JPH09107293A - Code amount control method and apparatus thereof - Google Patents

Abstract

(57) [Abstract] [PROBLEMS] To suppress variations in the amount of generated codes and always maintain the quality of reproduced images. A code amount calculator 201 calculates a target code amount of a processing unit to be encoded next from a remaining amount of a buffer memory 106 and an actual generated code amount of each processing unit.
The QP predictor 202 calculates an index value QP that is expected to obtain the target code amount of the processing unit calculated by the code amount calculator 201 from the generated code amount characteristics of a plurality of prepared quantization tables. The table creator 203
A quantization table corresponding to the index value QP sent from the QP predictor 202 is created. The created quantization table is sent to the quantizer via the memory 204 and the QP predictor 202. The quantizer has a code amount control unit 10
The next processing unit is encoded by using the quantization table sent from the QP predictor 202 of No. 7.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for controlling the amount of compressed (encoded) data after compressing the data.

[0002]

2. Description of the Related Art In recent years, data coding (compression) technology has already been widely adopted in data communication, storage (storage), etc., and is being standardized internationally.
In particular, image data is regarded as an important object for compression (encoding) because of its large data amount.

As a data coding method, a DCT (Discrete Cosine Transform) method which can realize a high compression rate is widely adopted. In this DCT method, image data is divided into a predetermined block size, DCT is performed for each block, each element in the block converted into a DCT coefficient is quantized using a quantization table, Basically, image data is compressed by performing entropy coding on the quantized DCT coefficient.

In the image compression by the DCT method, due to the difference in characteristics such as the ratio of high frequency components of the spatial frequency of the image,
The amount of data (the amount of generated code) after compression varies. In data compression, there are many cases where not only a high compression rate must be realized, but also the amount of data after compression must be controlled. This is because, in addition to the software reason that the image transmission conforms to a predetermined standard, for example, in order to stably send the data to the transmission path, which is generally provided in the data compression device. There is also a hardware reason such as a storage capacity of a buffer memory for temporarily storing encoded data.

For example, when the encoded data for one image (frame) exceeds the storage capacity of the entire buffer memory (hereinafter, this total storage capacity is referred to as a full buffer),
It becomes necessary to temporarily stop the compression of the image data, and the image data cannot be properly transmitted. Even if the encoded data for one image can be stored in the buffer memory, depending on the standard, for example, the encoded data amount (generated code amount) may be small in view of the transmission bit rate, the image transfer rate, and the like. For example, dummy data (fill bit) must be added to the encoded data to be transmitted. In order to suppress the deterioration of the quality of the image reproduced on the decoding side (reception side), it is not desirable to suppress the generated code amount more than necessary. For this reason, the generated code amount is also controlled in consideration of the restrictions on the hardware.

FIG. 7 is a conceptual diagram of a method of controlling the amount of generated codes which is performed according to the storage capacity of the buffer memory. This FIG.
Indicates that the encoded data is temporarily stored in the buffer memory 701 and then transmitted to the transmission path.

The control of the generated code amount, which is performed by the restriction of the storage capacity of the buffer memory, is performed by, for example, the buffer memory
The storage capacity of 1 is divided into 32 equal parts, and the total storage capacity of the buffer memory 701 and the current occupancy (the used storage capacity)
Quantization table used for quantization of DCT coefficients so that the usable storage capacity (remaining amount) is calculated from, and all encoded data generated when the next image data is compressed can be stored in the buffer memory 701. Was done by changing.

The reason why the generated code amount can be controlled by changing the quantization table is as follows. When the Huffman coding method is adopted as the coding method of the DCT coefficient, the DCT coefficient is coded as an event in which a run of zero coefficients (ineffective coefficient) and a value (level) of the following non-zero coefficient are combined. Therefore, by increasing the run length of the zero coefficient, the generated code amount becomes low.

Quantization of DCT coefficients in the DCT method is performed using a quantization table prepared corresponding to the block size for which DCT has been performed. With this quantization table, each DCT coefficient is linearly quantized with the step size assigned to the position in the block.
As the step size increases, the number of DCT coefficients quantized to the same value increases, and the number of AC coefficients that match the DC coefficient, that is, zero coefficient also increases. Therefore, the amount of generated code can be controlled by changing the quantization table (step size).

FIG. 8 is a diagram showing a hierarchical structure of image data. In compression of moving images, image data is frame, GO
It is divided into layers such as B (Group Of Block), macroblock, and block. Each frame in FIG. 8 represents a GOB, indicating that one frame of image data is composed of five GOBs.

When the image data has a hierarchical structure as shown in FIG. 8, it is possible to use different quantization tables for each GOB. In this case, for each GOB, the DCT is calculated from the remaining amount of the buffer memory at the time of compression.
The quantization table used to quantize the coefficients was determined.

[0012]

As described above, the generated code amount is controlled due to the restriction of the storage capacity of the buffer memory. In this conventional control, when the remaining amount of the buffer memory is large, the target code amount is large, and
If the remaining amount is small, reduce the target code amount,
Image data (frame) according to the remaining amount of buffer memory
The target code amount is determined for each GOB.
As a matter of course, a slight variation in the generated code amount cannot be avoided. In the conventional control, for example, when a certain amount of difference occurs in the generated code amount between adjacent image data, the target code amount is determined according to the remaining amount of the buffer memory. Therefore, it is generated for each image data (frame). There are many cases in which the code amount is repeatedly increased and decreased, and it takes a long time to converge the fluctuation of the generated code amount. There is also a problem that the generated code amount may be further changed depending on the changed state of the generated code amount. As described above, the conventional control has a problem that it is difficult to stabilize the generated code amount. Due to this problem, it is practically impossible to control the generated code amount for an image in which the generated code amount is greatly different, such as a moving image that moves rapidly.

As described above, the quality of the reproduced image largely depends on the generated code amount. When transferring images at a fixed transfer rate, if the generated code amount fluctuates greatly,
The quality of the entire reproduced continuous image is greatly deteriorated, and the time required to improve the image quality is increased. If the transfer rate is not fixed, the image update interval varies according to the generated code amount, and for example, the image update interval becomes longer in a fast moving state, resulting in an image that gives a great discomfort to people. . For these reasons, it has been desired to suppress the fluctuation of the generated code amount.

An object of the present invention is to suppress fluctuations in the generated code amount and always maintain the quality of reproduced images.

[0015]

In the code amount control method according to the first aspect of the present invention, the code amount of data compressed by using a specified quantization step size becomes a preset target code amount. The method is to divide the data into a plurality of predetermined processing units, and for each processing unit, quantize based on the target code amount and the code amount of another processing unit for which encoding has been completed. The quantization step size is determined, and compression processing is performed for each processing unit using the determined quantization step size.

In the code amount control method of the second aspect, one is selected from a plurality of quantization tables prepared in advance, and the code amount of the data compressed by using the selected quantization table is determined. A method for controlling the coded data after the compression process to have a target code amount set in advance based on the storage capacity of a storage unit in which the data is temporarily stored, and the data is divided into a plurality of units in a predetermined processing unit. Based on the target code amount for each processing unit and the code amount of another processing unit for which compression processing has been completed, one is selected from a plurality of prepared quantization tables, and the selected quantization is performed. The table is used to perform compression processing for each processing unit.

The code amount control device of the present invention selects one from a plurality of quantization tables prepared in advance, compresses the data, and temporarily stores the coded data after the compression process. In order to control the code amount of the data to be compressed to be a target code amount set in advance based on the storage capacity of the storage means, it is assumed that the data will be applied to a data compression device including A dividing unit that divides a plurality of predetermined processing units, and each processing unit,
Based on the target code amount and the code amount of another processing unit for which encoding has been completed, one is selected from a plurality of prepared quantization tables, and the processing unit is selected using the selected quantization table. And a control means for performing compression processing for each.

As described above, the data is divided into a plurality of processing units, and in consideration of the actual generated code amount of the processing unit after the compression processing, the quantization step size used for the processing unit that has not been compressed. By determining the (quantization table), the generated code amount of the processing unit that has not been compressed is adjusted, and the target code amount and the actual generated code amount of each processing unit in the processing unit that has completed the compression process. It is possible to absorb the error between and. As a result, it is possible to control the actual amount of generated code in the entire data with high precision, and it is possible to always keep the error between the actual amount of generated code and the target code amount in the entire data within a small range. .

[0019]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a data compression device 100 to which this embodiment is applied. The configuration and operation will be described with reference to FIG.

An image input device (not shown) is connected to the data compression device 100, and image data is sent from the image input device. The image data (input image data) sent from the image input device is input to the subtractor 101 and the switch unit 102. Note that the operation of each unit is slightly different depending on the color component of the input image data sent from the image input device, but here, for the sake of simplicity of description, it is assumed that there is one color component (monochrome image). To

The subtractor 101 receives the input image data and
The difference value (prediction error) from the motion-compensated previous frame image data output from the predictor 111 is calculated, and the calculated difference value (prediction error) is output to the switch unit 102. The switch unit 102 selects one of the input image data and the output of the subtractor 101 according to the control signal from the code amount control unit 107, and outputs it to the converter 103. Code amount control unit 1
07 causes the switch unit 102 to select the input image data only when a scene change occurs and refresh is performed (intra-frame mode).

The converter 103 divides the input image data or the prediction error input through the switch unit 102 into blocks of, for example, 8 × 8 pixels, performs DCT calculation, and D
Convert to CT coefficient. This DCT coefficient is the quantizer 10
4 is output.

The quantizer 104 quantizes the DCT coefficient input from the converter 103 using the quantization table input from the code amount control unit 107. After the DCT coefficient is quantized, the DCT coefficient is divided into a DC component and an AC component, and Huffman coding is performed. This Huffman coding is performed using, for example, a Huffman coding table prepared in advance. The Huffman-encoded DCT coefficient is encoded (compressed) data by the multiplexer 1.
05 and the inverse quantizer 108. As an alternative to Huffman coding, for example, arithmetic coding may be used.

In the multiplexer 105, in addition to the encoded data output from the quantizer 104, an interframe / intraframe indicating whether or not the encoding is performed in the intraframe mode by the code amount control unit 107. Identification flag f1, transmission / non-transmission identification flag f2 indicating the transmission timing of encoded data, and D
The index value QP of the quantization table used for the quantization of the CT coefficient and the motion vector amount mv are output from the predictor 111. The multiplexer 105 arranges the data input from each of these units in a predetermined format and outputs the data to the buffer memory 106. Hereinafter, the data output from the multiplexer 105 to the buffer memory 106 will be referred to as a generated code to distinguish it from other data.

The buffer memory 106 is a multiplexer 105.
The generated code output from is temporarily stored. The generated code once stored in the buffer memory 106 is then sequentially output to the transmission path.

The code amount control unit 107 includes a buffer memory 1
06, and the buffer memory 1 from the multiplexer 105
Based on the generated code output to 06, the quantization table used by the quantizer 104 is determined as described later, and the quantization table created according to this determination is output to the quantizer 104. It also controls the data compression apparatus 100 as a whole.

The inverse quantizer 108 has the same table as the Huffman coding table of the quantizer 104, and restores the coded data output by the quantizer 104 from the Huffman coding table to DCT coefficients. . After that, the values before quantization are calculated using the quantization table used for quantization of the restored DCT coefficients. The value of the DCT coefficient calculated at this time includes an error at the time of quantization.

The inverse transformer 109 performs an inverse DCT operation on the DCT coefficient input from the inverse quantizer 108 to obtain each D
Pixel data (prediction error) before conversion of CT coefficients is calculated. The value of the pixel data (prediction error) calculated at this time is a value expected to be restored on the receiving (decoding) side that receives the encoded data.

The adder 110 converts the pixel data (prediction error) output from the inverse converter 109 into the switch unit 11
The output from 2 is added, and the added value is calculated by the predictor 111.
Output to

The predictor 111 stores the image data of the previous frame, performs motion compensation on this image data in the inter-frame mode, and subtracts the image data after the motion compensation, And to the switch unit 112. The prediction error output from the inverse converter 110 to the adder 110 is a prediction error (difference value) between the image data subjected to motion compensation output to the subtractor 101 by the predictor 111 and the input image data. is there. The adder 110 adds the motion-compensated image data of the previous frame and the prediction error of the image data and the input image data,
Outputs the image data of the current frame.

The predictor 111 sequentially moves the motion compensation within a range of ± 15 pixels for each macroblock, and detects a position having the smallest accumulated error value within the moved range. Done in. The vector from the original position to the position where the accumulated value of the detected errors is the minimum is the motion vector amount mv. This motion vector amount mv is output from the predictor 111 to the multiplexer 105.

FIG. 2 is a block diagram of the code amount control unit 107. As shown in FIG. 2, a code amount calculator 201 that calculates a target generated code amount for each predetermined processing unit from the generated code input from the multiplexer 105 and the remaining amount of the buffer memory 106; A QP predictor 202 that predicts an index value QP of a quantization table that realizes the target generated code amount calculated by the code amount calculator 201;
A table creator 203 for creating a quantization table corresponding to the index value QP predicted by the P predictor 202;
The memory 204 stores a quantization table created by the table creator 203.

The operation of the code control unit 107 having the above configuration will be described with reference to FIGS. FIG. 3 is an explanatory diagram showing how image data is divided in processing units. FIG.
In, each frame represents a processing unit, and each processing unit is encoded from the upper side to the lower side in the order indicated by the numbers on the left side of the image (screen).

As shown in FIG. 3, in the present embodiment, 1
The image data of the frame is divided into areas 1 to 1 having the same area.
It is divided into six sixth processing units. Each processing unit in this embodiment corresponds to one GOB. The setting of the processing unit is not limited to this. A plurality of GOBs may constitute one processing unit, or a processing unit having an arbitrary shape and size may be set in consideration of the background portion of the image.

FIG. 4 is an explanatory diagram showing the relationship between the index value QP of the quantization table and the generated code amount (hereinafter, this will be referred to as the generated code amount characteristic). In the present embodiment, the sample image data is divided into processing units (GOB), the encoding is performed using a quantization table of a certain index value QP, and the index value QP and the generated code amount at that time are used as the basis. Then, the quantization table for each index value QP is determined.

Specifically, for example, the value of the index value QP which is the base is i, and the generated code amount at that time is code.
(Bit), the index value QP is 1 ≦ QP ≦ M (M is the maximum value (the number of prepared quantization tables)), and when the index value QP is i ≦ QP ≦ M, Tbl _QP =
When the index value QP is 1 ≦ QP ≦ i, the relationship of Tbl _QP = Tbl _QP-1 × 2 is satisfied so that the relationship of Tbl _QP-1 / 2 (Tbl is the generated code amount, the subscript is an index value) is satisfied. As described above, the step size of each element of the quantization table is changed, and the coding is performed by actually repeating the coding of the sample image data.

In the quantization table of each index value QP determined in this way, creation information for creating it is stored in the table creator 203. The table creator 203 uses the created information to generate the QP predictor 202.
Creates a quantization table for the index value QP specified by. The other QP predictor 202 calculates the index value QP corresponding to the target code amount for each processing unit sent from the code amount calculator 201 with reference to the generated code amount characteristic shown in FIG.

At this time, for example, when 250 bits are sent from the code amount calculator 201 as the target code amount of the processing unit, the QP predictor 202 determines from the generated code amount characteristics of FIG.
15 is calculated as the index value QP.

FIG. 5 is an explanatory diagram showing an example of a temporal change in the buffer occupation amount of the buffer memory 106. In FIG. 5, t _-1 , t ₀ , and t ₁ are the times when the image data was encoded. d _-1 and d ₀ are time intervals at which the image data is coded, which coincides with the frame intervals. FIG. 5 shows that one frame of encoded data (including side information such as the motion vector amount mv in this case) output from the multiplexer 105 at each time is stored, and the stored encoded data is stored in the transmission bit of the transmission path. It shows how the occupancy of the buffer memory 106 that sequentially outputs at a rate changes. The ideal remaining amount is an ideal remaining amount to prevent underflow.

In the present embodiment, the image transfer is performed by changing the image transfer rate (the number of frames per unit time), that is, the frame interval. The transmission bit rate of the transmission line is fixed, and the frame interval expands according to the generated code amount. In the intra-frame mode in which the DCT calculation is performed on the input image data as it is, the generated code amount is large, and thus the frame interval is wider than other frame intervals. In the present embodiment, the frame interval is determined so that the transfer rate is not lowered as much as possible and the buffer memory 106 is not overflowed. The target code amount for one frame is determined along with the determination of the frame interval.
The target code amount determined (calculated) at this time is a value at which transmission of the generated code for one frame is completely completed at the determined frame interval.

The code amount calculator 201 monitors the generated code amount for each frame and the occupied amount of the buffer memory 106,
Based on those monitoring results, the frame interval of the next frame,
And the target code amount is calculated. In the example of FIG. 5, the target code amount C ₀ is determined together with the frame interval d ₀ between times t ₀ and t ₁ .

In the present embodiment, as shown in FIG.
The frame is divided into six processing units for encoding. After calculating the target code amount C ₀ , the code amount calculator 201 calculates the target code amount C ₀ , and then based on the target code amount C ₀ and the generated code amount of each processing unit for which encoding has been completed, the target code of the next processing unit to be encoded. The quantity is calculated and output to the QP predictor 202.

The QP predictor 202 is a code amount calculator 201.
When the target code amount for each processing unit is input from, the index value QP expected to obtain the target code amount is calculated with reference to the generated code amount characteristic shown in FIG. The calculated index value QP is sent to the table creator 203, and the table creator 203 creates a quantization table corresponding to the value QP. The quantization table created by the table creator 203 is stored in the memory 204 and the QP predictor 202.
Is output to the quantizer 104 via. As a result, the processing unit to be encoded next is the DCT after the DCT operation using the quantization table created by the table creator 203.
The coefficients are quantized.

FIG. 6 is an operation flowchart of the compression process executed by the code amount control unit 107. This FIG.
3 is a flow representation of the operation of the code amount control unit 107 shown in FIG. 2 for controlling the generated code amount in encoding one frame of image data. A series of processing operations of the compression processing will be described in detail with reference to FIG.

First, in step 601, a target code amount for the entire frame to be encoded next is determined. This target code amount is determined by the image transfer rate, as described above.
The code amount calculator 201 determines based on the transmission bit rate of the transmission path and the occupied amount (remaining amount) of the buffer memory 106. For example, assuming that the current time is time t ₀ in FIG. 5, C ₀ is determined as the target code amount.

When the target code amount for the entire frame is determined, next, at step 602, a process for obtaining the index value QP corresponding to the target code amount is performed. In this process, the code amount calculator 201 divides the target code amount determined in step 601 into six equal parts (in this embodiment, one frame is divided into six processing units) and outputs the value to the QP predictor 202. Then, it is realized by causing the QP predictor 202 to calculate the index value QP corresponding to the value. The QP predictor 202 calculates the index value corresponding to the value sent from the code amount calculator 201 with reference to the generated code amount characteristic shown in FIG. At this time, if the value sent from the code amount calculator 201 is A, the index value QP corresponding to the value A will be calculated (see FIG. 4).

Upon completion of step 602, in step 603, the first processing unit is encoded. The index value QP obtained in step 602 is sent from the QP predictor 202 to the table creator 203. In the processing of step 603, the quantization table created by the table creator 203 is stored in the memory 204 and the QP predictor 20.
It is sent to the quantizer 104 via 2 and performed.

In step 604 following step 603, 2 is assigned to the variable i. This variable i is a code amount calculator 20 that indicates the processing unit currently being encoded in another processing unit that is sequentially encoded after the first processing unit.
1 is a variable for grasping and is held in the code amount calculator 201. In the processing after step 604, the encoding is performed for the processing unit indicated by the value of the variable i.

In step 605 following step 604, a process of obtaining the index value QP for the i-th processing unit is performed from the generated code amount of the processing unit coded so far and the target code amount determined in step 601. .

In this processing, for example, when the value of the variable i is 2, that is, the second processing unit is the object of encoding, the code amount calculator 201 encodes the first processing unit from the target code amount. The actual generated code amount is subtracted, and the value after subtraction is 5
(Number of processing units that have not been encoded yet) QP
It is realized by sending it to the predictor 202. The code amount calculator 201 grasps the actual generated code amount by inputting the generated code output from the multiplexer 105 and counting the data amount.

In step 606 following step 605, the i-th processing unit is encoded using the quantization table of the index value QP obtained in step 605.
When this encoding is completed, the process proceeds to step 607.

In step 607, it is determined whether or not the value of the variable i is equal to the number N (= 6) obtained by dividing the image data of one frame by the processing unit. When the encoding of the image data of one frame is completed, the determination is YES, and the series of processes is terminated here. Otherwise, the determination is no, the variable i is incremented in step 608, and the process returns to step 605.

Returning to the processing of step 605, the index value QP is obtained in consideration of the actual generated code amount of the newly encoded processing unit, and the quantization table of the obtained index value QP is used to The next processing unit is encoded. By repeating the processes of steps 605 to 608 until the determination of step 607 becomes YES, the encoding of all the processing units is completed.

As described above, in the present embodiment, according to the actual generated code amount of the processing unit coded up to that time,
The target code amount of the processing unit coded thereafter is adjusted. Therefore, the error between the actual generated code amount and the target code amount (determined in step 601) for each frame can always be kept within a small range. This makes it possible to suppress the fluctuation of the actual generated code amount within a small range even for an image in which the generated code amount greatly changes for each frame, such as a fast movement, and to accompany the fluctuation of the generated code amount. It is possible to always avoid deterioration of image quality on the receiving (decoding) side. Further, since it is possible to suppress unnecessary fluctuations in frame intervals (image update intervals), it is possible to avoid unnatural movement of a series of reproduced images on the receiving (decoding) side.

In this embodiment, when the target code amount of one frame is determined, basically, a value obtained by dividing the target code amount by the number of processing units dividing one frame is a target code of the processing unit. The amount is. However, especially when motion-compensated inter-frame prediction is performed, it is expected that the actual generated code amount will vary depending on the position of the processing unit on the image. For example, it can be expected that a background unit occupies a high ratio in a processing unit located at the edge of an image, such as the first processing unit, and clear image quality can be obtained with a small code amount. From this, it is desirable to set the weight for each processing unit and allocate the target code amount according to the weight of each processing unit so that the same image quality can be obtained in all the processing units. This is also the case when the remaining amount of the target code amount of the entire frame is assigned to another unprocessed processing unit after the coding is started and the coding of some processing units is completed. . In this way, when setting the weight for each processing unit, consider the viewpoints such as the part where the motion of the image is expected to be large and the part that is not large, the part to be reproduced with good image quality and the part that is not (eg background). It is desirable to set the shape of each processing unit and its position.

Further, in the present embodiment, the image data is encoded based on the DCT method, but the encoding method to which the present invention is applicable is not limited to this.
By setting the quantization step size for each processing unit, the present invention can be applied to other coding methods such as the run length coding method. The present invention can be widely applied to any encoding method that quantizes an original data value with a quantization step size, regardless of the type of data to be encoded.

[0057]

As described above, the present invention divides data into a plurality of processing units, and based on the actual generated code amount of the processing unit for which the compression processing has been completed, the processing unit which has not been compressed yet. In order to determine the quantization table (quantization step size) used for, the error between the target code amount and the actual generated code amount for each processing unit in the processing unit for which the compression processing has been completed, It can be absorbed by adjusting the generated code amount of the processing unit. As a result, it is possible to control the generated code amount with high accuracy, and it is possible to always keep the error between the actual generated code amount and the target code amount in the entire data within a small range. As a result, it is possible to avoid fluctuations in the generated code amount, and avoid deterioration in image quality and the like caused by fluctuations in the generated code amount.

[Brief description of the drawings]

FIG. 1 is a block diagram of a configuration of a data compression apparatus to which this embodiment is applied.

FIG. 2 is a block diagram of a code amount control unit.

FIG. 3 is an explanatory diagram showing how image data is divided in processing units.

FIG. 4 is an explanatory diagram showing a relationship between an index value QP and a generated code amount.

FIG. 5 is an explanatory diagram showing an example of a temporal change in buffer occupancy.

FIG. 6 is an operation flowchart of compression processing.

FIG. 7 is a conceptual diagram of a method of controlling a generated code amount.

FIG. 8 is a diagram showing a hierarchical structure of image data.

[Explanation of symbols]

 100 data compression device 103 converter 104 quantizer 106 buffer memory 107 code amount control unit 201 code amount calculator 202 QP predictor 203 table creator

Claims (3)

[Claims] 1. A method for controlling a code amount of data to be compressed using a specified quantization step size so as to be a preset target code amount, wherein the data is predetermined processing. Divide into a plurality of units, for each of the processing units, the target code amount, and determine the quantization step size based on the code amount of the other processing unit for which the encoding is completed, the determined quantization step A code amount control method, wherein a compression process is performed for each processing unit by using a size. 2. One of a plurality of quantization tables prepared in advance is selected, and the code amount of data compressed by using the selected quantization table is the coded data after the compression process. Is a method for controlling so that the target code amount is preset based on the storage capacity of the storage means that is temporarily stored, wherein the data is divided into a plurality of predetermined processing units, and each of the processing units is On the basis of the target code amount and the code amount of the other processing unit for which the compression processing has been completed, one is selected from the plurality of prepared quantization tables, and the selected quantization table is A code amount control method, wherein a compression process is performed for each processing unit by using the code amount control method. 3. A data compression apparatus comprising a storage means for selecting one of a plurality of quantization tables prepared in advance to perform data compression processing and temporarily storing the encoded data after the compression processing. Is a device for controlling so that the code amount of the data to be compressed is a target code amount that is preset based on the storage capacity of the storage means, and the data is processed in a predetermined processing unit. A dividing unit that divides into a plurality of units, one of the plurality of prepared quantization tables based on the target code amount and the code amount of another processing unit for which the encoding has been completed for each of the processing units. Choose one,
A code amount control apparatus comprising: a control unit that performs compression processing for each processing unit using the selected quantization table.