JP4567632B2 - Quantization step size change method, quantization step size calculation device, quantization step size change processing program, and recording medium thereof - Google Patents

Description

  The present invention relates to a video encoding technique, and more particularly to a technique for changing a quantization step size of a block to be encoded in video encoding in consideration of visual characteristics.

  In a video encoding method such as MPEG-2, an image can be divided into blocks, and the allocated code amount can be changed in units of blocks. This allocated code amount is controlled by the quantization step size. When the allocated code amount is increased, the quantization step size is decreased. Conversely, when the allocated code amount is decreased, the quantization step size is increased.

  The MPEG-2 test model (see Non-Patent Document 1) employs a mechanism for assigning a code amount in consideration of human visual characteristics in order to assign a code amount efficiently. This method uses a difference in sensitivity to human distortion. Distortion is conspicuous in a place where the luminance change is small (hereinafter referred to as a flat portion), and distortion is not conspicuous in a place where the luminance change is large (referred to as a texture portion) Is used. As a feature amount for discriminating the flat portion and the texture portion, the variance of the luminance signal in the encoding target block is used, and the quantization step size is changed based on the variance.

  However, when the comparison is simply performed by the variance, if the flat portion and the texture portion or the boundary between the flat portion and the flat portion exists in the block, the variance is larger than that of the simple flat portion. For this reason, if the quantization step size is increased as it is, distortion at the edge portion becomes conspicuous. Therefore, in the MPEG-2 test model, a block of 16 pixels × 16 pixels is divided into four small blocks of 8 pixels × 8 pixels, the variance for each of the divided small blocks is obtained, and the minimum variance of the four small blocks is determined as a block. Is distributed. According to such a method, if there is a flat region even at the edge portion, the dispersion becomes small and fine quantization is performed.

  FIG. 10 is a flowchart of the quantization step size calculation process in the MPEG-2 test model.

  First, the encoding target block is divided into four 8 × 8 pixel small blocks, and luminance distribution is obtained for each of the divided small blocks (step S100).

The minimum value of the obtained variances var ₀ , var ₁ , var ₂ , var ₃ is set as the feature amount act of the encoding target block (step S101).

act = min (var ₀ , var ₁ , var ₂ , var ₃ ) +1 (5)
Subsequently, the fluctuation amount Nact of the quantization step size is obtained from the following equation (step S102).

Here, act _avg is the average value of the feature values of the immediately previous encoded frame.

Using this result, the quantization step size QP of the encoding target block is calculated (step S103). The quantization step size QP of the encoding target block is the product of the basic quantization step size QP ₀ calculated by quantization control or the like and the quantization step size variation Nact obtained previously.

QP = QP ₀・ Nact (7)
In the MPEG-2 test model, quantization control considering visual characteristics is realized as described above.

  As prior art documents describing a technique for changing the quantization step size for each block, there are, for example, Patent Document 1 and Patent Document 2.

  Patent Document 1 describes a technique for changing the quantization step size in consideration of the flatness of a block with respect to the above-described MPEG-2 test model. However, the technique of Patent Document 1 is a technique closed to the encoding target block, and the technique of Patent Document 1 cannot solve the problem of the present invention described later.

In Patent Document 2, for the purpose of reducing the calculation cost, when calculating the feature value for changing the quantization step size, encoding is performed without calculating the feature value in the above MPEG-2 test model. A technique is described that realizes reduction of calculation cost by diverting values obtained in the process. However, this method is also a method closed to the encoding target block. That is, the feature amount used in the method of Patent Document 2 is calculated only from the value of the encoding target block, and the feature amount of the encoding target block is not calculated or corrected using the value of the neighboring block. Therefore, even the technique of Patent Document 2 cannot solve the problems of the present invention described later.
Japanese Patent Laid-Open No. 7-107481 JP 2005-252527 A MPEG-2 test model: MPEG-2, Test Model 5 (TM5), Doc.ISO / IEC JTC1 / SC29 / WG11 / N0400, Test Model Editing Committee, Apr. 1993.

  The conventional method of the MPEG-2 test model makes it possible to assign a code amount according to visual characteristics to some extent by using the feature amount of an image signal. However, depending on the characteristics of the image signal, it cannot be said that the code amount allocation consistent with the visual characteristics can be realized.

  FIG. 11 is a diagram for explaining the problems of the prior art. In this figure, the positional relationship of four macroblocks A to D is shown. In the prior art, the minimum value of luminance dispersion of the blocks in the macroblock is used as the feature amount. In the macroblocks B to D, since at least one block is a flat area, the feature amount of each macroblock is small. However, in the macro block A, the luminance dispersion of all the blocks increases due to the boundary of the face and the eyes. For this reason, the feature amount of the macroblock A becomes large. As a result, the macroblocks B to D are finely quantized while being a part of the same face, whereas the macroblock A is coarsely quantized, and the quantization distortion is concentrated on a part of the face. Disturbing.

  As described above, there is a possibility that the feature amount increases in a region where edges of a plurality of regions are complicated, and there is a problem that the visual characteristic and the code amount allocation are different. In particular, it becomes prominent when the bit rate is low or when the fluctuation amount is increased, and efficient code amount allocation cannot be performed.

  The present invention solves the above-described problems and uses the feature quantities of neighboring blocks to improve the determination accuracy of the area type of the block to be coded, and to efficiently allocate the code quantity without impairing visual characteristics. It aims at realizing.

  In order to solve the above problem, the present invention is characterized in that the quantization step size is changed using not only the feature amount of the block to be encoded but also the feature amount of the neighboring block.

  That is, in the conventional method, when a plurality of edges exist in a block, the block may be determined as a texture portion. On the other hand, in the present invention, a block that is determined to be a texture portion due to concentration of edges in a plurality of areas is determined in combination with a feature amount of a neighboring block, thereby determining whether or not it is an edge portion. Do things. Also, if there are many texture parts in the vicinity, the coding target block is likely to be a texture part where coding distortion is not conspicuous, so the quantization step size is made coarser and the generated code amount is saved. To be able to.

  As described above, the present invention can improve the determination accuracy of the region type (flat portion, texture portion, edge portion) of the block to be encoded by using the feature amount of the neighboring block, compared with the conventional method. Therefore, it is possible to assign an optimal code amount.

FIG. 1 is a quantization step size change processing flowchart showing an outline of the present invention. In FIG. 1, steps S2 and S3 are greatly different from the prior art. As shown in FIG. 1, in the present invention, in addition to the conventional feature quantity of the encoding target block, the feature quantity of the neighboring block of the encoding target block is calculated (steps S1 and S2). In the prior art, the variation of the quantization step size is calculated only from the feature amount of the block to be encoded. However, in the present invention, the magnitude relationship between the feature amount of the block to be encoded and the threshold and the feature of the neighboring block are calculated. Select neighboring blocks that have the same magnitude relationship with the threshold value, correct the feature quantity of the target block using the feature quantity of the neighboring block, and calculate the quantization step size variation from the feature quantity and (step S3), and which is changed in calculating the quantization step size based on (step S4).

There are the following methods for calculating the variation of the quantization step size. In the present invention, the latter method is used.
・ Method for correcting feature quantity of encoding target block ・ Method for calculating fluctuation amount of quantization step size based on magnitude relation with average value of feature quantity First, a method for correcting feature quantity of encoding target block will be described. . In this method, the feature amount of the encoding target block is corrected by the feature amount of the neighboring block, and the variation amount of the quantization step size is calculated using the corrected feature amount as the feature amount of the encoding target block.

FIG. 2 is a quantization step size change processing flowchart for correcting the feature amount of the encoding target block.
Step S10: The feature amount of the encoding target block is calculated.
Step S11: The feature amount of the neighboring block is calculated.
Step S12: Next, the feature amount of the encoding target block is corrected with the feature amount of the neighboring block.
Step S13: A variation amount of the quantization step size is calculated from the corrected feature amount.
Step S14: A quantization step size is calculated using the variation amount.

  In this method, when correcting the feature quantity of the encoding target block, a weighting factor is determined for each position of the neighboring block, and the feature quantity of each neighboring block is weighted by the weighting coefficient, and the feature of the encoding target block is determined. The amount may be corrected.

  Also, it has the same magnitude relationship between the feature value of the current block and the threshold value, with the average value of the feature value of the current frame, the average value of the feature value of the encoded frame, and a predetermined constant as threshold values. It is also possible to select a neighboring block and correct the feature amount of the encoding target block using only the selected neighboring block. For example, when the feature quantity of the neighboring block is large and the feature quantity of the encoding target block is small, the corrected feature quantity becomes large due to the influence of the neighboring block, and as a result, the feature quantity of the encoding target block is small. , There is a possibility to increase the quantization step size. Such a problem can be avoided by selecting a neighboring block to be used for correction.

  Next, a method for calculating the amount of change in the quantization step size from the magnitude relationship with the average value of the feature amounts will be described. In this method, the magnitude relationship between the threshold value of the feature amount of the encoding target block and the feature amount of the neighboring block is examined, and the variation in the quantization step size is calculated from the number of neighboring blocks having the same size relationship as the encoding target block. calculate.

FIG. 3 is a quantization step size change processing flowchart in the case of calculating the variation amount of the quantization step size from the magnitude relationship with the average value of the feature amounts.
Step S20: The feature amount of the encoding target block is calculated.
Step S21: The feature amount of the neighboring block is calculated.
Step S22: The feature amount of the encoding target block is compared with a predetermined threshold value.
Step S23: Also, the feature amount of the neighboring block is compared with a predetermined threshold value.
Step S24: Measure the number of neighboring blocks having the same magnitude relationship with the threshold as the encoding target block.
Step S25: The variation amount of the quantization step size is calculated from the value obtained by the measurement.
Step S26: A quantization step size is calculated using the fluctuation amount.

  Again, the average value of the feature values of the current frame, the average value of the feature values of the encoded frame, a predetermined constant, etc. are used as the threshold value. Also, when measuring the number of neighboring blocks that have the same magnitude relationship as the coding target block and the threshold, determine the weighting factor for each neighboring block position, measure the weighting factor, and calculate the number of neighboring blocks. Also good. For example, the up / down / left / right neighboring block in contact with the encoding target block is measured as 4, and the neighboring block existing on the diagonal is measured as 1.

  Also, the determination of the increase / decrease of the quantization step size can be determined by the magnitude relationship between the feature amount of the block to be encoded and the threshold value. For example, if the feature amount of the encoding target block is smaller than the threshold value, the quantization step size is set to decrease.

  Note that the weighting coefficient used in the present invention may be defined in advance according to the position of the neighboring block, or may be calculated according to the distance in contact with the encoding target block. In addition, as a feature amount of each block used in the present invention, a statistical amount (L1 variance, L2 variance, average value) of luminance signals in the block is assumed. Besides this, the statistic of the color difference signal can also be used.

  According to the above method, when the quantization step size is varied based on visual characteristics, the feature quantities of the block to be encoded and the neighboring blocks can be taken into account, and more efficient quantization control becomes possible. , Improvement in coding efficiency can be realized.

Examples of the related technology of the present invention (referred to as Example 1) and examples of the present invention (referred to as Example 2) are shown below. In Example 2, similarly to the conventional method, the macro block size and 16 × 16 pixels, and divides the macro block into small blocks of four 8 × 8 pixels, the variance calculated for each small block Let the minimum value be the feature quantity of the encoding target block. The quantization step size variation amount Nact is determined from this feature amount. The base quantization step size QP _fr is constant within the frame.

  FIG. 4 shows the relationship between the encoding target block (Current MB) and neighboring blocks (MB A to H). As shown in FIG. 4, the neighboring blocks use 8 blocks around the encoding target block.

FIG. 5 shows an example of weighting factors assigned to neighboring blocks. Weight coefficients w _i (i = 1, 2,..., 8) as shown in FIG. 5 are used according to the position of each neighboring block. However, if there is no neighboring block at the screen edge or the like, processing is performed with w _i = 0.

First, Example 1 will be described. FIG. 6 shows a flowchart of the first embodiment. First, the feature quantity var _j of all blocks in the encoding target frame is calculated (step S30). Subsequently, the feature amount is corrected by the following equation for all blocks (step S31). Note that the value before correction is used as the feature value of the neighboring block used for correction of the feature value.

Here, var _j represents the feature amount of the encoding target block, and varN _i represents the feature amount of the neighboring block (i = 1, 2,..., 8). Subsequently, an average value act of the corrected feature values (act with an overline in the formula; the same applies hereinafter) is calculated (step S32). From the corrected feature value act _j and the corrected feature value average value act, the quantization step size variation amount Nact _j of the block to be encoded is calculated (step S33).

The quantization step size QP of the encoding target block is the product of the base quantization step size QP _fr and the variation amount Nact _j obtained previously (step S34).

QP _j = QP _fr · Nact _j (10)
In this way, quantization control considering visual characteristics is realized.

  FIG. 7 is a diagram illustrating an apparatus configuration example according to the first embodiment. The quantization step size calculation apparatus 10 according to the first embodiment includes a block feature quantity calculation unit 11, a feature quantity storage unit 12, an encoding target block feature quantity extraction unit 13, a neighboring block feature quantity extraction unit 14, and an encoding target block feature quantity. A correction unit 15, a quantization step size variation calculation unit 16, and a quantization step size calculation unit 17 are provided.

  The block feature quantity calculation unit 11 calculates the feature quantities of all the blocks in the encoding target frame, and stores the calculated feature quantities in the feature quantity storage unit 12. The encoding target block feature quantity extraction unit 13 reads the feature quantity of the encoding target block stored in the feature quantity storage unit 12 and sends it to the encoding target block feature quantity correction unit 15. Also, the neighboring block feature quantity extraction unit 14 reads out the feature quantities of blocks in the vicinity of the encoding target block from the feature quantity storage unit 12 and sends them to the encoding target block feature quantity correction unit 15.

  The encoding target block feature amount correcting unit 15 corrects the feature amount of the encoding target block with the feature amount of the neighboring block. The quantization step size fluctuation amount calculation unit 16 calculates the fluctuation amount of the quantization step size of the encoding target block using the corrected characteristic amount of the encoding target block. The quantization step size calculation unit 17 calculates the quantization step size of the encoding target block using the variation amount calculated by the quantization step size variation calculation unit 16 and the base quantization step size. The above process is repeated for each encoding target block of the encoding target frame.

  In the first embodiment described above, the feature values of all neighboring blocks are not used as the feature values of the neighboring blocks used for correcting the feature values of the encoding target block, but the same method as the method of the second embodiment described later is used. Thus, it is possible to select a neighboring block having the same characteristics as the encoding target block and correct the feature quantity of the encoding target block using only the selected neighboring block.

  Next, Example 2 will be described. FIG. 8 shows a flowchart of the second embodiment. In this embodiment, the average value of the feature values of the encoding target frame is used as a threshold value for the magnitude relation comparison. The procedure is described below.

First, as in the first embodiment, the feature value var _j of all blocks in the encoding target frame is calculated (step S40). Subsequently, an average value var (in the equation, var with an overline; the same applies hereinafter) of the feature amount of the encoding target frame is calculated (step S41). Subsequently, the feature amount var _{j of the} block to be encoded is compared with the average value var within the frame (step S42).

Next, the feature value varN _i of the neighboring block and the average value var in the frame are compared (step S43), the number of neighboring blocks having the same magnitude relationship as the encoding target block is measured, and this value is set as the score score. (Step S44). For example, if var _j > var, the number of neighboring blocks having a feature amount larger than the average value var of the feature amount in the frame is examined and set as a score score. Here, when the score score is measured, the value to be added is weighted according to the position of the neighboring block.

  In this embodiment, the weighting factor shown in FIG. 5 is used. For example, if the neighboring block having the same magnitude relationship as the encoding target block is in the vertical and horizontal positions (MB B, MB D, MB E, MB G), 2 is added, and other neighboring blocks (MB A, MB A, In MB C, MB F, MB H), 1 is added. The variation amount of the quantization step size is calculated from the score score thus obtained (step S45).

In the present embodiment, a variation amount table (lookup table) corresponding to 0 to 12, which can be taken by the score score, is prepared in advance and converted into a variation amount Nact _j of the quantization step size according to each value. .

Nact _j = QP _offset [score] (11)
Note that the value of the fluctuation amount corresponding to the score score stored in the fluctuation amount table is determined in advance by, for example, experiments using various sample images, and is set in the fluctuation amount table.

The quantization step size QP of the encoding target block is calculated from the base quantization step size QP _fr and the variation amount Nact _j previously obtained (step S46). However, the expression differs depending on the magnitude relationship between the feature quantity var _j of the encoding target block and the intra-frame average value var.

  In this way, quantization control considering visual characteristics is realized.

  FIG. 9 is a diagram illustrating a device configuration example of the second embodiment. The quantization step size calculation device 20 according to the second embodiment includes a block feature amount calculation unit 21, a feature amount storage unit 22, an encoding target block feature amount extraction unit 23, a neighboring block feature amount extraction unit 24, a threshold storage unit 25, and a comparison. A unit 26, a comparison unit 27, a neighboring block number measurement unit 28, a quantization step size variation calculation unit 29, a variation table 30, and a quantization step size calculation unit 31.

  The block feature quantity calculator 21 calculates the feature quantities of all blocks in the encoding target frame, and stores the calculated feature quantities in the feature quantity storage unit 22. The encoding target block feature quantity extraction unit 23 reads the feature quantity of the encoding target block stored in the feature quantity storage unit 22 and sends it to the comparison unit 26. The neighboring block feature quantity extraction unit 24 reads each feature quantity of a block in the vicinity of the encoding target block from the feature quantity storage unit 22 and sends it to the comparison unit 27.

  The comparison unit 26 compares the feature amount of the block to be encoded with the threshold value stored in the threshold value storage unit 25. The comparison unit 27 also compares the feature amount of each neighboring block with the threshold value stored in the threshold value storage unit 25. In the second embodiment, the threshold value stored in the threshold value storage unit 25 is an average value of feature amounts of all blocks in the encoding target frame.

  Based on the comparison results of the comparison unit 26 and the comparison unit 27, the neighboring block number measuring unit 28 determines the size of the neighboring block in which the magnitude relationship between the neighboring block and the threshold is the same as the magnitude relationship between the feature amount of the encoding target block and the threshold. The number (score score) is measured. In the measurement of the score score, the value to be added is weighted according to the position of the neighboring block. The quantization step size fluctuation amount calculation unit 29 refers to the fluctuation amount table in which the fluctuation amount is set in advance according to the score score based on the measured number of neighboring blocks (score score), and determines the corresponding fluctuation amount. The amount of variation in the quantization step size. The quantization step size calculation unit 31 calculates the quantization step size of the encoding target block using the variation obtained by the quantization step size variation calculation unit 29 and the base quantization step size. The above process is repeated for each encoding target block of the encoding target frame.

  In this embodiment, the variation of the quantization step size is calculated by the product with respect to the base quantization step size, but it can also be realized by addition and subtraction. In addition, by calculating the feature values of all blocks first, the average value of the feature values of the encoding target frame is used as the average value of the feature values. However, the average value of the feature values of the immediately previous encoded frame is used. May be.

  In the second embodiment, the fluctuation amount of the score and the quantization step size is obtained from one fluctuation amount table (lookup table), but two types of tables for increase and decrease may be prepared. It can also be obtained from an expression using the score as a parameter. In addition, the amount of variation can be limited to only increasing or decreasing.

  The above quantization step size change processing can be realized by a computer and a software program. The program can be provided by being recorded on a computer-readable recording medium or via a network. is there.

It is a processing flowchart which shows the outline | summary of this invention. It is a quantization step size change process flowchart in the case of correct | amending the feature-value of an encoding object block. It is a quantization step size change process flowchart in the case of calculating the variation | change_quantity of quantization step size from the magnitude relationship with the average value of a feature-value. It is a figure which shows the relationship between an encoding object block and a neighboring block. It is a figure which shows the example of the weighting coefficient provided to the neighborhood block. 3 is a flowchart of the first embodiment. 1 is a diagram illustrating a device configuration example of Embodiment 1. FIG. 10 is a flowchart of Example 2. FIG. 6 is a diagram illustrating an apparatus configuration example according to a second embodiment. It is a flowchart of a quantization step size calculation process in the MPEG-2 test model. It is a figure explaining the problem of a prior art.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10,20 Quantization step size calculation apparatus 11,21 Block feature-value calculation part 12,22 Feature-value memory | storage part 13,23 Encoding object block feature-value extraction part 14 24 Neighboring block feature-value extraction part 15 Encoding object block feature-value Correction unit 16, 29 Quantization step size variation calculation unit 17, 31 Quantization step size calculation unit 25 Threshold storage unit 26, 27 Comparison unit 28 Neighboring block number measurement unit 30 Variation table

Claims (8)

A quantization step size changing method for changing a quantization step size of a block to be encoded in consideration of visual characteristics in a video encoding method for encoding in block units,
The process of calculating the variance of the luminance signal or chrominance signal in each block as a feature amount,
A process of comparing the feature quantity of the encoding target block with a predetermined threshold;
A process of comparing a feature quantity of a neighboring block of the encoding target block with the threshold;
Selecting a neighboring block whose magnitude relationship between the neighboring block and the threshold is the same as the magnitude relation between the feature quantity of the coding target block and the threshold;
Correcting the feature quantity of the encoding target block by a weighted average of the feature quantity of the selected neighboring block and the feature quantity of the encoding target block;
To determine the direction in which the feature amount of the encoding target block and the corrected Ru to increase or decrease the quantization step from the magnitude relationship between the threshold value, the amount of variation of the quantization step size according to the feature quantity of the encoding target block The process of calculating
And a step of calculating the quantization step size of the block to be encoded using the calculated variation amount of the quantization step size. A quantization step size changing method for changing a quantization step size of a block to be encoded in consideration of visual characteristics in a video encoding method for encoding in block units,
The process of calculating the variance of the luminance signal or chrominance signal in each block as a feature amount,
A process of comparing the feature quantity of the encoding target block with a predetermined threshold;
A process of comparing a feature quantity of a neighboring block of the encoding target block with the threshold;
A process of measuring the number of neighboring blocks in which the magnitude relationship between the neighboring block and the threshold is the same as the magnitude relation between the feature amount of the coding target block and the threshold;
Wherein determining the direction in which Ru increase or decrease the quantization step from the magnitude relation of the feature of the encoding target block and the threshold value, based on the amount of variation that is predetermined in accordance with the number of neighboring blocks of the quantization step size The process of calculating the variation,
And a step of calculating the quantization step size of the block to be encoded using the calculated variation amount of the quantization step size. The quantization step size changing method according to claim 2,
When measuring the number of neighboring blocks that have the same magnitude relationship between the feature quantity of the coding target block and the threshold, the weighting coefficient is determined from the position of the neighboring block, and the above-mentioned decision is made when measuring the number of neighboring blocks. A quantization step size changing method characterized by adding a weight coefficient. In the quantization step size changing method according to any one of claims 1 to 3,
As a threshold for comparing the magnitude relationship, the magnitude relationship is compared using either an average value of feature values of the current frame, an average value of feature values of an encoded frame, or a predetermined constant. How to change the quantization step size. A quantization step size calculation device that calculates a quantization step size of a block to be encoded in consideration of visual characteristics in a video encoding method that performs encoding in block units,
Means for calculating the variance of the luminance signal or chrominance signal in the block as a feature amount for each block;
Means for storing the calculated feature value;
Means for storing a predetermined threshold;
Means for comparing the feature quantity of the encoding target block read from the means for storing the feature quantity with a threshold value stored in the means for storing the threshold value;
Means for comparing a feature quantity of a neighboring block of the encoding target block read from the means for storing the feature quantity with a threshold value stored in the means for storing the threshold value;
Means for selecting a neighboring block in which the magnitude relationship between the neighboring block and the threshold is the same as the magnitude relation between the feature amount of the coding target block and the threshold;
Means for correcting the feature quantity of the encoding target block by a weighted average of the feature quantity of the selected neighboring block and the feature quantity of the encoding target block;
To determine the direction in which the feature amount of the encoding target block and the corrected Ru to increase or decrease the quantization step from the magnitude relationship between the threshold value, the amount of variation of the quantization step size according to the feature quantity of the encoding target block Means for calculating
Means for calculating the quantization step size of the block to be encoded using the calculated amount of variation in the quantization step size. A quantization step size calculation device that calculates a quantization step size of a block to be encoded in consideration of visual characteristics in a video encoding method that performs encoding in block units,
Means for calculating the variance of the luminance signal or chrominance signal in the block as a feature amount for each block;
Means for storing the calculated feature value;
Means for storing a predetermined threshold;
Means for comparing the feature quantity of the encoding target block read from the means for storing the feature quantity with a threshold value stored in the means for storing the threshold value;
Means for comparing a feature quantity of a neighboring block of the encoding target block read from the means for storing the feature quantity with a threshold value stored in the means for storing the threshold value;
Means for measuring the number of neighboring blocks in which the magnitude relation between the neighboring blocks and the threshold is the same as the magnitude relation between the feature quantity of the coding target block and the threshold;
Wherein determining the direction in which Ru increase or decrease the quantization step from the magnitude relation of the feature of the encoding target block and the threshold value, based on the amount of variation that is predetermined in accordance with the number of neighboring blocks of the quantization step size Means for calculating the amount of variation;
Means for calculating the quantization step size of the block to be encoded using the calculated amount of variation in the quantization step size.   A quantization step size change processing program for causing a computer to execute the quantization step size change method according to any one of claims 1 to 4.   A computer-readable recording medium recording a quantization step size change processing program for causing a computer to execute the quantization step size changing method according to any one of claims 1 to 4.

Images