KR100978465B1 - Bi-prediction coding method and apparatus, bi-prediction decoding method and apparatus, and recording midium - Google Patents

Abstract

The present invention relates to a bi-prediction coding method and apparatus, a bi-prediction decoding method and apparatus and a recording medium. According to the present invention, it is possible to solve the problem of the implementation of the two-prediction encoding of image compression and to transmit the motion vector more efficiently by using the fact that the motion occurs linearly. -Prediction coding method and apparatus, Bi-prediction decoding method and apparatus and recording medium are provided.

Description

Bi-PREDICTION CODING METHOD AND APPARATUS, BI-PREDICTION DECODING METHOD AND APPARATUS, AND RECORDING MIDIUM}

The present invention relates to a bi-prediction encoding method and apparatus, a bi-prediction decoding method and apparatus, and a recording medium, and more particularly, to reduce time-base correlation in video compression. Bi-prediction coding method and apparatus, Bi-prediction decoding method and apparatus and a recording medium.

According to the MPEG-1, MPEG-2 and MPEG-4 and ITU-T H.26x standards set by ISO / IEC JTC1 as the international standard for image compression, the past picture in coding the current picture. P-Picture coding method referring to (Picture) and B-Picture coding method referring to both past and future pictures are adopted and motion estimation is based on this. Encoding is done.

In addition, instead of encoding the motion vector of the current block as it is, to enhance the coding efficiency of the motion vector, prediction encoding of the motion vector is performed using the motion vectors of neighboring blocks to reflect the correlation with the motion vectors of neighboring blocks. have.

Therefore, in order to improve the coding efficiency, the accuracy of the motion vector and the minimization of the prediction error are important, but the compression efficiency of the motion vector data must be considered.

In order to minimize the motion prediction error, the bidirectional predictive coding method uses joint motion estimation to determine the optimal forward motion vector and backward motion vector pair. The accuracy of the motion vector is maximized by finding a very accurate motion vector pair through. However, such a bidirectional predictive encoding method requires a very high complexity and has a lot of difficulties in its implementation.

In order to avoid such implementation difficulties, in general, the bidirectional predictive coding method independently searches for a forward motion vector and a backward motion vector, and then searches for the optimal forward motion vector and the optimal backward motion vector. By using the two-way prediction encoding.

In general, however, the motion occurs linearly in many cases of the image. Even in this case, the transmission of the forward motion vector and the backward motion vector does not take full advantage of the fact that the motion is linear. By transmitting both word motion vectors, a large number of encoded bits of the motion vector may be generated, thereby degrading encoding performance.

Referring to the bidirectional predictive encoding related to conventional image compression in more detail, the conventional bidirectional predictive encoding uses the following equations (1) and (2) to calculate a predetermined cost value, and then doubles the minimum. A forward motion vector and a backward motion vector having a prediction cost of are selected as the bidirectional prediction vectors, and the optimal bidirectional prediction reference block is determined using the forward motion vectors.

[Equation 1]

[Equation 2]

는 비용(Cost)함수(Cost function)로 통상적으로 현재 블록과 예측 블록의 차이의 절대치 합(SAD)이나 비트율-왜곡 최적화 방법(Rate-Distortion Optimization) 등으로 결정된다.Where [Equation 1]

Cost is a cost function, which is typically determined by the absolute sum (SAD) of the difference between the current block and the predictive block, or by a rate-distortion optimization method.

는 현재 부호화할 블록,

는 [수학식 2]와 같이 포워드 참조픽춰 블록과 백워드 참조픽춰 블록의 가중합으로 만든 예측 블록이다. 그리고,

와

는 각각 포워드 움직임 벡터 및 백워드 움직임 벡터값이고,

와

는 포워드 탐색 크기와 백워드 탐색 크기이다.Also,

Is the block to be encoded currently,

Is a prediction block made by weighted sum of the forward reference picture block and the backward reference picture block as shown in [Equation 2]. And,

Wow

Are forward motion vector and backward motion vector, respectively,

Wow

Is the forward search size and the backward search size.

는 양방향 예측 부호화를 위한 최종 예측값이 된다. 여기서 양방향 예측 부호화를 위한 최종적인 포워드 움직임 벡터와 백워드 움직임 벡터를 찾기 위해서는 조인트 추정(Joint Estimation)을 수행하는데, 조인트 추정의 경우 움직임 탐색이

×

번 필요하여 매우 많은 연산량 및 참조픽춰 메모리 접근이 요구되므로, 실제 영상 압축 시스템에서 사용하기에는 현실적 문제가 많다.In Equation 1

Is the final prediction value for bidirectional predictive encoding. In this case, joint estimation is performed to find the final forward motion vector and the backward motion vector for bidirectional predictive encoding.

×

Since it requires twice, and requires a large amount of computation and access to reference picture memory, there are many practical problems to use in an actual image compression system.

Due to this problem, in general, bidirectional predictive coding independently searches for a forward motion vector and a backward motion vector as shown in Equation 3, and then forwards the result found in Equation 3 as shown in Equation 4. Bidirectional prediction is performed using the motion vector and the backward motion vector.

&Quot; (3) "

&Quot; (4) "

와

는 각각 포워드 참조픽춰의 움직임 벡터 위치의 블록 및 백워드 참조픽춰의 움직임 벡터 위치의 블록을 의미하고,

와

는 각각 포워드 참조픽춰에서 찾은 최종적인 움직임 벡터 위치의 블록 및 백워드 참조픽춰에서 찾은 최종적인 움직임 벡터 위치의 블록이다.In [Equation 3]

Wow

Means a block of the motion vector position of the forward reference picture and a block of the motion vector position of the backward reference picture, respectively.

Wow

Are the blocks of the final motion vector positions found in the forward reference picture and the blocks of the final motion vector positions found in the backward reference picture, respectively.

는 [수학식 3]에서 독립적으로 찾은 최적의 포워드 움직임 벡터와 백워드 움직임 벡터를 이용한 최적의 양방향 움직임 보상 블록을 나타낸다.Also, in [Equation 4]

Denotes an optimal bidirectional motion compensation block using an optimal forward motion vector and a backward motion vector found independently in Equation (3).

+

로 매우 낮은 복잡도가 요구되지만 대신 양방향 예측성능이 떨어지며 따라서 움직임 벡터 부호화시 비트 발생량이 많아진다.However, when the bidirectional motion estimation is independently performed as described above, the number of searches required is

+

However, very low complexity is required, but bidirectional prediction performance is lowered, resulting in higher bit generation in motion vector coding.

The conventional technique developed to solve the above problem is a symmetric mode in bidirectional prediction adopted by the Chinese Advanced Video System (AVS) standard. Symmetric mode transmits only the forward motion vector, and the backward motion vector is calculated by a predetermined equation at the decoder to obtain the forward motion vector and the backward motion vector required for the bidirectional prediction, and then use the bidirectional prediction. Obtain a reference block.

In this symmetric mode, since the backward motion vector is calculated and used symmetrically with the forward motion vector, only the forward motion vector needs to be transmitted. Therefore, the amount of bits for motion vector coding can be reduced.

If it is advantageous to transmit the forward motion vector and calculate the backward motion vector therefrom, the prior art reduces the amount of bits of the motion vector required for encoding, and effectively searches for forward and backward joint motion. In this way, it is possible to provide a bidirectional prediction method with little increase in the complexity of searching for a matching block of motion search.

However, in some cases it may be more advantageous to transmit a backward motion vector and calculate a forward motion vector therefrom, in which case the prior art also transmits a forward motion vector from which backward motion is derived. Since we have no choice but to calculate the vector, we have to accept inefficiency.

In addition, according to the bidirectional symmetric mode coding method according to the prior art, one of the reference blocks used for motion prediction of the current block is taken from the forward reference picture and the other from the backward reference picture. The time reference is before the current picture to which the current block belongs, and the backward reference picture is applied only when it is later in time. However, when the current block is motion predicted, it is often more advantageous for motion prediction that both the forward and backward reference pictures are temporally or later than the current picture to which the current block belongs, depending on the nature of the image. Even in this case, the motion prediction according to the prior art has a problem in that the compression efficiency is reduced by applying only reference pictures positioned before and after each time.

Therefore, the first object of the present invention is to solve the complexity problem of the implementation of the bi-prediction coding of image compression and to transmit the motion vector more efficiently by using the fact that the motion occurs linearly. Improves the encoding performance degradation problem caused by transmitting only the forward motion vector by the symmetric mode coding method, which is one of the conventional bidirectional predictive coding methods related to the compression of the video. Bi-prediction encoding method and apparatus that can perform more efficient encoding by facilitating joint estimation when performing actual bi-prediction encoding while reducing the amount of bits required for encoding. To provide.

In addition, a second object of the present invention is a bi-prediction decoding method capable of more efficient decoding in decoding data encoded according to the above-described bi-prediction encoding method and apparatus. And to provide an apparatus.

The first object is a bi-prediction encoding method using a plurality of reference pictures, comprising: (a) selecting a first selection motion vector from a first reference picture for a current block to be encoded; (b) calculating a first calculated motion vector for a second reference picture based on the first selected motion vector; (c) based on the first selection motion vector, the first calculation motion vector, a first selection motion prediction block corresponding to the first selection motion vector, and a second calculation motion prediction block corresponding to the first calculation motion vector. Calculating a first predictive encoding cost; (d) Repeating steps (a), (b) and (c), the first representative selection motion vector, the first representative calculation motion vector, and the first representative selection motion satisfying a predetermined criterion Selecting a first representative prediction coding cost according to a vector and the first representative calculation motion vector; (e) selecting a second selection motion vector from the second reference picture; (f) calculating a second calculated motion vector for the first reference picture based on the second selected motion vector; (g) based on the second selected motion vector, the second calculated motion vector, a second selected motion prediction block corresponding to the second selected motion vector, and a second calculated motion prediction block corresponding to the second calculated motion vector. Calculating a second predictive encoding cost; (h) repeating steps (e), (f), and (g) to perform a second representative selection motion vector, a second representative calculation motion vector, and the second representative selection motion that satisfy a predetermined criterion; Selecting a second representative prediction coding cost according to a vector and the second representative calculating motion vector; (i) if the first representative prediction coding cost is less than the second representative prediction coding cost, the first representative selection motion vector is selected as the encoding target motion vector, and the first representative calculation motion vector is selected as the uncoding target motion vector. And when the second representative prediction coding cost is less than the first representative prediction coding cost, selecting the second representative selection motion vector as the encoding target motion vector and the second representative calculating motion vector as the uncoding target motion vector. Steps; (j) encoding the motion vector to be encoded is achieved by a bi-prediction encoding method.

Here, the step (j) is bi-prediction encoding mode information for transmitting that the encoding is a bi-prediction encoding mode encoded by the bi-prediction encoding method. It may include the step of encoding.

In the step (b), the first calculated motion vector is obtained by multiplying a relative time distance between the current picture to which the current block belongs, the first reference picture and the second reference picture, and the first selected motion vector. Calculated; In the step (f), the second calculated motion vector may be calculated by multiplying the relative time distance between the current picture, the first reference picture and the second reference picture by the second selected motion vector.

(여기서, mv_cal는 상기 제1 산출 움직임 벡터 또는 상기 제2 산출 움직임 벡터이고, mv_sel는 상기 제1 선택 움직임 벡터 또는 상기 제2 선택 움직임 벡터이고, TD_B는 상기 제1 참조픽춰와 상기 제2 참조픽춰 중 상기 제1 선택 움직임 벡터 또는 상기 제2 선택 움직임 벡터가 선택된 어느 하나와 상기 현재픽춰 간의 시간상의 거리이고, TD_C는 상기 제1 참조픽춰 및 상기 제2 참조픽춰 중 상기 제1 산출 움직임 벡터 또는 상기 제2 산출 움직임 벡터가 산출된 어느 하나와 상기 현재픽춰 간의 시간상의 거리이다)에 의해 산출될 수 있다.Here, when the first reference picture and the second reference picture are located before and after each other in time with respect to the current picture, the first calculated motion vector and the first step in steps (b) and (f), respectively. 2 output motion vectors, respectively,

Where mv _cal is the first calculated motion vector or the second calculated motion vector, mv _sel is the first selected motion vector or the second selected motion vector, and TD _B is the first reference picture and the first motion vector. 2 is a temporal distance between any one of the first selected motion vector or the second selected motion vector among the reference pictures and the current picture, and TD _C is the first calculation of the first reference picture and the second reference picture; Motion distance or the second calculated motion vector is a distance in time between the calculated one and the current picture).

(여기서, mv_cal는 상기 제1 산출 움직임 벡터 또는 상기 제2 산출 움직임 벡터이고, mv_sel는 상기 제1 선택 움직임 벡터 또는 상기 제2 선택 움직임 벡터이고, TD_B는 상기 제1 참조픽춰와 상기 제2 참조픽춰 중 상기 제1 선택 움직임 벡터 또는 상기 제2 선택 움직임 벡터가 선택된 어느 하나와 상기 현재픽춰 간의 시간상의 거리이고, TD_C는 상기 제1 참조픽춰 및 상기 제2 참조픽춰 중 상기 제1 산출 움직임 벡터 또는 상기 제2 산출 움직임 벡터가 산출된 어느 하나와 상기 현재픽춰 간의 시간상의 거리이다)에 의해 산출될 수 있다.Further, when the first reference picture and the second reference picture are located together before or after in time from the current picture, the first calculated motion vector and the second in the steps (b) and (f). The output motion vectors are each

Where mv _cal is the first calculated motion vector or the second calculated motion vector, mv _sel is the first selected motion vector or the second selected motion vector, and TD _B is the first reference picture and the first motion vector. 2 is a temporal distance between any one of the first selected motion vector or the second selected motion vector among the reference pictures and the current picture, and TD _C is the first calculation of the first reference picture and the second reference picture; Motion distance or the second calculated motion vector is a distance in time between the calculated one and the current picture).

In addition, the first object of the present invention is a bi-prediction encoding apparatus that predictively encodes a plurality of reference pictures, wherein the first object is within a predetermined motion search range based on a current block to be encoded from the plurality of first reference pictures. A first motion vector selector which selects a first selected motion vector corresponding to each of the first reference pictures; A first motion vector calculator configured to calculate a first calculated motion vector corresponding to the first selected motion vector based on the first selected motion vector and a second reference picture corresponding to the first selected motion vector; ; The angle based on the first selection motion vector, the first calculation motion vector, a first selection motion prediction block corresponding to the first selection motion vector, and a first calculation motion prediction block corresponding to the first calculation motion vector. A first encoding cost calculator for calculating a first prediction encoding cost respectively corresponding to the first selected motion vector; A first encoding cost selecting unit configured to compare the plurality of first predictive encoding costs calculated corresponding to each of the first selected motion vectors, and select a first representative prediction encoding cost that satisfies a preset condition; A second motion vector selector which selects a second selection motion vector corresponding to each second reference picture from each second reference picture; A second motion vector calculator configured to calculate a second calculated motion vector corresponding to the second selected motion vector based on the second selected motion vector and the first reference picture corresponding to the second selected motion vector; Wow; The angle based on the second selection motion vector, the second calculation motion vector, the second selection motion prediction block corresponding to the second selection motion vector, and the second calculation motion prediction block corresponding to the second calculation motion vector. A second encoding cost calculator for calculating second predictive encoding costs respectively corresponding to the second selected motion vector; A second encoding cost selecting unit configured to compare the plurality of second prediction encoding costs calculated corresponding to the second selection motion vectors, and select a second representative prediction encoding cost that satisfies a preset condition; When the first representative prediction coding cost is smaller than the second representative prediction coding cost, any one of the first selected motion vectors corresponding to the first representative prediction coding cost is selected as an encoding target motion vector, and the second representative A motion vector selecting unit that selects one of the second selected motion vectors corresponding to the second representative prediction coding cost as an encoding target motion vector when a prediction coding cost is smaller than the first representative prediction coding cost; It may also be achieved by a bi-prediction encoding apparatus comprising a motion vector encoder for encoding the motion vector to be encoded.

On the other hand, the second object is a bi-prediction decoding method for decoding a bi-prediction coded data using a plurality of reference pictures, (a) analyzing the encoded data Determining whether the current block to be decoded is in a bi-prediction encoding mode; (b) if it is determined that the current block is in the bi-prediction encoding mode, reconstructing a decoding target motion vector by decoding the encoded data; (c) a temporal distance between the reconstructed decoding object motion vector, the current picture to which the current block belongs, and a first decoding reference picture corresponding to the decoding object motion vector, and a time view between the current picture and the second decoding reference picture. Calculating an undecoded target motion vector corresponding to the second reference picture based on a distance of a; (d) generating a prediction block for the current block based on the reconstructed decoding object motion vector and the calculated decoded object motion vector, and reconstructing the current block based on the generated prediction block. It is achieved by a Bi-prediction decoding method characterized in that.

Here, in step (c), the decoded object motion vector may be calculated by a product of a relative time distance between the current picture, the first reference picture and the second reference picture, and the product of the decoding object motion vector. .

(여기서, mv_cal는 상기 비복호화 대상 움직임 벡터이고, mv_sel는 상기 복호화 대상 움직임 벡터이고, TD_B는 상기 제1 참조픽춰와 상기 현재픽춰 간의 시간상의 거리이고, TD_C는 상기 제2 참조픽춰와 상기 현재픽춰 간의 시간상의 거리이다)에 의해 산출될 수 있다.When the first reference picture and the second reference picture are positioned before and after each other in time with respect to the current picture, in step (c), the undecoded target motion vector is expressed by the following equation.

Where mv _cal is the undecoded object motion vector, mv _sel is the decoding object motion vector, TD _B is the distance in time between the first reference picture and the current picture, and TD _C is the second reference picture. And a distance in time between the current picture and the current picture).

(여기서, mv_cal는 상기 비복호화 대상 움직임 벡터이고, mv_sel는 상기 복호화 대상 움직임 벡터이고, TD_B는 상기 제1 참조픽춰와 상기 현재픽춰 간의 시간상의 거리이고, TD_C는 상기 제2 참조픽춰와 상기 현재픽춰 간의 시간상의 거리이다)에 의해 산출될 수 있다.Here, when the first reference picture and the second reference picture are located together before or after in time from the current picture, in step (c), the undecoded target motion vector is expressed by the following equation.

Where mv _cal is the undecoded object motion vector, mv _sel is the decoding object motion vector, TD _B is the distance in time between the first reference picture and the current picture, and TD _C is the second reference picture. And a distance in time between the current picture and the current picture).

In addition, the second object of the present invention is a bi-prediction decoding apparatus that decodes bi-prediction coded data using a plurality of reference pictures, the entropy of entropy decoding the coded data. A decoder; A decoding controller which analyzes the decoded data to determine whether a current block to be decoded is encoded by a bi-prediction encoding mode; A decoding object motion vector reconstructing unit which reconstructs a decoding object motion vector from the decoded data when it is determined by the decoding control unit as the bi-prediction encoding mode; Based on the reconstructed decoding object motion vector, the temporal distance between the current picture to which the current block belongs and the first reference picture corresponding to the decoding object motion vector, and the distance in time between the current picture and the second reference picture. An undecoded target motion vector calculator for calculating an undecoded target motion vector corresponding to the second reference picture; A motion compensator for generating at least one prediction block based on the reconstructed decoding target motion vector and the calculated undecoded target motion vector; It may also be achieved by a bi-prediction decoding apparatus comprising a current block reconstruction unit for reconstructing the current block based on the decoded data and the prediction block.

According to the present invention, the first calculated motion vector for the second reference picture is calculated on the basis of the first selected motion vector selected from the first reference picture in selecting the encoding target motion vector and the uncoding target motion vector for the current block. In addition, the second calculated motion vector for the first reference picture is calculated based on the second selected motion vector selected from the second reference picture, the encoding efficiency is selected and encoded, and the encoding efficiency of the transmitted motion vector is further increased. This can improve the coding efficiency of prediction errors.

In addition, a problem of deterioration in encoding performance caused by transmitting only a forward motion vector by a symmetric mode encoding method, which is one of the conventional bidirectional predictive encoding methods related to video compression, is improved. It is possible to perform more efficient encoding by facilitating joint estimation when performing actual bi-prediction coding while reducing the amount of bits required.

1 and 2 are diagrams for explaining a bi-prediction encoding method according to the present invention;
3 to 5 illustrate examples of a temporal relationship between a current picture, a first reference picture, and a second reference picture in a bi-prediction encoding method according to the present invention;
6 is a diagram illustrating a configuration of a bi-prediction encoding apparatus according to the present invention.
FIG. 7 is a diagram illustrating an example of a configuration of a motion prediction unit of the bi-prediction encoding apparatus of FIG. 6.
8 is a diagram for explaining a bi-prediction decoding method according to the present invention;
9 is a diagram illustrating a configuration of a bi-prediction decoding apparatus according to the present invention.
FIG. 10 is a diagram illustrating an example of a configuration of a motion vector decoder of the bi-prediction decoding apparatus of FIG. 9.

Hereinafter, with reference to the accompanying drawings will be described in detail the present invention.

1 and 2 are flowcharts of a bi-prediction encoding method according to the present invention. Referring to FIGS. 1 and 2, when an N × M block (hereinafter, referred to as a “current block”) to be encoded is input (S10), the first reference picture is based on the current picture to which the current block belongs. And a second reference picture are determined. Herein, the size of the block N × M may include a case in which N is the same as or different from M.

Then, a motion vector (hereinafter, referred to as a 'first selected motion vector') is selected within a preset motion search range from the first reference picture (S11). Then, a motion vector (hereinafter, referred to as a 'first calculated motion vector') for a second reference picture corresponding to the first reference picture is calculated based on the first selected motion vector (S12). Here, the first selection motion vector selected from the first reference picture is a motion vector that is actually encoded and transmitted when it is selected as an encoding target motion vector in a later-described process, and the first calculated motion vector is based on the transmitted encoding target motion vector. In the decryption process As a motion vector to be calculated, a motion vector that is not encoded in the encoding process, that is, a motion vector that is not transmitted.

Then, a first selected motion prediction block for the first selected motion vector is generated, and a first calculated motion prediction block for the first calculated motion vector is generated (S13). A predictive coding cost (hereinafter, referred to as a 'first predictive coding cost') is calculated based on the first selected motion vector, the first calculated motion vector, the first selected motion prediction block, and the first calculated motion prediction block. S14).

Thereafter, steps S14 to S14 are repeatedly performed on the selected first reference picture and the second reference pictures to calculate a plurality of first prediction encoding costs. When the calculation of the first predictive encoding cost for the first reference picture and the second reference pictures is finished (S15), the first predictive encoding cost that satisfies a predetermined criterion among the plurality of calculated first predictive encoding costs is determined. Select. Here, as an example, the selection criterion of the first prediction encoding cost is provided so that any one having the lowest encoding cost is selected.

The first selected prediction vector, the first selected motion vector and the first calculated motion vector used for calculating the selected first prediction encoding cost, the first representative prediction coding cost, the first representative selection motion vector, and One representative motion vector is selected (S16).

Meanwhile, in response to the above process, a motion vector (hereinafter referred to as a 'second selection motion vector') is selected from the second reference picture (S17), and the first reference picture is based on the selected second selection motion vector. A motion vector (hereinafter, referred to as a 'second calculated motion vector') is calculated (S18).

Then, a second selective motion prediction block for the second selective motion vector is generated, and a second calculated motion prediction block for the second calculated motion vector is generated (S19). A prediction encoding cost (hereinafter, referred to as a second prediction encoding cost) is calculated based on the second selection motion vector, the second calculation motion vector, the second selection motion prediction block, and the second calculation motion prediction block ( S20).

In addition, a plurality of second prediction encoding costs are calculated by repeating steps S20 through S20 for the first reference picture and the second reference pictures. Then, when the calculation of the second predictive encoding cost is finished (S21), a second predictive encoding cost is selected among the calculated plurality of second predictive encoding costs that satisfy a predetermined criterion. Here, as an example, the selection criterion of the second prediction encoding cost is provided so that any one having the lowest encoding cost is selected.

The second selected prediction vector, the second selected motion vector and the second calculated motion vector used for calculating the selected second prediction encoding cost, the second representative prediction motion vector, and the second representative selection motion vector, respectively. 2 is selected as the representative output motion vector (S22).

When the first representative prediction encoding cost and the second representative prediction encoding cost are selected through the above process, the first representative prediction encoding cost is compared by comparing the first representative prediction encoding cost and the second representative prediction encoding cost (S23). If it is less than the second representative prediction coding cost, the first representative selection motion vector is selected as the encoding target motion vector, and the first representative calculating motion vector is selected as the uncoding target motion vector (S25).

On the other hand, when the second representative prediction coding cost is smaller than the first representative prediction coding cost, the second representative selection motion vector is selected as the encoding target motion vector and the second representative calculating motion vector is selected as the uncoding target motion vector (S24). .

When selection of the encoding target motion vector and the uncoding target motion vector is completed through the above process, a prediction block for the current block is generated using the selected encoding target motion vector and the uncoding target motion vector. In addition, an excess block that is a deviation between the current block and the prediction block is generated (S26).

Then, the encoding target motion vector and the excess block are encoded (S27). At this time, in the encoding process, the bi-prediction encoding mode information having information indicating that the image is encoded by the above-described encoding method, that is, the bi-prediction encoding method according to the present invention is encoded together. Can be sent. As a result, in the decoding process, an uncoded target motion vector that is not transmitted is calculated using the transmitted target encoding motion vector.

In the above process, the first reference picture and the second reference picture may be determined to correspond to each other. When the second selection motion vector is selected from the second reference picture to which the first calculation motion vector belonged by the first selection motion vector belongs, the second calculation motion is performed using the second selection motion vector selected from the second reference picture. The first reference picture used when calculating the vector may be a first reference picture to which the first selection motion vector belongs.

Hereinafter, a method of calculating a first calculated motion vector and a second calculated motion vector in a bi-prediction encoding method according to the present invention will be described in detail with reference to FIGS. 3 to 5.

The first calculated motion vector according to the present invention is calculated by multiplying a first temporal motion vector by a relative time distance between the current picture to which the current block belongs, the first reference picture, and the second reference picture. Similarly, the second calculated motion vector is calculated by multiplying the second temporal motion vector by a relative time distance between the current picture to which the current block belongs, the first reference picture and the second reference picture. Hereinafter, a description will be given with reference to the first selection motion vector and the first calculation motion vector.

3 (a) and 3 (b) show a case in which the first reference picture and the second reference picture are positioned before and after each other in time with respect to the current picture, and (a) shows that the first reference picture is in time with respect to the current picture. Previously, the second reference picture is shown in the case where each is later in time of the current picture, and (b) shows the opposite case.

As shown in FIG. 3, when the first reference picture and the second reference picture are positioned before and after each time in the center of the current picture, the first calculated motion vector is calculated through Equation 5 below.

[Equation 5]

Where mv _cal is the first calculated motion vector, mv _sel is the first selected motion vector, TD _B is the distance in time between the first reference picture and the current picture, and TD _C is the time in time between the second reference picture and the current picture. Is the distance.

In addition, TD _C may be expressed as Equation 6 below.

&Quot; (6) "

Here, TD _D is a distance in time between the first reference picture and the second reference picture, and TD _B is a distance in time between the current picture and the first reference picture.

4A and 4B show a case in which both the first reference picture and the second reference picture are co-located in time with respect to the current picture, and (a) shows that the first reference picture is from the second reference picture. It shows the case where it is located later in time, and (b) shows the case where the second reference picture is located later in time from the first reference picture.

As shown in FIG. 4, when both the first reference picture and the second reference picture are located in time before the current picture, the first calculated motion vector is calculated through Equation 7 below.

[Equation 7]

Where mv _cal is the first calculated motion vector, mv _sel is the first selected motion vector, TD _B is the distance in time between the first reference picture and the current picture, and TD _C is the time in time between the second reference picture and the current picture. Is the distance.

When TD _C of Equation 7 is expressed using TD _D which is the distance in time between the first reference picture and the second reference picture, and TD _B which is the distance in time between the current picture and the first reference picture, FIG. In the case of 4 (a), it can be expressed as [Equation 8], and in the case of FIG. 4 (b) as [Equation 9].

[Equation 8]

[Equation 9]

5 (a) and 5 (b) show a case in which both the first reference picture and the second reference picture are located together later in time from the current picture, and (a) shows that the first reference picture is from the second reference picture. The case where it is located before in time is shown, (b) shows the case where the 2nd reference picture is located before on time from a 1st reference picture.

As shown in FIG. 5, even when both the first reference picture and the second reference picture are positioned before in time from the current picture, the first calculated motion vector is calculated through Equation 7 above.

When TD _C of Equation 7 is expressed using TD _D which is the distance in time between the first reference picture and the second reference picture, and TD _B which is the distance in time between the current picture and the first reference picture, FIG. In the case of 5 (a), it can be expressed as [Equation 8], and in the case of FIG. 5 (b) as [Equation 9].

Here, when [Equation 5] to [Equation 9] is applied to the second calculation motion vector, the first reference picture and the second reference picture are changed. That is, the first reference picture in Equations 5 to 9 is a reference picture from which a motion vector is selected, and the second reference picture is a reference picture from which a motion vector is calculated. Therefore, in the calculation of the second calculated motion vector, the motion vector is selected in the second reference picture and the motion vector is calculated in the first reference picture. In the calculation of the second calculated motion vector, Equations 5 to 9 ], The first reference picture and the second reference picture are replaced.

Hereinafter, a bi-prediction encoding apparatus according to the present invention will be described in detail with reference to FIGS. 6 and 7.

Referring to FIG. 6, the bi-prediction encoding apparatus according to the present invention includes an image input unit 10, a motion predictor 16, a motion compensator 17, and a motion vector encoder 18. The redundant data encoder 12, the redundant data decoder 13, the entropy encoder 14, the multiplexer 19, and the encoding controller 11 may be included.

The video input unit 10 receives a digital image which is not compressed into an image to be encoded. Here, the image data input to the image input unit 10 is composed of blocks divided into predetermined sizes. The image data input through the image input unit 10 is subtracted through the prediction block output from the motion compensator 17, that is, the compensation value and the subtractor, and output to the surplus data encoder 12.

When the image data is input through the image input unit 10, the encoding controller 11 determines a coding type according to whether to perform motion compensation on the input image data, for example, intra coding and inter coding, and controls correspondingly. Perform the action.

The redundant data encoder 12 quantizes the image data output from the subtractor, that is, transform coefficient values obtained by transform encoding the excess block according to a predetermined quantization process, and N, which is two-dimensional data composed of quantized transform coefficient values. XM Generate data. In this case, a discrete cosine transform (DCT) method may be used as an example of a transformation method applied by the redundant data encoder 12.

Referring to FIG. 6, the bi-prediction encoding apparatus according to the present invention includes an image input unit 10, a motion predictor 16, a motion compensator 17, and a motion vector encoder 18. The redundant data encoder 12, the redundant data decoder 13, the entropy encoder 14, the multiplexer 19, and the encoding controller 11 may be included.

The video input unit 10 receives a digital image which is not compressed into an image to be encoded. Here, the image data input to the image input unit 10 is composed of blocks divided into predetermined sizes. The image data input through the image input unit 10 is subtracted through the prediction block output from the motion compensator 17, that is, the compensation value and the subtractor, and output to the surplus data encoder 12.

When the image data is input through the image input unit 10, the encoding controller 11 determines a coding type according to whether to perform motion compensation on the input image data, for example, intra coding and inter coding, and controls correspondingly. Perform the action.

The surplus data encoder 12 quantizes image data output from the subtractor, that is, transform coefficient values obtained by transform encoding the excess block according to a predetermined quantization process, and is two-dimensional data composed of quantized transform coefficient values. Referring to FIG. 6, the bi-prediction encoding apparatus according to the present invention includes an image input unit 10, a motion predictor 16, a motion compensator 17, a motion vector encoder 18, The redundant data encoder 12, the redundant data decoder 13, the entropy encoder 14, the multiplexer 19, and the encoding controller 11 may be included.

The video input unit 10 receives a digital image which is not compressed into an image to be encoded. Here, the image data input to the image input unit 10 is composed of blocks divided into predetermined sizes. The image data input through the image input unit 10 is subtracted through the prediction block output from the motion compensator 17, that is, the compensation value and the subtractor, and output to the surplus data encoder 12.

When the image data is input through the image input unit 10, the encoding controller 11 determines a coding type according to whether to perform motion compensation on the input image data, for example, intra coding and inter coding, and controls correspondingly. Perform the action.

The redundant data encoder 12 quantizes the image data output from the subtractor, that is, transform coefficient values obtained by transform encoding the excess block according to a predetermined quantization process, and N, which is two-dimensional data composed of quantized transform coefficient values. XM Generate data. In this case, a discrete cosine transform (DCT) method may be used as an example of a transformation method applied by the redundant data encoder 12.

Since the image data input and encoded by the redundant data encoder 12 may be used as a reference picture for motion compensation for image data input after or before, the redundant data encoder 12 may be used through the redundant data decoder 13. Inverse quantization and inverse transform encoding are performed. The image data output from the redundant data decoding unit 13 is stored in the memory unit 15. When the image data output from the redundant data decoding unit 13 is differential image data, the output of the motion compensation unit 17 is output. The data, i.e., added to the prediction block, is then stored in the memory unit 15.

Meanwhile, the motion predictor 16 predicts motion using a plurality of reference pictures, that is, the first reference picture and the second reference pictures. The motion predictor 16 selects a motion vector to be encoded and a motion vector to be uncoded through the aforementioned bi-prediction encoding method, and the motion compensator 17 encodes a motion vector to be encoded and an uncoded object. A prediction block, that is, a compensation value, for the current block is calculated using the motion vector.

7 is a diagram showing the configuration of the motion predictor 16 according to the present invention. Referring to FIG. 7, the first motion vector selector 110a may respectively select a first selected motion corresponding to the first reference picture within a preset motion search range from the first reference pictures based on the current block to be encoded. Select the vector.

The first motion vector calculator 111a calculates a first calculated motion vector corresponding to the second reference picture by using the first selected motion vector selected by the first motion vector selector 110a. Here, the above-described Equation 5 to Equation 9 may be used for calculating the first calculated motion vector, and a detailed description thereof will be omitted.

The second motion vector selector 110b selects a second selection motion vector corresponding to the second reference picture, respectively, from the second reference pictures extracted by the reference picture extractor. The second motion vector calculator 111b calculates a second calculated motion vector corresponding to the first reference picture by using the second motion vector selected by the second motion vector selector 110b. Here, the above-described equations (5) to (9) may be used as the calculation process of the second calculated motion vector, and a detailed description thereof will be omitted.

The first motion prediction block generation unit 112a may include a first selection motion prediction block corresponding to the first selection motion vector and a first calculation motion vector based on the first selection motion vector and the first calculation motion vector. 1 Generate a calculated motion prediction block.

The first encoding cost calculator 113a calculates a first prediction encoding cost by using the first selected motion vector, the first calculated motion vector, the first selected motion prediction block, and the first calculated motion prediction block. Here, the first prediction encoding cost is calculated for each of the first selective encoding costs respectively selected from the plurality of first reference pictures.

Here, the first encoding cost selecting unit 114a selects any one of the first predictive encoding costs calculated by the first encoding cost calculating unit 113a that satisfies a preset condition. As an example, one having the least encoding cost is selected.

In addition, the first encoding cost selecting unit 114a may determine the selected first predictive encoding cost, the first selected motion vector for the selected first predictive encoding cost, and the first calculated motion vector as the first representative predictive encoding cost, and the first. A representative selection motion vector and a first representative output motion vector are selected.

In the same manner, the second motion prediction block generator 112b may generate a second selected motion prediction block and a second calculated motion vector corresponding to the second selected motion vector based on the second selected motion vector and the second calculated motion vector. Generate a second calculated motion prediction block corresponding to.

The second encoding cost calculator 113b uses the second selected motion vector, the second calculated motion vector, the second selected motion prediction block, and the second calculated motion prediction block to calculate a plurality of second prediction encoding costs. Calculate.

Here, the second encoding cost selecting unit 114b selects any one of the second predictive encoding costs calculated by the second encoding cost calculating unit 113b that satisfies a preset condition. As an example, one having the least encoding cost is selected.

In addition, the second encoding cost selecting unit 114b converts the selected second prediction encoding cost, the second selected motion vector for the selected second prediction encoding cost, and the second calculated motion vector into the second representative prediction encoding cost, and the second. The representative selection motion vector and the second representative output motion vector are selected.

The first representative prediction encoding cost and the second representative prediction encoding cost selected by the first representative encoding cost selecting unit and the second representative encoding cost selecting unit are compared by the motion vector selecting unit 115. If the first representative prediction coding cost is less than the second representative prediction coding cost, the motion vector selecting unit 115 selects the first representative selection motion vector as a code and a target motion vector and decodes the first representative calculation motion vector. Selected as the target motion vector.

On the other hand, when the second representative prediction coding cost is less than the first representative prediction coding cost, the motion vector selecting unit 115 selects the second representative selection motion vector as the encoding target motion vector and selects the second representative calculating motion vector for the uncoding object. Select by motion vector.

The motion compensator 17 performs motion compensation and outputs the prediction block to the subtractor by using the motion vector to be encoded and the motion vector to be uncoded through the above process. Here, the motion compensator 17 determines and outputs a prediction block for the current block based on the encoding target motion vector and the uncoding target motion vector.

On the other hand, the motion vector encoder 18 encodes and outputs the encoding target motion vector selected by the motion predictor 16. Here, the motion vector encoder 18 may encode bi-prediction encoding mode information for transmitting that the encoded motion vector is encoded by a bi-prediction encoding method. .

The data encoded by the motion vector encoder 18 is input to the multiplexer 19 together with the image data encoded by the entropy encoder 14, and is generated and transmitted in the form of a compressed bitstream.

Hereinafter, a bi-prediction decoding method according to the present invention will be described in detail with reference to FIG. 8.

First, when a compressed bitstream that is encoded data is input (S30), entropy decoding is performed on the input bitstream (S31). Next, the entropy decoded data is analyzed to determine whether the current block to be decoded is in the bi-prediction encoding mode (S32). Here, the determination of whether the mode is a bi-prediction encoding mode is the existence of the bi-prediction encoding mode information encoded in the process of encoding through the above-described bi-prediction encoding method. You can check whether or not.

Here, when it is determined in the bi-prediction encoding mode, the decoding target motion vector is recovered from the entropy decoded data (S33). Here, the decoding target motion vector is an encoding target motion vector encoded by the above-described bi-prediction encoding method.

Then, the distance in time between the current picture to which the current block to be decoded belongs and the reference picture corresponding to the motion vector to be decoded (hereinafter, referred to as a 'first decoded reference picture'), a reference picture different from the current picture (hereinafter, A non-decoded target motion vector for the second decoded reference picture is calculated using a time distance between the second decoded reference picture and the reconstructed decoded motion vector (S34).

Here, the calculation of the decoded target motion vector is performed by using the first selected motion vector (or the second selected motion vector) in the above-described bi-prediction encoding method. Vector), that is, using [Equation 5] to [Equation 9].

Here, the decoded motion vector, the first decoded reference picture and the second decoded reference picture in the bi-prediction decoding method are respectively the first selected motion vector and the first decoded method in the bi-prediction coding method. 1 is applied to [Equation 5] to [Equation 9] corresponding to the first reference picture from which the selection motion vector is selected and the second reference picture corresponding to the calculated first calculation motion vector, thereby decoded motion vectors. Is calculated.

As described above, when the decoded motion vector is calculated, a prediction block is generated using the decoded motion vector and the decoded motion vector (S35). Here, in the prediction block, a pair of prediction blocks may be generated corresponding to each of the first decoded reference picture and the second decoded reference picture.

When the prediction block is generated, the encoded data is subjected to an entropy decoding process and a redundant data decoding process to generate a surplus block for the current block, and the surplus block and the prediction block are summed to generate a reconstructed image of the current block (S36). ).

On the other hand, when the data input in step S32 is not the bi-prediction encoding mode, decoding is performed through a predetermined decoding process, for example, a known decoding method (S37).

Hereinafter, a bi-prediction decoding apparatus according to the present invention will be described in detail with reference to FIG. 9. As illustrated in FIG. 9, the bi-prediction decoding apparatus according to the present invention includes an entropy decoding unit 30, a redundant data decoding unit 31, a motion vector decoding unit 35, and a motion compensation unit ( 34, a memory 33, and a decryption controller 32.

The entropy decoding unit 30 entropy decodes and outputs coded and input data. Here, the entropy decoding unit 30 entropy-decodes and outputs entropy-coded quantized transform coefficient values, a motion vector to be decoded, information about a motion vector, and the like.

The surplus data decoding unit 31 dequantizes the image data by inverse quantization and inverse transformation of the entropy decoded transform coefficient values.

The decoding control unit 32 extracts the encoding type of the encoded and input data from the data decoded by the entropy decoding unit 30. Here, in the case where the encoded and input data is data encoded by the above-described bi-prediction encoding method, the bi-prediction coded together in the bi-prediction encoding process. The encoding mode information is received, and the decoding control unit 32 recognizes that the corresponding data is encoded through a bi-prediction encoding method based on the reception of the bi-prediction encoding mode information. .

When the motion vector decoder 35 determines that the decoding control unit 32 is in the bi-prediction encoding mode, the motion vector decoder 35 restores a decoding target motion vector among the data decoded by the entropy decoding unit 30. . Referring to FIG. 10, the motion vector decoder 35 may include a decoding target motion vector reconstructor 130 and an undecoded target motion vector calculator 133.

When the decoding target motion vector reconstruction unit 130 determines that the decoding control unit 32 is in the bi-prediction encoding mode, the motion vector decoded by the entropy decoding unit 30 is the decoding target motion vector. To be recognized.

The decoded object motion vector calculating unit 133 decodes the decoded object motion vector using a distance in time between the decoded object motion vector and the current picture of the first decoded reference picture and the second decoded reference picture stored in the memory 33. Calculate the target motion vector. Here, since the calculation of the decoded target motion vector has been described in the above-described Bi-prediction decoding method, a detailed description thereof will be omitted.

Meanwhile, the motion compensator 34 generates a prediction block by using the decoding target motion vector and the undecoding target motion vector output from the motion vector decoder 35. As described above, in the prediction block, a pair of prediction blocks may be generated corresponding to each of the first decoded reference picture and the second decoded reference picture.

The predicted block generated by the motion compensator 34 is added through an excess block and an adder output through an entropy decoding process and an excess data decoding process to generate a reconstructed image of the current block (S36). In addition, the reconstructed image of the generated current block is stored in the memory unit 33 for motion compensation.

Although the preferred embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements of those skilled in the art using the basic concepts of the present invention defined in the following claims are also provided. It belongs to the scope of rights.

Claims (11)

In a bi-prediction coding method,
(a) selecting a first selection motion vector from a first reference picture for the current block to be encoded;
(b) calculating a first calculated motion vector for a second reference picture based on the first selected motion vector;
(c) based on the first selection motion vector, the first calculation motion vector, a first selection motion prediction block corresponding to the first selection motion vector, and a second calculation motion prediction block corresponding to the first calculation motion vector. Calculating a first predictive encoding cost;
(d) Repeating steps (a), (b) and (c), the first representative selection motion vector, the first representative calculation motion vector, and the first representative selection motion satisfying a predetermined criterion Selecting a first representative prediction coding cost according to a vector and the first representative calculation motion vector;
(e) selecting a second selection motion vector from the second reference picture;
(f) calculating a second calculated motion vector for the first reference picture based on the second selected motion vector;
(g) based on the second selected motion vector, the second calculated motion vector, a second selected motion prediction block corresponding to the second selected motion vector, and a second calculated motion prediction block corresponding to the second calculated motion vector. Calculating a second predictive encoding cost;
(h) repeating steps (e), (f), and (g) to perform a second representative selection motion vector, a second representative calculation motion vector, and the second representative selection motion that satisfy a predetermined criterion; Selecting a second representative prediction coding cost according to a vector and the second representative calculating motion vector;
(i) if the first representative prediction coding cost is less than the second representative prediction coding cost, the first representative selection motion vector is selected as the encoding target motion vector, and the first representative calculation motion vector is selected as the uncoding target motion vector. And when the second representative prediction coding cost is less than the first representative prediction coding cost, selecting the second representative selection motion vector as the encoding target motion vector and the second representative calculating motion vector as the uncoding target motion vector. Steps;
(j) encoding the encoding target motion vector, wherein the bi-prediction encoding method is included. The method of claim 1,
Step (j) encodes bi-prediction encoding mode information for conveying that encoding is a bi-prediction encoding mode encoded by the bi-prediction encoding method. Bi-prediction encoding method comprising the step of. The method of claim 2,
In the step (b), the first calculated motion vector is calculated by multiplying the relative time distance between the current picture to which the current block belongs, the first reference picture and the second reference picture by the first selected motion vector. ;
In the step (f), the second calculated motion vector is calculated by multiplying the relative time distance between the current picture, the first reference picture and the second reference picture by the second selected motion vector. Bi-prediction coding method. The method of claim 3,
When the first reference picture and the second reference picture are located before and after each other in time with respect to the current picture, the first calculated motion vector and the second calculation in steps (b) and (f), respectively. The motion vectors are each equation

Where mv _cal is the first calculated motion vector or the second calculated motion vector, mv _sel is the first selected motion vector or the second selected motion vector, and TD _B is the first reference picture and the first motion vector. 2 is a temporal distance between any one of the first selected motion vector or the second selected motion vector among the reference pictures and the current picture, and TD _C is the first calculation of the first reference picture and the second reference picture; Bi-prediction encoding method characterized in that the motion vector or the second calculated motion vector is calculated by any one of the calculated time and the current picture. The method of claim 3,
When the first reference picture and the second reference picture are located together before or after in time from the current picture, the first calculated motion vector and the second calculated motion in steps (b) and (f). Vectors are equations

Where mv _cal is the first calculated motion vector or the second calculated motion vector, mv _sel is the first selected motion vector or the second selected motion vector, and TD _B is the first reference picture and the first motion vector. 2 is a temporal distance between any one of the first selected motion vector or the second selected motion vector among the reference pictures and the current picture, and TD _C is the first calculation of the first reference picture and the second reference picture; Bi-prediction encoding method characterized in that the motion vector or the second calculated motion vector is calculated by any one of the calculated time and the current picture. A computer-readable recording medium having recorded thereon a program code for performing a bi-prediction encoding method according to any one of claims 1 to 5. In a bi-prediction encoding apparatus,
A first motion vector selecting unit selecting a first selected motion vector corresponding to each of the first reference pictures within a preset motion search range based on a current block to be encoded from a plurality of first reference pictures;
A first motion vector calculator configured to calculate a first calculated motion vector corresponding to the first selected motion vector based on the first selected motion vector and a second reference picture corresponding to the first selected motion vector; ;
The angle based on the first selection motion vector, the first calculation motion vector, a first selection motion prediction block corresponding to the first selection motion vector, and a first calculation motion prediction block corresponding to the first calculation motion vector. A first encoding cost calculator for calculating a first prediction encoding cost respectively corresponding to the first selected motion vector;
A first encoding cost selecting unit configured to select a first representative prediction encoding cost that satisfies a predetermined condition by comparing the plurality of first prediction encoding costs calculated corresponding to each of the first selected motion vectors;
A second motion vector selector which selects a second selection motion vector corresponding to each second reference picture from each second reference picture;
A second motion vector calculator configured to calculate a second calculated motion vector corresponding to the second selected motion vector based on the second selected motion vector and the first reference picture corresponding to the second selected motion vector; Wow;
The angle based on the second selection motion vector, the second calculation motion vector, the second selection motion prediction block corresponding to the second selection motion vector, and the second calculation motion prediction block corresponding to the second calculation motion vector. A second encoding cost calculator for calculating second predictive encoding costs respectively corresponding to the second selected motion vector;
A second encoding cost selecting unit configured to compare the plurality of second prediction encoding costs calculated corresponding to the second selection motion vectors, and select a second representative prediction encoding cost that satisfies a preset condition;
When the first representative prediction coding cost is smaller than the second representative prediction coding cost, any one of the first selected motion vectors corresponding to the first representative prediction coding cost is selected as an encoding target motion vector, and the second representative A motion vector selecting unit that selects one of the second selected motion vectors corresponding to the second representative prediction coding cost as an encoding target motion vector when a prediction coding cost is smaller than the first representative prediction coding cost;
And a motion vector encoder for encoding the motion vector to be encoded. The method of claim 7, wherein
The motion vector encoder,
Encoding bi-prediction encoding mode information for transmitting that the encoding of the encoding target motion vector is a bi-prediction encoding mode encoded by a bi-prediction encoding method. Bi-prediction encoding apparatus, characterized in that. The method of claim 8,
The first motion vector calculator calculates the first calculated motion vector by multiplying a first temporal motion vector by a distance in time between the current picture to which the current block belongs, the first reference picture and the second reference picture. and;
The second motion vector calculator calculates the second calculated motion vector by multiplying the second time-dependent motion vector by a distance in time between the current picture to which the current block belongs, the first reference picture and the second reference picture. Bi-prediction coding apparatus, characterized in that. 10. The method of claim 9,
When the first reference picture and the second reference picture are located before and after each other in time with respect to the current picture, the first motion vector calculating unit and the second motion vector calculating unit are arranged in the first calculated motion vector and the first picture. The second output motion vector is respectively expressed as

Where mv _cal is the first calculated motion vector or the second calculated motion vector, mv _sel is the first selected motion vector or the second selected motion vector, and TD _B is the first reference picture and the first motion vector. 2 is a temporal distance between any one of the first selected motion vector or the second selected motion vector among the reference pictures and the current picture, and TD _C is the first calculation of the first reference picture and the second reference picture; Bi-prediction encoding apparatus, characterized in that it is calculated by the motion vector or the second calculated motion vector is the distance in time between the calculated one and the current picture. 10. The method of claim 9,
When the first reference picture and the second reference picture are located together before or after in time from the current picture, the first motion vector calculator and the second motion vector calculator include the first calculated motion vector and the first motion picture. 2 output motion vectors, respectively

Where mv _cal is the first calculated motion vector or the second calculated motion vector, mv _sel is the first selected motion vector or the second selected motion vector, and TD _B is the first reference picture and the first motion vector. 2 is a temporal distance between any one of the first selected motion vector or the second selected motion vector among the reference pictures and the current picture, and TD _C is the first calculation of the first reference picture and the second reference picture; Bi-prediction encoding apparatus, characterized in that it is calculated by the motion vector or the second calculated motion vector is the distance in time between the calculated one and the current picture.

Images