and a predetermined variable-length code length, the orthogonal transformation image data DATA for each block of this frame stored in the storage unit 10 is quantized (main quantization) and encoded.
(Real coding). Specifically, when converting to a block code amount based on the information on the code length of the variable length coding, if an error is detected between the frame converted code amount and the estimated code amount based on the feedback information, the conversion is performed again. The increase / decrease of the quantization level is adjusted to converge to the frame target code amount.

As described above, in the encoding apparatus shown in FIG.
Is provided, the first code amount control unit 12 and the second code amount control unit 14 control the code amount according to the distribution of the local code amount in one image frame. It can be done easily. That is, the aim is to encode adjacent coding blocks with uniform coding parameters, and to allocate the code amount to each frame in accordance with the generated code amount estimated by the first coding unit 11, and In addition, it is possible to divide the global (per unit time) target generated code amount after performing control so as to satisfy the reproduction data buffer usage rate of the decoder.

Further, by controlling the coding control mechanism only for code blocks in which the local information amount estimated from the code generation amount for each coding block is relatively small, adjustment to the target generated code amount for each image frame is performed. Can be realized.

When encoding is performed with a target of a variable transfer rate, it is rarely necessary to greatly control the code generation amount. However, in the present invention, the target code amount is set to the generated code amount. Since the estimated approximate value is used, the correction of the generated code amount performed by the second encoding control unit 14 is a small amount. Therefore, from the viewpoint of the balance between the deterioration of the image quality due to the code amount control and the controllable range of the generated code amount, the requantization gradation is preferentially performed in order from the coding block in which the code amount distribution in the image frame is relatively small. Is preferable to facilitate the adjustment of the generated code amount. That is, as in the present invention, an evaluation parameter value u that gives a high priority between the predetermined threshold value TH and the minimum value of the estimated code amount is set,
Based on the evaluation parameter value, if the generated code adjustment amount is small, the quantization parameter is changed only for the block ranked in the group having the largest evaluation parameter value u, and if the generated code adjustment amount is large, the evaluation parameter value u is changed. It is preferable to control the re-quantization gradation by sequentially and uniformly changing the quantization parameter from the lower block to the easier adjustment of the generated code amount.

As described above, according to the present invention, in a storage medium or the like having a data channel with a variable transfer rate, it is possible to make the most of its advantages and achieve highly efficient image encoding and uniform reproduction image quality encoding. , The amount of generated code can be controlled in accordance with the restrictions of the media and the restrictions of the decoder.

The encoding control of the present invention is based on the international standard I
It can be applied to SO10918-1 (JPEG: still image data compression coding technology), and can be applied to the international standard ISO.
It is also applicable to 11172-2 (MPEG: moving picture coding technology). That is, the present invention can be applied to encoding of a still image and to encoding of a moving image.

In particular, if encoding and transmission can be performed in accordance with the information amount of each image frame of a moving image, more efficient transmission of information is possible. In other words, if the media serving as the transmission channel can expand and contract the transmission information capacity per unit time according to the user's request, the transmission channel is assumed to have a relatively large amount of information of the moving image to be transmitted. There is no need to secure or encode, and encoding can be performed while maintaining uniform image quality.
However, if no encoding control is performed, reproduction by the decoder cannot be guaranteed as described above. In order to maximize the advantages of media in which the transmission channel media can expand and contract the transmission information capacity per unit time in accordance with the user's request, that is, media that features a variable data transfer rate, the present invention uses An optimized coding method and code amount control method can be provided.

When a moving image is encoded, the orthogonal transform unit 9 performs orthogonal transform of the image data before encoding.
(As a premise), the difference (offset) between the current frame and the previous frame is detected as a motion vector, and the compression efficiency can be further improved by using motion compensation.

FIG. 6 is a diagram showing an example of the configuration of a coding apparatus that also takes such motion compensation into account. Referring to FIG. 6, in the encoding apparatus, in the encoding apparatus in FIG.
Further, a motion vector detecting unit 107 for detecting a motion vector G from image frame data input sequentially in time.
A type determining unit 106 for determining a motion compensation type / coding type E; a motion compensation frame error calculation unit 102 for calculating a motion compensation frame error; and a storage unit 108 for temporarily storing the detected motion vector G. Have been.

Here, the motion vector is, as described above,
The parameters representing the “magnitude” and “direction” of the motion used when performing the block motion compensation. The motion compensation type is a parameter representing the inter-frame motion compensation type. There are motion compensation, backward inter-frame motion compensation, bidirectional inter-frame motion compensation, and no motion compensation. The coding type is a parameter indicating a block coding type selected in association with the motion compensation type, and includes inter-frame differential coding, intra-frame coding, and the like.

In the coding apparatus having such a configuration, when the image frame data successive in time arrives sequentially, the motion vector detecting section 107 sets all the motion compensation types which may be employed for coding on a block-by-block basis. Is the block displacement amount that minimizes the difference from the reference frame.
Is calculated as Motion compensation frame error calculator 102
Performs motion compensation for each block using the motion compensation type / coding type E determined by the type determination unit 106 according to the motion vector G detected by the motion vector detection unit 107, and calculates the inter-frame difference for each block. . Note that, when the intra-frame encoding is selected as the encoding type, the processing of the motion compensation frame error calculation unit 102 is not performed (bypassed). That is, if the INTRA encoding that does not use the inter-frame difference is selected as the encoding type of the block, the processing in this stage is bypassed, and the processing is switched to the processing of simply reading the image data of the block from the storage unit.

That is, most of the techniques for compressing a moving image encode the difference between frames and the like, so that the decoding time of the image data is not the same as the reproduction time of the image data. Frame or several frames later. For example, ISO117172-2
If an encoder / decoder conforming to (MPEG) is used, intra-frame encoding or motion compensation type inter-frame encoding can be selected for each frame, and a different code amount is assigned to each frame. For this reason, a general usage of a decoder for this code is to decode data while buffering data for a maximum of 0.5 to 1 second.

The code amount information, the motion vector G, the motion compensation type / coding type E, and the orthogonally transformed image data for each block are stored in the storage units 10 and 108 for realizing a delay of one frame time or more. It is used at the time of conversion. In addition, regarding the encoding type, the first encoding unit 1
1, first code amount control unit 12, second code amount control unit 14
Also used in

Based on the coding parameters determined by these mechanisms, the image data coded by the second coding unit 15 includes a motion vector G, motion compensation / coding type information E, and a requantization parameter Q. And multiplexed by the multiplexer 129, converted into a bit stream SR, and output.

In the above-described processing, a conversion table as shown in FIG. 5 can be prepared for each encoding type, and a quantization parameter can be determined according to the encoding type. In this case, the values in this table are successively modified by the previous training with the same coding type.

[0083]

As described above, according to the first to twelfth aspects of the present invention, for the data in a predetermined frame, the estimated code amount for each frame in the predetermined frame is determined for each block. In the block having a large estimated code amount, the code amount is not adjusted, and the code amount is adjusted for a block having a small estimated code amount to some extent. Therefore, encoding is performed in accordance with the information amount of the image frame. In this case, it is possible to perform optimal encoding control so as to suppress image degradation (to make the image degradation less noticeable) and not to hinder, for example, reproduction processing in a decoder.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a configuration example of an encoding device according to the present invention.

FIG. 2 is a diagram showing a relationship between an estimated code amount (local code information amount) d and a correction coefficient A of a reference quantization parameter _Qref .

FIG. 3 is a diagram illustrating a relationship between an evaluation parameter value u and a reference quantization parameter Q _ref and an estimated code amount (local code information amount) d.

FIG. 4 is a diagram illustrating a more specific configuration example of the encoding device in FIG. 1;

FIG. 5 is a diagram illustrating an example of a conversion characteristic between a code amount and a quantization parameter.

FIG. 6 is a diagram illustrating a more specific configuration example of the encoding device in FIG. 1;

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 code amount estimating means 2 code amount discriminating means 3 code amount adjusting means 4 encoding means 9 orthogonal transform unit 10 storage unit 11 first encoding unit 12 first code amount control unit 13 code amount adjustment suitability determining unit 14 th 2 code amount control unit 15 second encoding unit 102 motion compensation frame difference calculation unit 106 encoding type determination unit 107 motion vector detection unit 108 storage unit

Claims (12)

[Claims] 1. An encoding control device for performing encoding control on data in a predetermined frame, wherein an estimated code amount for each block in the predetermined frame is not more than a predetermined threshold value for the data in the predetermined frame. Code amount determining means for determining whether or not each of the blocks has an estimated code amount equal to or less than a predetermined threshold, and a code amount adjusting unit for adjusting the generated code amount of the block according to the estimated code amount. Encoding control device comprising: 2. The coding control device according to claim 1, wherein the predetermined threshold value is an average value of estimated code amounts of blocks in the predetermined frame. 3. The encoding control device according to claim 1, wherein the code amount adjusting unit is configured to determine the estimated code amount of a block whose estimated code amount is equal to or less than the predetermined threshold and whose estimated code amount is a minimum value. An evaluation parameter value that gives a high priority between the predetermined threshold value and the minimum value is set, and based on the evaluation parameter value, a block ranked in the group with the largest evaluation parameter value if the generated code adjustment amount is small. Only the quantization parameter is changed, and if the amount of generated code adjustment is large, the quantization parameter is sequentially and uniformly changed up to the block having the lower evaluation parameter value to adjust the generated code amount of the block. Control device. 4. The encoding control device according to claim 1, wherein the code amount adjusting unit is configured to determine whether a total code amount of the entire predetermined frame is equal to a target amount allocated to the frame. An encoding control device for adjusting a generated code amount of a block whose estimated code amount is equal to or less than a predetermined threshold value according to the estimated code amount so as to approach the code amount. 5. The coding control device according to claim 4, wherein the estimated code amount for each block in the predetermined frame and the target code amount in the predetermined frame are obtained by coding the current predetermined frame by temporary coding means. An encoding control device which is determined by encoding or is estimated based on a previous frame. 6. An encoding device for encoding data in a predetermined frame for each block, a code amount estimating means for estimating a code amount for each block for data in a predetermined frame, A code amount determining unit that determines whether the estimated code amount for each block estimated by the code amount estimating unit is equal to or less than a predetermined threshold for each block, and for a block whose estimated code amount is equal to or less than a predetermined threshold, A code amount adjusting means for adjusting the generated code amount of the block according to the estimated code amount, and a code amount adjusted for each block by the code amount adjusted by the code amount adjusting means. An encoding unit for performing an encoding process on data in a predetermined frame. 7. The coding apparatus according to claim 6, wherein the code amount estimating means estimates the code amount by performing temporary coding on a current frame, or based on a previous frame. A coding apparatus for estimating a code amount of a current frame. 8. The coding apparatus according to claim 6, wherein said code amount adjusting means is configured to calculate the estimated code amount such that a total code amount of the entire predetermined frame gradually approaches a target code amount allocated to the frame. An encoding apparatus characterized in that, for a block whose is equal to or smaller than a predetermined threshold value, the generated code amount of the block is adjusted according to the estimated code amount. 9. The coding apparatus according to claim 8, wherein said code amount adjusting means sets an evaluation parameter value which gives a high priority in the middle between said predetermined threshold value and a minimum value of said estimated code amount, and evaluates said evaluation parameter value. Based on the parameter values, if the generated code adjustment amount is small, the quantization parameter is changed only for the block ranked in the group with the largest evaluation parameter value, and if the generated code adjustment amount is large, the blocks are sequentially reduced to blocks with lower evaluation parameter values. An encoding apparatus for controlling a re-quantization gradation by uniformly changing a quantization parameter to encode a predetermined frame. 10. An encoding control method for performing encoding control on data in a predetermined frame, wherein for the data in the predetermined frame, a degree of an estimated code amount for each frame in the predetermined frame is determined for each block. A coding control method characterized in that the code amount is not adjusted for a block having a large estimated code amount, and the code amount is adjusted for a block having a small estimated code amount to some extent. 11. The encoding control method according to claim 10, wherein the adjustment of the code amount is performed such that a total code amount of the entire predetermined frame gradually approaches a target code amount allocated to the frame. An encoding control method for a block whose amount is equal to or smaller than a predetermined threshold value, depending on the amount of code amount. 12. The coding control method according to claim 11, wherein the estimated code amount for each block in the predetermined frame and the target code amount in the predetermined frame are obtained by coding the current predetermined frame by temporary coding means. A coding control method characterized in that the coding control method is determined by coding or is estimated based on a previous frame.