JP2001148856A - Code quantity controller, code quantity control method, and code quantity control program recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image encoder that encodes an image in a prescribed processing time in the case of encoding the image so as not to exceed an object code quantity and that identifies a natural image from an artificial image so as to obtain an image with high quality and less distortion even in the case of the artificial image. SOLUTION: Decreasing an object code quantity predicted by a quantization width prediction section 111 as an error quantity predicted by an error range prediction section 113 increases so as to prevent an object code quantity from being highly exceeded. Furthermore, smoothing a distribution of a block assigned code quantity prediction table of a block assignment code quantity setting section 112 accompanying the increase in the error predicted by the error range prediction section 113 corrects a state where the assignment of the code quantity is nearly zero.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a code amount control device,
The present invention relates to a code amount control method and a recording medium storing a code amount control program.

[0002]

2. Description of the Related Art In recent years, an image compression technique using highly efficient variable length coding has been used.

[0003] Digital still cameras and the like have been developed as products to which the image compression technique is applied, and in such products, the number of recorded images is specified. On the other hand, the encoding method using variable length encoding has a problem that a fixed number of recorded images cannot be guaranteed because the data amount varies depending on the image.

In order to solve the above problems, the following method is known.

That is, the quantization width of the quantization used for controlling the compression ratio in the variable length coding is temporarily determined, the encoding process is performed using the quantization width, and the total code amount of the obtained image is obtained. When,
By repeating the comparison with the target code amount one or more times, the optimum quantization width for approaching the target code amount is predicted. Next, a variable length code is output using the quantization width obtained by the prediction.

[0006] Japanese Patent Application Laid-Open No. 10-336661 discloses an image encoding apparatus, an image encoding method, and a recording medium on which an image encoding program is recorded. In order to solve the problem, paying attention to the characteristic of the natural image that the frequency component patterns of adjacent pixels are similar, an operation of predicting a quantization width from a result of scanning by thinning out between adjacent pixels or divided blocks is performed. By doing so, it is possible to reduce the amount of prediction calculation, improve the performance of the encoding process, and improve the target code, as compared with the related art in which encoding is performed at least once in order to predict the generated code amount of an image. A method of encoding with a code amount error equal to or less than a certain amount from the amount is disclosed.

FIG. 3 is a graph showing the relationship between the quantization width and the generated code amount. As shown in FIG. 3, there is a correlation between the quantization width and the generated code amount depending on the amount of the unique frequency component included in the image. A characteristic curve is drawn on a graph, and the quantization width Q and n The relational expression with the code amount L when an original image composed of 8 × 8 pixel blocks is encoded can be expressed by Expression (1), where d is a prediction error due to prediction of the code amount.

L = f (Q) · n + d. . . (1) When the code amount L when the quantization width Q takes Qα and Qβ respectively is Lα and Lβ, the characteristic curve f () is obtained from the measurement of these two points as shown in FIG. Characteristic curve f ()
The quantization width Qp, which is required to generate the target code length Lp, on the trajectory of is represented by the reciprocal f ′ of f () in equation (1).
If () is taken and if the maximum code amount error of the prediction error d that can be taken when processing the image currently being processed when L = Lp is D, the equation (2) can be obtained.

Qp = f ′ ((Lp−D) / n). . . (2) In a natural image, since the similarity is seen in the code amount of each of the adjacent pixel blocks, even when the prediction is performed by thinning out the pixel blocks, the maximum code amount error D of the result is reduced to a substantially constant rate or less. Since it can be considered that D ≒ A · Lp (A is a constant less than 1), the optimum quantization width Qp for obtaining the target code length Lp can be uniquely obtained from Expression (3). .

Qp = f ′ ((1−A) / n · Lp). . . (3) FIG. 4 is an explanatory diagram of a code amount prediction procedure of the conventional code amount control device.

Reference numeral 401 denotes an input pixel code amount predicting unit which divides and reads pixel information into a plurality of blocks, performs frequency conversion, quantizes, predicts the code amount, and calculates the code amount of the block on the entire screen. Reference numeral 411 denotes a quantization width prediction unit, and code amounts Lα and Lβ when the quantization width of the quantization of the input pixel code amount prediction unit 401 takes Qα and Qβ, respectively.
By summing up the code amount of, the optimum quantization width Qp for obtaining the target code length Lp of the curve characteristic f ′ () of equation (3) is obtained. Reference numeral 412 denotes a block allocation code amount prediction unit. The target code amount is L based on the allocation allocation of the code amount to each block predicted by the input pixel code amount prediction unit 401.
The assigned code amount when p is obtained is obtained.

The above detailed description is disclosed in Japanese Patent Application Laid-Open No. 10-3366.
No. 61, the description is simplified.

[0013]

However, in the above-described conventional configuration, the image is divided into a plurality of blocks and compression-encoded because the image has strong similarity in information between adjacent pixels. Assuming the characteristic of a natural image that the generated code amount in the case is similar, the code amount is predicted from the thinned sampling result between input pixels.

However, the artificial image does not apply to natural images composed of images of landscapes and living things.
There is a type of image, such as a geometric image, in which information similar between adjacent pixels has little similarity, and the prediction error may be larger than that of a natural image.

If the prediction error of the code amount becomes too large,
The code amount does not fall within the prediction range. If the prediction error is eliminated by using a method of reducing the frequency component by correcting the code amount, the frequency component is excessively reduced in a portion of the screen where the generated code amount is large, so that the block distortion appears large enough to be visually identified. .

In the case where the compression coding is performed by estimating the natural image in advance with the maximum value of the prediction error having the same size as that of the artificial image, the compression coding of the natural image which is normally performed is required to be larger than the target size. Since the coding method is too small and the coding method is inversely inverted, the image quality of the entire screen of the natural image is degraded more than necessary at the time of decoding.

Encoding a natural image without deteriorating the conventional image quality so that the amount of code of the entire coded image does not exceed the target amount of code while suppressing the above-described phenomenon that block distortion occurs in the artificial image. However, the present invention is a problem to be solved.

[0018]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and it is an object of the present invention to provide an encoding apparatus capable of performing efficient encoding by identifying and compressing image quality characteristics of an artificial image. Aim.

[0019]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The maximum value of the prediction error has the characteristic of increasing or decreasing according to the degree of discreteness of the code amount for each block of the entire image.

[0020]

(Equation 1)

It can be obtained as a solution U of equation (4),
The prediction error D in the equation (2) is obtained as a value proportional to the standard deviation σ.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

As described above, the code amount for each block generated from an artificial image of a type in which the similarity between pixels is not very similar is more discrete than the code amount of a natural image, and the pixels are thinned out. Causes a large error in the code amount prediction by.

FIG. 1 shows the configuration of a code amount control device according to the present invention.

In the code amount control apparatus of FIG. 1, reference numeral 11 denotes a code amount control apparatus, and 12 denotes original image information which is image data before compression encoding read by the code amount control apparatus 11;
Is a target code amount for which the code amount control device 11 sets the target code amount, 14 is an optimal prediction quantization width obtained and output from the calculation result of the code amount control device 11, and 15 is a code amount control device 11. A block allocation code amount obtained and output from the calculation result, and 16 is a control signal.

Hereinafter, each element constituting the code amount control device 11 will be described.

The code amount control unit 11 includes an inter-pixel reading unit 1
01, a frequency conversion unit 102, a quantization unit 103, a code amount prediction unit 104, a prediction quantization width generation unit 105,
Quantization width prediction unit 111 and block allocation code amount setting unit 1
12, an error range prediction unit 113, and an encoding control unit 121
Consists of

The operation of the code amount control device configured as described above will be described below.

When the control signal 16 is input, the encoding control unit 121 controls all the processes of the code amount control device of FIG. 1 in the following order.

The pixel-to-pixel reading unit 101 stores the original image information 12
Is read as thinning information while thinning between blocks. The frequency converter 102 converts the frequency of the thinning information. The quantization unit 103 quantizes the result of the frequency conversion.

Here, the prediction quantization width generation unit 105
The compression ratio is a quantization width close to generating the target code amount from a value obtained by dividing the original image data size obtained from the original image information 12 by the target code amount obtained from the original image information 12, and
A quantization width Qα for generating a code amount larger than the target code amount and a quantization width Qβ for generating a code amount smaller than the target code amount are calculated and used as the quantization width of the quantization unit 103.

The code amount prediction unit 104 predicts and obtains a code length from the run length of the DC component and the AC component as a result of the quantization.

The quantization width prediction unit 111 includes a code amount prediction unit 1
04, the code lengths Lα and Lβ when the prediction quantization widths Qα and Qβ are used are obtained from the cumulative total of the code length prediction results, and f of the equation (2) which is a curve expression of the quantization width and the code length is obtained. '()
Ask for. As a specific method, a table of the relationship between the quantization width and the generated code amount is prepared in advance.

The block allocation code amount setting unit 112 obtains the allocation code amount when the target code amount is Lp, based on the allocation allocation of the code amount to each block predicted by the code amount prediction unit 104.

However, if the degree of discreteness of the code obtained by the error range prediction unit 113 according to the equation (4) is a large value that is far from the value that can be taken by the natural image, an image having the characteristics of an artificial image Can be considered. The code amount maximum error D in the equation (2) can be regarded as a value proportional to the discreteness of the code length of the block in the screen.

FIG. 5 is a graph showing the relationship between the quantization width of the natural image and the artificial image, the generated code amount, and the error. As shown in FIG. Becomes larger in proportion to the degree of discreteness of the code amount of, the target code amount predicted by the quantization width prediction unit 111 is reduced with an increase in the prediction value of the error range prediction unit 113, so that the target code amount increases. Thus, it is possible to suppress a phenomenon that a large amount of frequency component suppression occurs in the artificial image.

FIG. 2 shows an image encoding device 21 using the code amount control device of the present invention.

The following is the same as the code amount control device of FIG. 11 is the code amount control device of FIG. 1, 12 is original image information, 13 is a target code amount, 14 is a prediction quantization width, 16
Is a control signal. Further, the frequency conversion unit 102 in FIG.
The quantization unit 103 and the encoding control unit 121 have the same functions as the code amount control device in FIG. 1 and are processing devices that can be shared.

Reference numeral 22 denotes an original image information buffer for storing original image information, and outputs the original image information 12. A code amount setting unit 23 outputs a target code amount 13. 201, a quantization width storage unit; 202, an image encoding unit; a block reading unit 202a, which is a component, and a frequency conversion unit 102
, A quantization unit 103 and an entropy encoding unit 202b. 211 is a block allocation code amount storage unit, and 212 is a code amount correction unit.

When the control signal 16 is input, the encoding control section 121 controls all the processes of the image encoding apparatus shown in FIG. 2 in the following order.

First, the code amount control device 11 reads the original image information 12 from the original image information buffer, reads the target code amount 13 from the code amount setting unit 23, predicts the code amount, and sets the prediction quantization width 14 The block allocation code amount 15 is recorded in the block width storage unit 201 and the block allocation code amount storage unit 211.

Next, in the image encoding unit 202, the original image information 12 is input to the block reading unit 202 a, frequency conversion is performed by the frequency conversion unit 102, and quantization is performed by the quantization unit 103 using the value of the quantization width storage unit 201. And entropy encoding section 202
b and output to the coded information buffer 24. Here, the entropy encoding unit 20 in the code amount correction unit 212
2b is compared with the total code amount allocation predicted by the block allocation code amount storage unit 211, and if the total code amount exceeds the actual code length, the entropy coding unit 202b outputs By limiting the code amount by suppressing the frequency components in the output code, the code amount can be suppressed so that the code amount does not exceed the target value even if it is slight.

With the above configuration, the code amount generated from the quantization width predicted by the code amount control device in FIG. 1 does not always need to exceed the target value, and the code amount causes block distortion due to suppression of frequency components. Since it is sufficient if the level is not a small amount, a code amount closer to the target code amount can be generated.

FIG. 6 shows an example of code amount allocation for an artificial image.

As shown in FIG. 6, an extremely small code amount distribution table value 601 in which the code amount is allocated near zero may be created, and the code amount of the code amount distribution table value 601 is actually encoded. The code amount distribution is estimated to be extremely smaller than the code amount distribution 602 at the time of the correction, and when the target code amount of the image encoding apparatus in FIG. A block 603 in which the deterioration of the image is noticeable occurs.

As a means for solving the above problem, the steepness of the difference between the distribution values of the block allocation code amount prediction table for suppressing the code amount following the increase of the error range is reduced, thereby reducing the code amount distribution. A corrected code amount distribution table value 604 that corrects the state where the allocation of the code amount of the table value 601 is near zero is created, and the corrected code amount distribution table value 6
04 is used as a code amount distribution table.

According to the above configuration, in an artificial image having a steep code amount distribution, there is no block that is expected to have an extremely small code amount allocation near zero, so that the code amount can be reduced by suppressing frequency components when a prediction error occurs. Can be prevented from occurring.

In the embodiment, the description has been made that the pixel-to-pixel reading unit 101 performs the thinning between the blocks. However, the thinning between the pixels may be used for the prediction calculation. In addition, although the prediction quantization width is set to two types, any one or more types may be used. A table of the relationship between the quantization width and the generated code amount may be prepared for each color component. Also, a table of the relationship between the quantization width and the generated code amount may be prepared for each color component.

As a code amount control method, a configuration may be added in which an arithmetic table for performing variable length coding optimal for generation of a target code amount is frequently assigned to high frequency components in accordance with the increase in the error range.

The frequency converter 102, the quantizer 103, and the encoding controller 121 of the image encoding apparatus shown in FIG. 2 share the same apparatus as the code amount controller shown in FIG. Another device may be used, and the pixel thinning process may be a combination of thinning between pixels and thinning between pixels.

The present invention relates to an image coding method, which operates in the same manner as the above-described image coding apparatus and has the same effects. A minimum configuration of a digital still camera as one embodiment of the present invention will be described with reference to a block diagram shown in FIG.

In FIG. 7, 701 is a microcomputer, 702 is a ROM, 703 is a RAM, 704 is an imaging unit, and 705 is a display unit. Microcomputer 70
1 loads and executes a program for controlling the amount of code from the ROM 702 and coding the code to a target size or less. The image information is read from the imaging unit 704, and the image is encoded and stored using the RAM 703 as a work area.

FIG. 8 shows a flow of the code amount control procedure of FIG.

An initial value setting step 801 sets a target code amount and image data size. Reference numeral 802 denotes a quantization width generation step for prediction. The compression ratio is a quantization width close to generating the target code amount from a value obtained by dividing the size of the image captured by the imaging unit 704 in the initial value setting step, and
A quantization width Qα for generating a code amount larger than the target code amount and a quantization width Qβ for generating a code amount smaller than the target code amount are calculated and used as the quantization width of the quantization unit 103.

Reference numeral 803 denotes a pixel-to-pixel reading step in which the image of the image pickup unit 704 is divided into blocks each having a plurality of pixels, and input while performing predetermined thinning as shown in FIG.
A frequency conversion step 804 converts the frequency of the pixel read in the pixel-to-pixel reading step 803. A quantization step 805 quantizes the result of the frequency conversion in the frequency conversion step 804 with a quantization width Qα or Qβ. 8
A code amount prediction step 06 predicts the code amount from the DC component and the AC component amount when the quantization result of the quantization step 805 is encoded, and calculates the code amount.

Reference numeral 807 denotes a code amount distribution totaling step, which is a RAM.
On the work area 703, a table of the distribution of the code amount predicted in the code amount prediction step 806 for each block is created. Numeral 808 denotes a step of determining the end of quantization. The calculation between steps 805 to 807 is repeated in two cases, that is, when the quantization width is Qα and when the quantization width is Qβ. Reference numeral 809 denotes a step for determining the end of the image, which is repeated until all code amount prediction sampling is completed. 810
In the error range prediction (error correction value setting) step, the amount of error included in the result of the code amount distribution totalizing step 807 after all code amount prediction sampling is completed is

[0057]

(Equation 2)

The correction value is obtained as a correction value proportional to the solution U in the equation (5).

Reference numeral 811 denotes a quantization width prediction step, in which the quantization widths are Qα, Qβ
Are calculated respectively when L is
Correction is performed using the correction values obtained in the two-point error range prediction step 810, and an optimum quantization width is obtained as a target code amount.
Reference numeral 812 denotes a block allocation code amount setting step, in which a code amount distribution for each block is obtained from the value of the predicted code amount, and suppression of the code amount when the target code amount is exceeded in proportion to the increase in the error range. Is performed, the steepness of the difference between the distribution values of the block allocation code amount prediction table is reduced.

The present invention provides a code amount predicting step of recording the predicted code amount for each block, an error range predicting step of obtaining a code amount prediction error range by the thinning out from the discreteness of the code amount, A quantization width prediction step for obtaining a quantization width whose encoding result is closest to a desired code amount based on the dependency of the sum of the amounts and the quantization width used at the time of prediction; and A code amount control method comprising: obtaining an optimal quantization width that allows an encoding result to be within a target code amount by specifying a value obtained by subtracting a value of the code amount prediction error range from the target code amount.

According to the present invention, there is provided a computer-readable recording medium having recorded thereon a program relating to an image encoding method, wherein a program for causing a computer to execute the image encoding method is recorded. It is a storage medium.

As described above, according to the present embodiment, by providing the error range prediction unit and correcting the quantization width and the block allocation code amount, the result of coding by the code amount prediction becomes the target value. Since it does not greatly exceed, the distortion of the image can be corrected.

[0063]

As described above, the present invention provides an error range prediction unit to estimate the fluctuation of the code amount prediction error, and corrects the prediction quantization width and the block allocation code amount to achieve the target at the time of compression encoding. A method of encoding with a code amount error equal to or less than a certain value from the code amount is shown.

Since it is possible to estimate the fluctuation of the prediction error of the code amount of each image, it is possible to maintain the coding quality of the natural image and to reduce the deterioration of the image due to the block distortion due to the reduction of the frequency component even in the artificial image having a larger prediction error. Can be prevented.

In the case of predicting the code amount to keep the data size after encoding at a certain level or less, an image code capable of correcting the occurrence of image distortion in an artificial image by only one type of code amount prediction configuration. The method can be realized.

[Brief description of the drawings]

FIG. 1 is a diagram showing a code amount control device according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating an image encoding device using an image encoding method according to an embodiment of the present invention.

FIG. 3 is a graph showing a relationship between a quantization width and a generated code amount.

FIG. 4 is an explanatory diagram of a code amount prediction procedure of a conventional code amount control device.

FIG. 5 is a graph showing the relationship between the quantization width of a natural image and an artificial image, the generated code amount, and an error.

FIG. 6 is a diagram showing an example of code amount allocation of an artificial image.

FIG. 7 is a minimum configuration diagram of a digital still camera.

FIG. 8 is a flowchart of a code amount control procedure.

FIG. 9 is a diagram illustrating an example of a pixel thinning range.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 11 Code amount control apparatus 12 Original image information 13 Target code amount 14 Predictive quantization width 15 Block allocation code amount 16 Control signal 21 Image encoding device 22 Original image information buffer 23 Code amount setting unit 24 Encoding information buffer 101 Pixel thinning Reading unit 102 Frequency conversion unit 103 Quantization unit 104 Code amount prediction unit 105 Quantization width generation unit for prediction 111 Quantization width prediction unit 112 Block allocation code amount setting unit 113 Error range prediction unit 121 Coding control unit 201 Quantization width Storage unit 202 Image encoding unit 202a Block reading unit 202b Entropy encoding unit 211 Block allocation code amount storage unit 212 Code amount correction unit 401 Input pixel code amount prediction unit 411 Quantization width prediction unit 412 Block allocation code amount prediction unit 701 micro Computer 702 ROM 703 RAM 704 Image unit 705 Display unit 801 Target code amount setting 802 Predictive quantization width generation 803 Reading between pixels 804 Frequency conversion 805 Quantization 806 Code amount prediction 807 Code amount distribution total 808 Quantization end determination 809 Image end determination 810 Error range Prediction (error correction value setting) 811 Quantization width prediction 812 Block allocation code amount setting

Continued on the front page F term (reference) 5C059 KK03 KK22 LB07 MA01 ME01 PP19 SS11 TA06 TA53 TA57 TB08 TC03 TC19 TC38 TD14 UA02 UA33 5C078 BA57 CA02 CA21 DA00 DA01 DA07 DA11 DA21 DB04 DB07 EA00

Claims (10)

[Claims] A code amount control device for predicting a code amount before encoding an image so that the code amount of the entire encoded image does not exceed a target code amount. A pixel reading unit that divides the image into blocks of pixels and performs input while performing predetermined thinning, a frequency converting unit that obtains a frequency component of a pixel obtained by the pixel reading unit, and at least one type of the frequency component. Quantization means for quantizing with the above quantization width, code amount prediction means for predicting a code amount of the image after encoding with the quantization width from the quantization result, and code amount prediction means Code amount distribution totalizing means for recording the obtained predicted code amount for each block; error range predicting means for obtaining a code amount prediction error range by the thinning from the discreteness of the code amount for each block; Painting From the relationship between the total sum and the quantization width used at the time of prediction, a quantization width prediction unit for obtaining a quantization width whose encoding result is closest to a desired code amount, and the quantization width prediction unit includes the desired code amount. A code amount control device capable of obtaining an optimal quantization width in which an encoding result is contained, wherein a value obtained by subtracting a value of the code amount prediction error range from the target code amount is specified. 2. A method for determining the allocation of the allocated code amount for each of the divided blocks from the value of the predicted code amount, and suppressing the code amount so that the target code amount does not exceed the allocated code amount. Block allocation code amount setting means capable of controlling the above, and making a large correction to reduce the steepness of distribution to the entire average level of the block allocation code amount prediction table in proportion to the increase amount of the error range. And a code amount control device for preventing the code amount allocation from being predicted near zero. 3. The code amount of the entire coded image is allowed to slightly exceed the target code amount in advance, and an optimum quantization width is obtained by using a smaller estimated value of the error range. The encoding of the image is performed from the encoding width, and the encoding after the encoding of the image is performed by reducing the frequency component when performing the encoding exceeding the total of the code amount allocation for each block in the middle of the encoding. 3. The code amount control device according to claim 2, wherein the amount is a value not exceeding the target code amount. 4. An image using the code amount control device according to claim 1, wherein the image is encoded with an optimum quantization width determined so that the code amount of the entire image does not exceed the target code amount. Encoding device. 5. A code amount control method for predicting a code amount before encoding an image so that the code amount of the entire encoded image does not exceed a target code amount. A pixel-to-pixel reading step of inputting while dividing into blocks of pixels and performing predetermined thinning, a frequency conversion step of obtaining a frequency component of a pixel obtained by the pixel-to-pixel reading step, and at least one of the frequency components A quantization step of quantizing with the above quantization width, and a code amount prediction step of recording, from the quantization result, a predicted code amount of an image having been encoded with the quantization width for each of the blocks, An error range prediction step for obtaining a code amount prediction error range by the thinning from the code amount discreteness, and a coding result obtained from a dependency of a sum of the prediction code amounts and a quantization width used at the time of prediction. A quantization width prediction step of obtaining a quantization width closest to a desired code amount, and the quantization width prediction step calculates a value obtained by subtracting a value of the code amount prediction error range from the target code amount to the desired code amount. A code amount control method characterized in that an optimum quantization width in which a coding result falls within a target code amount is obtained by designating the code amount. 6. The code amount is obtained by calculating the allocation of the allocated code amount for each of the divided blocks from the value of the predicted code amount and suppressing the code amount so that the target code amount does not exceed the allocated code amount. A code for setting a block allocation code amount that enables the control of (a), and a code for greatly adding a correction to alleviate the steepness of distribution to the entire average level of the block allocation code amount prediction table in proportion to the increase amount of the error range. Quantity control means. 7. The code amount of the entire coded image is allowed to slightly exceed the target code amount in advance, and an optimum quantization width is obtained by using a smaller estimated value of the error range. The encoding of the image is performed from the encoding width, and the encoding after the encoding of the image is performed by reducing the frequency component when performing the encoding exceeding the total of the code amount allocation for each block in the middle of the encoding. 7. The code amount control means according to claim 6, wherein the amount is a value not exceeding the target code amount. 8. An image using the code amount control means according to claim 6, wherein the image is encoded with an optimum quantization width determined so that the code amount of the entire image does not exceed the target code amount. Encoding means. 9. A method according to claim 5, wherein said predictive code amount calculating step uses a standard deviation for each block of the entire image for calculating a prediction error of a predictive code amount of a block represented by the following equation. Control method. 10. A computer-readable recording medium on which a program relating to a code amount control method is recorded, wherein the computer-readable recording medium stores a program for causing a computer to execute the image encoding method according to claim 5. Recording medium characterized by the following.