JP2000134626A - Motion vector encoding and decoding method and storage medium recording the same method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a motion vector encoding/decoding method and a storage medium which records the same method that can reduce the size of data to be used for encoding/decoding of a motion vector. SOLUTION: At first steps (S1 to S4), a length component of a motion vector is encoded. At second steps (S5 to S8), a direction component of the motion vector is encoded. Here, the second steps includes a step (S6) which calculates an area to which the direction component of the motion vector belongs in a predetermined coordinate system divided into plural areas in advance and which is the coordinate system where the direction of the motion vector are coordinated to a coordinate position and obtains area information displaying the area, and a step (S7) for encoding the area information and the position information.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a motion vector encoding / decoding method for encoding and decoding a motion vector used for motion compensation inter-frame prediction in a moving picture encoding method, and a recording medium on which the method is recorded.

[0002]

2. Description of the Related Art Conventionally, when encoding a moving image, as a technique for reducing redundancy in the time direction, an inter-frame predicting a motion vector on a frame at another time from an already encoded frame is known. Predictive coding is known. In order to increase the prediction efficiency in the time direction, a motion-compensated inter-frame prediction method for predicting a motion vector using a motion-compensated image signal as a prediction signal is used. Since the motion vector is necessary for motion compensation at the time of decoding, it is encoded and transmitted to the decoding side.

Normally, since motion compensation is performed in units of regions such as blocks, individual motion vectors are generated for each region. It is known that these motion vectors have a high correlation between neighboring areas. Therefore, a method is used in which the motion vector of the coding target region is predicted from the motion vector of the neighboring region of the coding target region, and the prediction error is transmitted to reduce the redundancy of the motion vector (that is, the amount of redundant code). Can be

[0004] For example, IS that defines a moving picture coding method
In O / IEC 11172-2 (MPEG-1), motion compensation is performed in block units, and the motion vector of the current block is predicted from the motion vector of the block immediately before the current block. The motion vector used in MPEG-1 is a vector in an orthogonal coordinate system, and is composed of a horizontal component x and a vertical component y. The following formulas are used to calculate the respective constituents xd and yd of the prediction error vector of each component of the motion vector. .

[0005] x_d= X_i-X_i-1 (1) y_d= Y_i-Y_i-1 (2) Here, xi and yi are the motion vectors of the current block.
Is a component of torr, x _i-1, Y_i-1Is the previous block
This is a component of the motion vector.

The prediction error vector (x _d , y _d ) is shown in FIG.
Has a distribution concentrated near (0,0),
The smaller the value of each component, the higher the frequency. Therefore, a prediction error vector in the vicinity of (0, 0) having a high appearance frequency is encoded with a short code length, thereby improving the encoding efficiency as a whole.

For example, Japanese Patent Application No. Hei.
There is a technique disclosed in Japanese Patent Application No. 0-194612. According to this conventional technique, a prediction error vector is decomposed into a length component and a direction component, each component is subjected to variable-length coding, and transmitted to the decoding side. Specifically, as shown in FIG. 13, a trajectory obtained by connecting four points on the x-axis and the y-axis located at a distance r from the origin by a straight line with respect to a vector having a length component r (> 0) At the top, equally spaced points are defined. And
Starting from a point on the positive x-axis as a base point, numbers 0 to r-
1, r to 2r-1, 2r to 3r-1, and 3r to 4r-1 are sequentially numbered, and their numbers are used as direction components of the vector.

According to this example, the length component is determined such that the directional component takes a value in the range of [0, 4r-1] with respect to r, and the range of the directional component varies with the length component r. I do.
For this reason, a plurality of tables corresponding to length components are prepared as encoding tables for encoding direction components. Then, when encoding the directional component, the encoding table of the directional component is switched according to the length component r, thereby improving the encoding efficiency.

On the other hand, when decoding an encoded motion vector, a length component and a directional component of a prediction error vector, which has been subjected to variable length encoding, are extracted from an encoded data sequence and decoded, respectively. , (4), a motion vector of the block to be decoded is obtained by adding the prediction error vector to the motion vector of the block decoded immediately before.

[0010] _{x i = x i-1 +} x d (3) y i = y i-1 + y d (4) where, x _i, y _i is the component of the motion vector of the current block, x _{i -1} and y _i-1 are motion vector components of the block decoded immediately before.

[0011]

By the way, in the above-mentioned prior art, the expression range [0, 4r-
1] is four times the length component, so the size of the direction component encoding / decoding table is four times the length component. That is, in the related art, when the maximum expression range of the length component of the motion vector is enlarged, the encoding /
There is a problem that the size of the decoding table increases.
Specifically, in the above example, assuming that the maximum expression range of the length component is r _max , the size T _size of the encoding / decoding table of the direction component is obtained by the following equation (5).

[0012]

(Equation 1)

The present invention has been made in view of the above circumstances, and a motion vector encoding / decoding method capable of reducing the size of an encoding / decoding table used for encoding / decoding a motion vector, and the method. An object of the present invention is to provide a recording medium on which is recorded.

[0014]

In order to achieve the above object, the present invention has the following arrangement. That is, in the encoding method according to the first aspect of the present invention, in the motion vector encoding method for encoding a motion vector used for motion compensation inter-frame prediction of moving image encoding, a length component of the motion vector is calculated. A first step of encoding; and a second step of encoding a directional component of the motion vector, wherein the second step comprises: (a) the directional component of the motion vector corresponds to a coordinate position; Calculating a region to which the directional component of the motion vector belongs in a predetermined coordinate system that is divided into a plurality of regions in a given coordinate system, and obtaining region information representing the region; Calculating the coordinate position of the directional component in the set area to obtain position information representing the coordinate position; and (c) encoding the area information and the position information. And wherein the door.

According to the present invention, a space (coordinate system) divided in advance into a plurality of regions is defined. Then, when encoding the directional component, an area to which the directional component belongs is checked, and a code indicating the area is transmitted. Then, the position in each area is encoded and transmitted. Here, since the coordinate system that associates the directional component of the vector with the coordinate position is divided into a plurality of regions, when focusing on the length component of a certain vector, the coordinate position (direction Component expression range)
Decreases according to the number of divisions. Therefore, it is possible to reduce the size of the table for encoding the direction component in each area. For example, when the coordinate system is divided into four, the size T ′ _size of the encoding table is obtained from Expression (6).

[0016]

(Equation 2)

If the number of coordinate positions belonging to each area is the same, only the association between the coordinate position and the code is changed.
A table in a certain area can be used as a table in another area. That is, it is possible to generate an encoding table of information indicating a position in an area from a single table, and a plurality of encoding tables corresponding to each area are not required. In particular, when the number of divisions of the coordinate system is 4, the maximum value r _{max of} the length component and the range that the directional component can take are substantially equal, and the staple that is the basis for encoding the length component and the directional component is shared. It is possible to do.

According to a second aspect of the present invention, in the encoding method, when encoding the directional component, the encoding method is performed from a plurality of encoding tables prepared in advance according to the length component of the motion vector. A step of selecting an encoding table corresponding to a length component of a motion vector to be converted. According to this, the table size can be set according to the length component of the motion vector, and the directional component can be encoded using an encoding table corresponding to the length component of the vector.

According to a third aspect of the present invention, in the encoding method, a plurality of encoding tables for encoding the directional component according to a length component of a motion vector to be encoded are encoded into one encoding table. The method has a step of generating from a table. According to this, since a plurality of tables for encoding the directional components belonging to each area are generated from one table, it is possible to easily generate an encoding table for each area. However, the correspondence between the coordinate position representing the direction component and the code is corrected according to each area.

According to a fourth aspect of the present invention, the plurality of encoding tables are generated from the same encoding table as an encoding table for encoding the length component of the motion vector. It is characterized by the following. According to this, if the range that the length component can take is the same as the range that the direction component can take, the table size will be the same, and the coding for coding the length component of the motion vector will be described. An encoding table for each area is generated from the same encoding table as the table. However, since the length component is associated with the code in the length component encoding table, it is necessary to correct the directional component (coordinate position) so as to be associated with the code.

According to a fifth aspect of the present invention, there is provided an encoding method according to the first aspect, wherein all motion vectors having the same value as the length component of the motion vector to be encoded are associated with a predetermined motion vector expression range. Evaluating the containment relationship;
Redefining an encoding table for encoding the directional component of the motion vector according to the result of the evaluation,
It is characterized by having.

According to this, by evaluating the inclusion relationship between the length component of the motion vector to be coded and the predetermined motion vector expression range, the range of the length component and the vector expression range are evaluated. To understand the mismatch area. The code for the coordinate position existing in the non-coincidence area is regarded as a redundant code, and the code corresponding to this coordinate position is deleted from the encoding table. If the predetermined condition is satisfied, the code length is shortened and the coding table is generated again. As a result, the size of the encoding table is reduced, and the code length is reduced.

According to a sixth aspect of the present invention, in the motion vector decoding method for decoding a motion vector of a moving image, a first step of decoding a length component of the motion vector; A second step of decoding the directional component of the motion vector, wherein the second step comprises: (a) a coordinate system in which the directional component of the motion vector is associated with a coordinate position; Decoding region information representing a region to which the directional component of the motion vector belongs in the divided predetermined coordinate system;
(B) decoding position information representing a coordinate position of the directional component in the region; and (c) obtaining a directional component of the motion vector from the region information and the position information. It is characterized by.

According to the present invention, a space divided in advance into a plurality of regions is defined. Then, the directional component is decoded to obtain a code indicating the area to which the directional component belongs. Subsequently, the position in each area is decoded. Here, since the coordinate system that associates the directional component of the vector with the coordinate position is divided into a plurality of regions, when focusing on the length component of a certain vector, the coordinate position representing the directional component belonging to each region is determined by the number of divisions. Decrease accordingly. Therefore, it is possible to reduce the size of the table for decoding the direction component in each area.

If the number of coordinate positions belonging to each area is the same, only the association between the coordinate positions and the codes is changed.
A table in a certain area can be used as a table in another area. That is, a decoding table of information indicating a position in an area can be generated from one certain table, and a plurality of decoding tables corresponding to each area are not required. In particular, when the number of divisions of the coordinate system is 4, the maximum value r _{max of} the length component and the range that the directional component can take are substantially equal, and the staple that is the basis for decoding the length component and the directional component is shared. It becomes possible.

According to a seventh aspect of the present invention, in the decoding method, when decoding the directional component, a decoding target to be decoded is selected from a plurality of encoding tables prepared in advance according to a length component of a motion vector. The method further comprises the step of selecting a decoding table according to the length component of the motion vector. According to this, the table size can be set according to the length component of the motion vector, and the directional component can be decoded using the decoding table corresponding to the length component of the vector.

[0027] According to the decoding method of the present invention, a plurality of decoding tables for encoding the directional component in accordance with the length component of the motion vector to be decoded are generated from one decoding table. It is characterized by having a step. According to this, since a plurality of tables for decoding the directional components belonging to each area are generated from one table, a decoding table for each area can be easily generated. However, the correspondence between the coordinate position representing the direction component and the code is corrected according to each area.

A decoding method according to a ninth aspect of the invention is characterized in that the plurality of decoding tables are generated from the same decoding table as a decoding table for decoding the length component of the motion vector. . According to this, if the range that the length component can take and the range that the direction component can take are the same, the table size will be the same,
A decoding table for each area is generated from the same decoding table as the decoding table for decoding the length component of the motion vector. However, since the length component is associated with the code in the decoding table of the length component, it is necessary to correct the direction component (coordinate position) as being associated with the code.

In the decoding method according to the present invention, the inclusive relation between all motion vectors having the same value as the length component of the motion vector to be decoded and a predetermined motion vector expression range. And a step of redefining a decoding table for decoding the directional component of the motion vector according to the result of the evaluation.

According to this, the inclusion relation between the length component of the motion vector to be decoded and the predetermined motion vector expression range is evaluated, whereby the range of the length component and the vector expression range are compared. Understand the mismatch area. And
The code for the coordinate position existing in the mismatch area is regarded as a redundant code, and the code corresponding to this coordinate position is deleted from the decoding table. If the predetermined condition is satisfied, the code length is shortened and the decoding table is generated again. This reduces the size of the decoding table,
Moreover, the code length is shortened.

The recording medium according to the eleventh aspect of the present invention includes a step of encoding a length component of a motion vector, and a step of encoding a direction component of the motion vector in a coordinate system in which the component is associated with a coordinate position. Calculating a region to which the directional component of the motion vector belongs in a predetermined coordinate system divided into a plurality of regions to obtain region information representing the region; and calculating the coordinate position of the directional component in the calculated region. Calculating and obtaining position information representing the coordinate position, encoding the area information and the position information,
A program for executing the program is recorded.

According to a twelfth aspect of the present invention, there is provided a recording medium, comprising: a step of decoding a length component of a motion vector, wherein the direction component of the motion vector is a coordinate system associated with a coordinate position. Decoding area information indicating an area to which a directional component of the motion vector belongs in a predetermined coordinate system divided into areas, and decoding position information indicating a coordinate position of the directional component in the area. Obtaining a directional component of the motion vector from the area information and the position information.

[0033]

Embodiments of the present invention will be described below with reference to the drawings. In the present embodiment, a coding method for performing motion compensation on a block basis is assumed, a motion vector is predicted in the same manner as in Japanese Patent Application No. 10-194612, and a prediction error vector of the predicted motion vector is coded. Will be described.

Although the motion vector expresses the moving direction and the moving amount of the image, in this embodiment,
It is assumed that a prediction error vector is included in the motion vector. Therefore, in this embodiment, a case will be described in which a prediction error vector is encoded. However, the present invention can also be applied to a case in which an original motion vector expressing a moving direction and a moving amount of an image is coded. it can. In the following description, a motion vector (prediction error vector) is a two-dimensional vector.

The encoding method according to this embodiment will be described below with reference to the flowchart shown in FIG. Step S1: First, a motion vector of a block to be encoded is detected. Step S2: Next, the prediction error vector (x _d , y _d ) is calculated in the orthogonal coordinate system by the above-described equations (1) and (2).

Subsequently, the following processing (steps S3 to S3)
8), the calculated prediction error vector (x _d , y _d )
Is converted into a length component and a direction component for encoding. Step S3: The length component of the prediction error vector (x _d , y _d ) is calculated. In the present embodiment, the length component is calculated by combining a plurality of calculation formulas according to the length component by Expression (7).

[0037]

(Equation 3)

Where r is a length component and x _d , y
_d represents the horizontal and vertical components of the prediction error vector. Step S4: Next, the calculated length component is encoded using an encoding table to which short codes are assigned in order from 0 according to the length component.

Subsequently, the following processing (steps S5 to S5)
The direction component is encoded according to 8). First, an outline of this encoding will be described. As described below, the encoding method differs depending on the length component r. (A) No sign when r = 0. (B) When 0 <r ≦ 2, 2-bit fixed-length coding is used. (C) When r> 2, both 2-bit fixed length coding and variable length coding are used.

Here, a coordinate system used when encoding the direction component will be described with reference to FIG. This coordinate system is for associating a direction component with a coordinate position, and is assigned a coordinate position corresponding to the norm r obtained by the above equation (7). For example, when r = 0, the origin (0,0) of the coordinate system is associated with this direction component. Also, when r = 1, the coordinate position (1, 0),
Any of the four points (0, 1), (-1, 0), and (0, -1) is associated with each other, and four types of directions are expressed. Similarly, r = 1.4 (= √2) and r = 2 also represent four types of directions.

For example, when r = 3, the absolute value of the x component and the absolute value of the y component such as the coordinate positions (3,0), (2,1), (1,2) are obtained from Expression (7). Are assigned to coordinate positions whose sum is equal to 3. Similarly, when r = 4, the coordinate position where the sum of the absolute value of the x component and the absolute value of the y component is 4, such as the coordinate position (3, 1), (2, 2), (1, 3) Is assigned.

Here, as shown in FIG. 2, the coordinate system shown in FIG. 1 is radially divided into four regions A to D with its origin at the center, and a code is assigned to each region. In this embodiment, as shown in FIG. 4 described later, a code “01” is assigned to the area A, a code “11” is assigned to the area B, a code “00” is assigned to the area C, and a code “10” is assigned to the area D. Assign. In this case, for example, the symbol “01” indicates that the coordinate position belonging to the area A is between the left 45 ° direction and the right 45 ° direction with respect to the positive y-axis direction.

When the coordinate system is divided into four parts as described above, the size T _size ′ of the encoding table of the information indicating the position in each area can be obtained from the above equation (6). That is, 4
When divided, the table size is reduced to ４, and the table size is reduced. This table contains the length component r
Can be used in common in each area. Therefore, only one table needs to be prepared for each length component.

That is, by generating an encoding table of information indicating a position in an area from a single table, a plurality of encoding tables become unnecessary. In particular, as described above, when the number of divisions of the coordinate space is 4, the maximum value r _max of the length component is substantially equal to the range that the direction component can take. It can be shared for encoding components, and there is no increase in the number of tables due to division.

Next, the encoding of the direction component will be described. Step S5: It is determined whether or not the length component (norm) r of the prediction error vector obtained by the above equation (7) is zero. As a result of this determination, if r = 0, the encoding of the direction component ends, and the direction component is not encoded (that is, there is no sign). Step S6: If the length component r is not zero (Step S5: No), a table (direction component table) for encoding the direction component is generated as follows.

That is, when r ≦ 2 (r = 1, √2,
2), encoding of direction components is performed using the tables shown in FIGS. Here, FIG.
3 show the codes corresponding to the four coordinate positions when r = 1, and FIG. 3B shows the case where r = 1.4 shown in FIG.
3C show codes corresponding to the four coordinate positions when r = 2 shown in FIG. As shown in FIGS. 3A to 3C, each norm r
, The encoding tables have the same size, and a common code can be used. However, the coordinate positions corresponding to the codes are different.

When r> 2, the direction component represents a code (area information) indicating an area shown in FIG. 4 described later and a coordinate position in each area calculated by the following equation (8). The area information and the position information are encoded by the position information theta. At this time, as will be described later, the area information is encoded using a fixed length code, and the position information is variable length encoded.

[0048]

(Equation 4)

The table for encoding the direction component is generated according to the magnitude relation between the norm r and the motion vector expression range Range. Hereinafter, a description will be given of a method of generating a direction component encoding table when r ≦ Range. FIG. 4 shows the relationship between the area information code indicating the up, down, left, and right directions and the fixed length code in this case. Also, as the sign of the position information,
The sign of the length component is used. However, the following conditional expression (9)
If a) and (9b) hold, the sign calculated by equation (10) is used.

V [theta]. len ≠ V [| theta | -1]. len (9a) | theta | = r / 2 (9b)

[0051]

(Equation 5)

Note that V [i]. len represents the code length of the code corresponding to the symbol i in the length component coding table. Also, V [i]. code represents a code corresponding to the symbol i in the length component coding table. The code length at this time is V [| theta | -1]. Let it be len. Where V
[| Theta | -1]. If len = 1, the code length is “2”.

Here, equations (9a), (9b) and (10) will be described. Equation (9a) indicates that the code length of the position information (| theta |) is different from the code length of the position information (| theta | -1), and indicates that the position information theta has changed. Equation (9b) represents the case where the value of half the norm r is equal to the position information theta, and represents the coordinate position that gives the maximum value as the position information in each region. Specifically, a coordinate position (position information) existing in a region surrounded by a dotted line shown in FIG. 5 satisfies these expressions. When these expressions (9a) and (9b) are satisfied, the code (code) calculated by the expression (10) is used without using the encoding table as it is.

In the equation (10), when the position information theta> 0, the code length is V [| theta | -1]. len means the number of bits given, and when position information theta ≦ 0, the original code V [theta]. From the lower side of code, V [the
ta]. len-V [| theta | -1]. This means that the bit length given by len is truncated. In this case, the location information thet
The code length is the same as the code length when a> 0.

FIG. 6 shows an example of a specific table generation method.
This will be described with reference to FIG. When the original base table is as shown in FIG. 6, if the motion vector expression range Range is “5” (that is, the norm r =
5), as can be seen from FIG. 4, the coordinate position (3, -2), (4,-
There are five points: 1), (5, 0), (4, 1), and (3, 2).

Therefore, from the equation (8), the position information thet
The range of “a” is “−2 to +2”, and the size of the table illustrated in FIG. 7A is obtained. Here, by applying equation (10) to the base table shown in FIG. 6, the respective codes “101” and “111” for the position information theta = −2 and +2 shown in FIG. 7A are obtained. . This allows
Each code length of the code corresponding to the position information theta = -2, +2 is reduced by one bit.

Similarly, if the original table is as shown in FIG. 6, if the motion vector expression range Range is "9" (ie, norm r = 9), FIG.
The size of the table illustrated in (b) is obtained. in this case,
Each code length of the code corresponding to the position information theta = -4, +4 is reduced by 2 bits.

Next, a description will be given of a method of generating a direction component encoding table when r> Range. FIG. 8 shows the relationship between the area information code indicating the up, down, left, and right directions and the fixed length code in this case.
Shown in In the figure, the dotted line represents the motion vector expression range.
Boundary of expression range when Range is set to “9” (RANG
E = 9). In this example, for example, when r = 10,
The coordinate position (0, 10) (not shown) exceeds the motion vector expression range.

As described above, when r> Range, depending on the direction component, its coordinate position exceeds the expression range of the motion vector, so that the expression of the direction component beyond this expression range becomes unnecessary. Therefore, it is not necessary to prepare codes for portions beyond the motion vector expression range, and it is not necessary to prepare codes for position information of all directional components. Therefore, in this case, it is possible to reduce the size of the table by omitting, from the codes in the length component coding table, codes that exceed the motion vector expression range.

More specifically, in FIG.
In the area (code = 01) indicated by, when r = 13, the coordinate positions (−3, 10), (−2, 11), (−)
(1,12), (0,13), (1,12), (2,1)
1), (3, 10) are the range of motion vector expression Range
(= 9). In this case, the codes corresponding to r = −3 to +3 are omitted from the table shown in FIG.
The table shown in (c) is obtained.

At this time, the lower j bits of the code shown in FIG. 6 are cut out to obtain the code shown in FIG. 7C.
Here, j is a value determined by the norm r.
In the example shown in FIG. 7C, for the code corresponding to r = -4, +4, the lower three bits of the code corresponding to r = -4, +4 shown in FIG. 6 are cut out, and r = −5 to −4. 7, +5
For +7, the lower 4 bits are cut out. As described above, for the portion exceeding the motion vector expression range, the codes are omitted, the table size is reduced, and the code length of each code is shortened. As described above, the encoding table of the direction component is generated.

Here, the description returns to FIG. Step S7: As shown in FIG. 3D and FIG. 4, the area information is encoded using a 2-bit fixed length code representing the up, down, left, and right directions. Step S8: The position information theta in the area is variable-length coded using the table generated as described above.

Here, the encoding method of the directional component will be supplementarily described by taking the vector (5, 3) shown in FIG. 9 as an example. In the case of the vector in this example, the area information indicating the up, down, left, and right directions is encoded with a fixed length code “00”. Also, the location information theta
Becomes “3” from equation (8), and is subjected to variable-length encoding.
The variable length code of this position information is generated from the length component coding table based on the value of r. As described above, the directional component of the prediction error vector is encoded.

Hereinafter, the above-described encoding method (procedure) will be summarized. 1. A motion vector of the current block is detected (step S1). 2. A prediction error vector (x _d , y _d ) is calculated (step S2). 3. The length component r is calculated (step S3). 4. Variable length coding is performed on the length component r (step S4).

5. Check the length component r (step S
5) If the length component r is 0 (step S5: Ye)
s) Since there is no need to transmit the direction component, the encoding process ends. 6. Generate an encoding table for the direction component (step S
6). 7. Encode the direction component area information (step S
7). 8. The position information of the direction component is encoded (step S
8). The encoding method according to this embodiment has been described above.

Next, a decoding method according to this embodiment will be described with reference to FIG. On the decoding side, similarly to the encoding side, a decoding table of the position information of the directional component is generated. The contents of this table are the same as the encoding side,
Using this table, conversion from code to position information is performed as follows.

Step S11: The length component r is decoded from the code. Step S12: It is determined whether or not the length component r is zero.
Here, if r is zero, the process proceeds to step S17, where r
Is not zero (steps S12 to S12)
17) is executed.

Hereinafter, a method of decoding a motion vector will be described by taking as an example a case where the directional component of a prediction error vector is variable-length coded (r> 2). Step S13: First, a decoding table of the position information is generated. Step S14: Subsequently, the area information is decoded. Step S15: Subsequently, the position information is decoded.

Step S16: Next, the horizontal / vertical components of the vector are calculated. That is, | x |
And | y | are determined. The position information theta is x
Since the component and y component show the component whose absolute value is small,
The value of either component is known. Also, the length component r = |
Since z | + | y |, the absolute value of the other component is known. Finally, the sign is determined from the area information.

As a specific example, norm r = 8, area information =
A case will be described in which a code consisting of 00, position information theta = 3 is decoded to obtain a vector (5, 3). If the area information is "0
0 "indicates that | x |> | y |.
Therefore, from equation (8), the position information theta represents y, and y =
Get 3. At this time, | x | = 5 from r = | x | + | y |. In the area represented by the area information “00”, x> 0
Therefore, the sign of x becomes positive, and x = 5 is obtained. Thus, the prediction error vector (5, 3) is decoded.

Step S17: Next, a motion vector of the block to be decoded is calculated. That is, in other words, the motion vector (x _i, y _i ) of the block to be decoded is obtained by adding the prediction error vector (x _d, y _d ) to the motion vector (x _i−1, y _i−1 ) of the block decoded immediately before. Is decoded.

The procedure of the decoding method will be summarized below. 1. The length component r is decoded (step S11). 2. Check the length component r, and if the length component r is 0,
It is assumed that x _d = 0 and y _d = 0, and the process proceeds to step S17 described later (step S12). 3. An encoding table of the direction component (position information) is generated (step S13). 4. The area information of the direction component is decoded (step S1).
4). 5. The position information of the direction component is decoded (step S1).
5). 6. The horizontal / vertical components of the prediction error vector are calculated (step S16). That is, a prediction error vector is calculated from the length information r and the area information and the position information of the direction component. 7. The motion vector of the decoding target block is calculated (step S17).

Although the embodiment of the present invention has been described above, the present invention is not limited to this embodiment, and the present invention is applicable to any design change or the like without departing from the gist of the present invention. included. For example, in the above-described embodiment, the coordinate system for associating the direction component with the coordinate position is 4
Although the area is divided into three areas, the number of divisions may be set as appropriate without being limited to this.

Although the smaller of the x component and the y component is used as the position information according to the equation (8), the present invention is not limited to this, and the larger one may be used. Further, when the conditional expressions represented by the expressions (9a) and (9b) are satisfied, a table for encoding the direction component is generated, and the code is redefined. May be changed.

In the above-described embodiment, a case has been described as an example where a two-dimensional vector is encoded. However, the present invention can be extended to a case where a three-dimensional vector or more is encoded. Furthermore, the encoding table of the directional component is generated according to the magnitude relationship between the expression range Range of the motion vector and the norm r. Needless to say, the table used for encoding the length component can be used as it is. Good.

Further, the above-mentioned encoding / decoding method is executed on a computer such as a workstation, for example, and a program describing this encoding / decoding method is recorded on a computer-readable recording medium. You may leave.

[0077]

As described above, according to the present invention, the following effects can be obtained. That is, according to the first aspect of the present invention, a coordinate area to which a directional component of a motion vector belongs is calculated, area information representing the area is calculated, and a coordinate position in the calculated area is calculated. Is calculated, and the region information and the position information are encoded, so that the size of an encoding table used for encoding a motion vector can be reduced.

According to the second aspect of the present invention, when the direction component is encoded, the encoding table corresponding to the length component of the motion vector to be encoded is selected. The direction component can be encoded using a table corresponding to the length component.

According to the third aspect of the present invention, 1
Since the encoding table corresponding to the length component of the motion vector to be encoded is generated from the two encoding tables, the encoding table for each area can be easily generated, and the length of each area can be easily calculated. Encoding can be performed using a table of a size corresponding to the component.

According to the fourth aspect of the present invention, a table for encoding the directional component is generated from the same encoding table as the encoding table for encoding the length component of the motion vector. Therefore, the types of encoding tables can be reduced.

According to the fifth aspect of the present invention, the inclusion relationship between the length component of the motion vector to be encoded and the expression range of the motion vector is evaluated, and the direction component is encoded according to the evaluation result. Since the encoding table for encoding is redefined, the size of the encoding table can be further reduced.

According to the invention of claim 6, the length component of the motion vector is decoded, the coordinate area to which the direction component of the motion vector belongs and the position information are decoded, and the direction component is decoded. Therefore, it is possible to reduce the size of the decoding table used for decoding the motion vector.

According to the seventh aspect of the present invention, when the direction component is decoded, a decoding table corresponding to the length component of the motion vector to be decoded is selected.
The direction component can be decoded using a table corresponding to the length component of the vector.

According to the invention of claim 8, 1
Since the decoding table corresponding to the length component of the motion vector to be decoded is generated from the two decoding tables, the decoding table for each region can be easily generated, and the decoding table corresponding to the length component for each region can be generated. Decoding can be performed using the size table.

According to the ninth aspect of the present invention, a table for decoding a directional component is generated from the same decoding table as a decoding table for decoding a length component of a motion vector. The number of types of decoding tables can be reduced.

According to the tenth aspect of the present invention,
Since the inclusion relationship between the length component of the motion vector to be decoded and the expression range of the motion vector is evaluated, and the decoding table for decoding the directional component is redefined according to the result of the evaluation, the decoding table Can be further reduced.

According to the eleventh aspect of the present invention,
Encoding the length component of the motion vector, obtaining area information indicating an area to which the directional component of the motion vector belongs, obtaining position information indicating a coordinate position of the directional component, And a step of encoding the position information.

According to the twelfth aspect of the present invention,
Decoding a length component of the motion vector, decoding area information indicating an area to which the direction component of the motion vector belongs, decoding position information indicating a coordinate position of the direction component, Obtaining a directional component of a motion vector from the position information.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining a coordinate system expressing directional components in a motion vector encoding / decoding method according to an embodiment of the present invention.

FIG. 2 is a diagram for explaining a divided region of a coordinate system expressing a directional component in the motion vector encoding / decoding method according to the embodiment of the present invention;

FIG. 3 is a diagram for explaining a table for encoding direction components of a motion vector according to the embodiment of the present invention;

FIG. 4 is a diagram illustrating a relationship (r) between a length component r of a motion vector and a vector expression range Range according to the embodiment of the present invention;
≦ Range).

FIG. 5 is a diagram for explaining a method of generating a direction component encoding table according to the embodiment of the present invention;

FIG. 6 is a diagram showing an example of a direction component encoding table (base table) according to the embodiment of the present invention;

FIG. 7 is a diagram showing a direction component encoding table according to the embodiment of the present invention;

FIG. 8 shows a relationship (r) between a length component r of a motion vector and a vector expression range Range according to the embodiment of the present invention.
> Range).

FIG. 9 is a diagram for describing a specific example of encoding according to the embodiment of the present invention.

FIG. 10 is a flowchart illustrating an encoding method according to the embodiment of the present invention.

FIG. 11 is a flowchart illustrating a decoding method according to the embodiment of the present invention.

FIG. 12 is a diagram showing a distribution of a prediction error vector.

FIG. 13 is a diagram for explaining an encoding method according to the related art.

[Explanation of symbols]

A to D: divided region r: length component (norm) of motion vector (prediction error vector) Range: motion vector expression range S1 to S8, S11 to S17 ... step

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor Satoshi Hayama 3-19-2 Nishi-Shinjuku, Shinjuku-ku, Tokyo Nippon Telegraph and Telephone Corporation F-term (reference) 5C059 KK08 KK19 MA00 MA05 ME01 ME13 NN03 NN31 RC16 SS20 SS26 TA21 TA47 TC12 TD16 UA02 UA05

Claims (12)

[Claims] 1. A motion vector coding method for coding a motion vector used for motion compensation inter-frame prediction in video coding, wherein: a first step of coding a length component of the motion vector; A second step of encoding a directional component of the motion vector, wherein: (a) a coordinate system in which the directional component of the motion vector is associated with a coordinate position; Calculating a region to which the directional component of the motion vector belongs in a predetermined coordinate system divided into a region and obtaining region information representing the region; and (b) calculating a coordinate position of the directional component in the calculated region. A motion vector encoding method, comprising: calculating and obtaining position information representing the coordinate position; and (c) encoding the area information and the position information. 2. A method according to claim 1, wherein when the direction component is encoded, a code corresponding to the length component of the motion vector to be encoded is selected from a plurality of encoding tables prepared in advance according to the length component of the motion vector. 2. The method according to claim 1, further comprising the step of selecting a coding table. 3. The method according to claim 1, further comprising the step of generating a plurality of encoding tables for encoding the directional component from one encoding table in accordance with a length component of a motion vector to be encoded. Item 6. The motion vector encoding method according to Item 1. 4. The motion according to claim 3, wherein the plurality of encoding tables are generated from the same encoding table as an encoding table for encoding a length component of the motion vector. Vector encoding method. 5. Evaluating the inclusion relationship between all motion vectors having the same value as the length component of the motion vector to be coded and a predetermined motion vector expression range; The motion vector encoding method according to claim 4, further comprising: redefining an encoding table for encoding a direction component of the motion vector in response to the encoding table. 6. A motion vector decoding method for decoding a motion vector of a moving image, wherein: a first step of decoding a length component of the motion vector; a second step of decoding a directional component of the motion vector; Wherein the second step comprises the steps of: Decoding area information indicating an area to which a directional component belongs; (b) decoding position information indicating a coordinate position of the directional component within the area; and (c) decoding the area information and the position information. Obtaining a directional component of the motion vector from the following. 7. When decoding the directional component, a decoding table corresponding to the length component of the motion vector to be decoded is selected from a plurality of encoding tables prepared in advance according to the length component of the motion vector. 7. The method according to claim 6, further comprising the step of selecting. 8. The method according to claim 6, further comprising the step of generating a plurality of decoding tables for encoding the direction components from one decoding table in accordance with a length component of a motion vector to be decoded. The described motion vector decoding method. 9. The method according to claim 8, wherein the plurality of decoding tables are generated from the same decoding table as a decoding table for decoding the length component of the motion vector. 10. A step of evaluating the inclusion relationship between all motion vectors having the same value as the length component of the motion vector to be decoded, and a predetermined motion vector expression range, according to a result of the evaluation. And redefining a decoding table for decoding a directional component of the motion vector by using the method. 10. The method according to claim 9, further comprising: 11. A step of encoding a length component of a motion vector, wherein the direction component of the motion vector is a coordinate system associated with a coordinate position, and is encoded in a predetermined coordinate system divided into a plurality of regions in advance. Calculating a region to which the directional component of the motion vector belongs to obtain region information representing the region; and calculating a coordinate position of the directional component in the calculated region to obtain position information representing the coordinate position. And a step of encoding the area information and the position information. A computer-readable recording medium recording a program for executing the steps. 12. A step of decoding a length component of the motion vector, wherein the directional component of the motion vector is a coordinate system associated with a coordinate position, and the directional component of the motion vector is divided into a plurality of regions in a predetermined coordinate system. Decoding area information indicating an area to which a directional component of a motion vector belongs; decoding position information indicating a coordinate position of the directional component in the area; and determining the motion from the area information and the position information. A step of obtaining a directional component of a vector; and a computer-readable recording medium recording a program for executing the steps.