KR20090103663A - Method for coding data of motion vector and apparatus for the same - Google Patents

Abstract

PURPOSE: A method for coding data of a motion vector and an apparatus for the same are provided to reduce the information for motion estimation by encoding a reset satisfaction condition for motion vector related data. CONSTITUTION: A method for coding data of a motion vector comprises the steps of: setting up subunits in which encoding units are split into two regions calculating a PMV(Predictive Motion Vector) for plural subunits(260); confirming whether or not the plural PMVs satisfies a preset condition; and generating the encoding data for the motion vector by including the information that indicates the omission of the subunit motion vector if the plural PMVs satisfy the preset condition(350).

Description

METHOD FOR CODING DATA OF MOTION VECTOR AND APPARATUS FOR THE SAME}

The present invention relates to a video encoding method, and more particularly, to a video encoding method capable of improving the compression ratio of information on a motion vector.

The recently standardized H.264 / AVC standard has excellent rate-distortion performance while maintaining a block-based encoding scheme like other conventional video compression standards. In particular, the H.264 / AVC standard divides a macroblock as one coding unit into subunits of a plurality of variable block sizes more precisely than the conventional video compression standard, and then estimates a motion vector for the divided block to compensate for motion. do.

Specifically, for a macroblock as a coding unit, as shown in FIG. 1, SKIP, P16 × 16, P16 × 8, P8 × 16, and P8 × 8 (P8 × 8, P8 × 4, P4 × 8, P4 × 4) Motion vectors are estimated and compensated in units of blocks. In H.264 / AVC, when using this variable block size motion compensation technique, up to 16 motion vectors are generated in one macroblock, one coding unit. This shows that a large amount of bits are required for motion vector coding. In order to predict the variable block size motion, a macroblock is composed of sub-units of which size to compensate for motion, as well as a positive effect due to more precise sub-compensation of motion compensation. A vector and a negative effect on bit overhead required for coding unit splitting information indicating how a given coding unit is divided into subunits, which are motion compensation units, are all determined.

As described above, when the data is divided into variable block sizes, the smaller the coding unit is, the smaller the residual data is after the motion compensation. However, since motion vectors corresponding to the number of divided subunits are generated, motion vectors for a large number of subunits are generated. Therefore, a problem arises in that excess data to be encoded decreases but the bit amount increases due to an increase in the number of motion vectors. As a result, a problem arises in which the effect of sub-movement compensation cannot be properly utilized.

SUMMARY OF THE INVENTION The present invention has been made in view of the foregoing, and an object thereof is to provide an encoding / decoding method and apparatus capable of effectively compressing information indicating a motion vector of a coding unit and data for surplus data.

In order to achieve the above object, a method of encoding motion vector information according to an aspect of the present invention is a method of encoding information about a motion vector of a coding unit, wherein a subunit obtained by dividing a coding unit into at least two regions is set. And calculating a predicted motion vector (PMV) for a plurality of subunits, checking whether the plurality of predicted motion vectors (PMV) satisfy a predetermined condition, and performing the plurality of predicted motion vectors. When the vector PMV satisfies the predetermined condition, the method includes generating coded data for the motion vector, including information indicating that the sub-unit motion vector is omitted.

According to an aspect of the present invention, there is provided a method of encoding motion vector information based on a plurality of motion vector prediction modes, and estimating a motion vector according to each mode, and applying the estimated motion vector to a reference picture. Restoring video information of the video data; calculating surplus data between the restored video information and the original video information; encoding and decoding the excess data; and encoding the encoded and decoded surplus data to the restored video. Applying, selecting a mode having a minimum cost among a process of generating reconstructed image information, calculating a cost between the reconstructed image information and original image information, and a plurality of motion vector prediction modes And encoding the minimum cost as the first cost, and applying a predicted motion vector (PMV) to a reference picture. Restoring the above video information, and calculating a second cost between the restored video information and the original video information, wherein the predetermined condition is that the second cost is relatively smaller than the first cost. It is preferable that it is a condition which has a value.

If the predetermined condition is not satisfied, generating encoded data for the motion vector, including a motion vector, a reference picture, and residual data estimated for the mode having a minimum cost; Can be.

The predetermined condition is that the prediction motion vector (PMV) of the sub-unit and the motion vector searched by the sub-unit have the same value, or the reference pictures of the motion vectors of the sub-units included in the coding unit are all the same reference picture. It may be a condition made.

Further, the reference picture may be a picture that is timely neighboring to a picture currently being encoded, or any one of pictures stored in a frame memory, and in the latter case, the encoded data is at least one of the pictures. It is preferable to include information indicating a reference picture.

The operation of calculating the predicted motion vector (PMV) may include obtaining a motion vector for subunits located nearby, calculating a median value of the motion vector for the subunits located nearby, and calculating the subunit. It may include the process of predicting the predicted motion vector (PMV) of.

According to another aspect of the present invention, there is provided a method for recovering information on a motion vector, the method for recovering information on a motion vector from a bitstream including an identification code indicating a unit from which the motion vector is obtained. Parsing the identification code included in the processor to obtain a Motion Vector Coding Mode (MVCM), and if the mode is a mode indicating omission of the sub-unit motion vector, A motion compensation block is performed by calculating a predictive motion vector (PMV) according to a previously promised condition, obtaining information about a reference picture, and applying the predictive motion vector (PMV) to the reference picture. It includes the process of restoring.

According to another aspect of the present invention, a method for restoring information on a motion vector may include encoding a motion vector and a coded surplus when the motion vector encoding mode (MVCM) is not a mode indicating omission of a sub-motion vector. and reconstructing the image information of the coding unit by reconstructing the residual data, identifying the reference picture, and applying the motion vector and the surplus data to the reference picture.

The process of calculating a predicted motion vector (PMV) may include obtaining a motion vector for subunits located in the periphery, calculating a median value of the motion vector for the subunits in the periphery, and calculating the median value. It is preferable to include the step of setting to the predicted motion vector (PMV).

The reference picture may be a picture neighboring in time with a picture currently being encoded.

An apparatus for encoding information about a motion vector according to another aspect of the present invention is an apparatus for encoding information about a motion vector of a coding unit by using a motion estimator and a motion compensator, wherein a plurality of motion vector prediction modes ( Instructs the search of a motion vector and operation of a sub-predictive motion vector (PMV) for MVPM (Motion Vector Predictive Mode), calculates the cost of the reconstructed coding unit, and calculates the cost of the motion estimation unit and the motion compensation unit. The controller controls an operation and a motion vector of each of the motion vector prediction modes (MVPM) corresponding to the motion vector prediction mode (MVPM) input by the controller, and then provides the motion compensation unit to provide a motion compensation unit. A motion estimator for calculating a subunit predicted motion vector (PMV) using the vector and then providing the motion compensation unit And a motion compensator for restoring image information of a coding unit by applying the motion vector received from the video estimation unit to the reference picture, wherein the control unit indicates to omit the motion vector for the coding unit when a predetermined condition is satisfied. Information to be included in the encoded data.

If the predetermined condition is not satisfied, the controller includes a motion vector prediction mode (MVPM) having a minimum cost, a motion vector searched based on the mode (MVPM), and surplus data (residual). It is preferable to generate encoded data for the vector.

The predetermined condition is that the cost based on the subunit predicted motion vector PMV has a value smaller than the minimum cost calculated based on the motion vector predictive mode MVPM, and the subunit predicted motion vector PMV. PMV) and the motion vector obtained through the subunit search may be the same condition.

The predetermined condition may be a condition in which all of the reference pictures of the sub-unit motion vector are the same reference picture.

The reference picture may be a picture neighboring in time with a picture currently being encoded.

Preferably, the reference picture is any one selected picture from among pictures stored in a frame memory, and the encoded data includes information indicating at least one reference picture.

According to another aspect of the present invention, an apparatus for encoding information about a motion vector includes: a redundant data encoder for encoding surplus data between original image information and reconstructed image information of a coding unit; and reconstructing the encoded excess data. The apparatus may further include a redundant data decoder, and the controller may calculate a cost using the number of bits of the redundant data provided from the redundant data encoder and a distortion provided from the redundant data decoder.

Preferably, the motion estimator obtains a motion vector for subunits located nearby, calculates a median value of the motion vector for subunits located nearby, and calculates the predicted motion vector of the subunit.

An apparatus for restoring information on a motion vector according to another aspect of the present invention is an apparatus for restoring information on a motion vector from a bitstream, the apparatus for restoring information about a motion vector from a bitstream, wherein the motion vector coding mode (MVCM; MVCM; Motion Vector Coding) A control unit for acquiring a mode, controlling a reconstruction of a motion vector corresponding to the mode (MVCM) and the reconstruction of image information of a coding unit, and a condition promised in advance with the mode (MVCM) and the encoder provided from the controller. With reference to the motion vector decoder for restoring motion information including the encoded motion vector and the reference picture, and the mode (MVCM) provided from the controller, the sub-prediction motion vector (PMV) is referred to. A motion vector predictor calculating a vector) and applying the motion vector to the reference picture to restore image information. A motion compensator, a surplus data decoder for restoring the encoded surplus data, and an adder for adding video information output from the motion compensator and the surplus data from the redundant data decoder.

When the mode (MVCM) is a mode for instructing to omit a motion vector of a sub-unit, the controller may control the motion vector predictor according to a condition previously promised with an encoder to obtain information about a reference picture. have.

The motion vector predictor is configured to obtain a motion vector for subunits located nearby, calculate an intermediate value of the obtained motion vectors, and set the intermediate value as a predicted motion vector (PMV) of the subunit. desirable.

The reference picture may be a picture neighboring in time with a picture currently being encoded.

When the mode is not a mode for instructing to omit the sub-motion vector, the controller controls the motion vector decoder and the redundant data decoder according to a condition previously promised by the encoder, and the motion vector decoder Preferably, the motion vector and the reference picture are obtained from the input data, and the redundant data decoder obtains the redundant data by referring to the input bitstream.

According to the present invention, information about motion prediction can be effectively reduced by encoding whether or not a predetermined condition on data related to a motion vector is satisfied.

In addition, the motion vector information of the sub-units can be encoded by encoding information indicating one macroblock encoding mode without the need to encode and transmit the motion vector information of the plurality of sub-units. It is possible to restore it.

1 illustrates an example of a motion vector prediction mode according to the H.264 / AVC standard.

2 is a block diagram of a video encoding apparatus to which a motion information encoding apparatus according to an embodiment of the present invention is applied;

3 is a flowchart illustrating a procedure of a method of encoding motion vector information according to an embodiment of the present invention;

4 is an exemplary diagram of a macroblock for encoding motion vector information according to an embodiment of the present invention;

5 is a flowchart illustrating a procedure of a method of encoding motion vector information according to another embodiment of the present invention;

6 is a flowchart showing a procedure of a method of encoding motion vector information according to another embodiment of the present invention;

7 is a block diagram illustrating a video decoding apparatus to which a motion vector decompression device is applied according to an embodiment of the present invention;

8 is a flowchart illustrating a procedure of a method of decoding motion vector information according to an embodiment of the present invention.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. Specific details appear in the following description, which is provided to help a more general understanding of the present invention, and it is common knowledge in the art that such specific matters may be changed or modified within the scope of the present invention. It is self-evident to those who have.

An embodiment of the present invention illustrates the application of the encoding method and apparatus to the H.264 / AVC standard. Therefore, in the embodiment of the present invention, the coding unit uses a macroblock unit that is a unit for encoding pixels included in a picture of an image in the H.264 / AVC standard.

In an embodiment of the present invention, as a unit for obtaining a motion vector, a unit obtained by dividing a coding unit (eg, a macroblock) into at least two is defined as a subunit. Furthermore, in the embodiment of the present invention illustrating the H.264 / AVC standard, the subunit may be a unit consisting of 16 × 8, 8 × 16, 8 × 8, 8 × 4, 4 × 8, 4 × 4 pixels, and the like. have.

Although, according to an embodiment of the present invention, a coding unit is illustrated as a macroblock while an encoding and decoding method based on the H.264 / AVC standard is illustrated as a method for encoding and decoding a video, and subunits are 16 × 8, 8 It illustrated by the unit which consists of x16, 8x8, 8x4, 4x8, 4x4 pixel. However, the present invention is not limited thereto and may be variously applied to a method of encoding and decoding a video.

Furthermore, in an embodiment of the present invention, a picture being encoded is defined as a current picture (CP), and a macroblock being encoded is defined as a current macroblock (CMB). As shown in FIG. 1, a mode for estimating a motion vector in predetermined units (16 × 16, 16 × 8, 8 × 16, 8 × 8, 8 × 4, 4 × 8, and 4 × 4 pixel units). Is defined as a Motion Vector Predictive Mode (MVPM), and a motion vector predicted using motion vectors of neighboring coding units (or sub-units) is defined as a Predictive Motion Vector (PMV). The coding mode determined according to the motion vector coding method of the present invention is defined as a motion vector coding mode (MVCM).

Meanwhile, in the process of encoding a video, as one coding unit such as a macroblock is divided into a predetermined number of subunits and motion compensated, a motion vector is generated for each subunit.

In particular, in the case of a macroblock corresponding to a region having little motion or a macroblock corresponding to a region moving in the same direction, there may be a significant correlation between the plurality of motion vectors.

Therefore, in the encoding method of the present invention, the relationship between the motion vector of the neighboring macroblock and the motion vector of the current macroblock is checked, and only information on whether the relationship satisfies a predetermined condition is encoded. Correspondingly, in the decoding process of the present invention, information on whether a predetermined condition is satisfied is checked, and the motion vector of the macroblock is restored by applying the predetermined condition.

2 is a block diagram of a video encoding apparatus to which a motion information encoding apparatus according to an embodiment of the present invention is applied. 2, a video encoding apparatus to which a motion information encoding apparatus of an embodiment of the present invention is applied includes a motion estimator 110, a motion compensator 120, a subtractor 125, a reference picture memory 130, The redundant data encoder 140, the redundant data decoder 150, the adder 155, the entropy encoder 160, the motion vector encoder 170, the controller 180, and the multiplexer 190 are included. .

The motion estimator 110 receives one of the motion vector prediction modes (MVPM) shown in FIG. 1 from the controller 180. In addition, referring to at least one reference picture stored in the reference picture memory 130, a unit corresponding to the received mode (MVPM) (for example, 16 × 16, 16 × 8, 8 × 16, 8 × 8, 8). Motion vectors are searched in units of 4x4, 4x8 and 4x4 pixels). In this case, the motion estimation unit 110 may directly receive information indicating a reference picture from the control unit 180 and search for a motion vector by referring to the received reference picture. In addition, as an alternative thereto, the motion estimator 110 may simply receive only a motion vector prediction mode (MVPM) from the controller 180. In this case, the motion estimator 110 calculates an error value with respect to a current picture and pictures that are located in a timely vicinity, and searches for a motion vector based on a reference picture having a minimum error value. . The motion estimation unit 110 also provides the motion vector and the index information of the reference picture.

In addition, the motion estimation unit 110 is instructed by the control unit 180 to calculate the subunit prediction motion vector PMV, and calculates the prediction motion vector PMV. The motion estimator 110 provides the calculated result, that is, the predicted motion vector PMV, to the motion compensator 120. For example, the motion estimation unit 110 is instructed to calculate a prediction motion vector (PMV) in units of 8 × 8 pixels from the controller 180, and predicts motion in units of 8 × 8 pixels with respect to the current macroblock (CMB). Calculate the vector PMV.

Although, in the embodiment of the present invention, the motion estimator 110 illustrates calculating the predicted motion vector (PMV), the present invention is not limited thereto. For example, the controller 180 or the motion vector encoder It is also possible for 170 to calculate the predicted motion vector PMV.

The motion compensator 120 receives index information about a motion vector and a reference picture from the motion estimator 110. The macroblock is generated by applying a motion vector to the reference picture.

The subtractor 125 receives the macroblock in the reference picture corresponding to the input macroblock from the motion compensator 120 and performs the difference operation with the input macroblock when the input macroblock is predictively encoded between the pictures. (residue) Print data.

The surplus data encoder 140 converts the surplus data output from the subtractor 125 into DCT transform and quantization operation and then converts the surplus data into encoded data. The surplus data decoder 150 restores the encoded surplus data and outputs the encoded surplus data.

The adder 155 combines the macroblock reconstructed by the motion compensator 120 and the surplus data output through the surplus data decoder 150 to generate a reconstructed image and then stores the reconstructed image in the reference picture memory 130. Save it.

The entropy encoder 160 entropy encodes and outputs the surplus data output from the surplus data encoder 140, and the motion vector encoder 170 outputs the motion vector and the motion vector output from the motion estimator 110. Information about a reference picture is encoded and output.

Meanwhile, the controller 180 collectively controls the operations of the respective functional units. In particular, the controller 180 determines whether the motion vector of the current macroblock (CMB) satisfies a predetermined condition, and determines a motion vector encoding mode (MVCM).

That is, the controller 180 controls the current macroblock CMB to search for the motion vector in units corresponding to the motion vector prediction mode MVPM shown in FIG. 1. The operation of the motion estimator 110, the motion compensator 120, and the reference picture memory 130 is controlled to generate a motion compensated picture based on the found motion vector.

Through such control, the controller 180 obtains data that is a criterion of a predetermined condition. In addition, when the obtained data satisfies a predetermined condition, the controller 180 provides the multiplexer 190 with a mode for instructing to omit motion vector information without information of a separate motion vector. Here, the predetermined condition may be a condition in which the cost based on the predicted motion vector PMV has a value that is relatively smaller than the cost based on the motion vector prediction mode MVPM having the minimum cost. Therefore, the controller 180 obtains a result of calculating and comparing the cost between the original image of the current macroblock (CMB) and the motion compensated macroblock as data that is a criterion of the predetermined condition.

The multiplexer 190 aligns the information about the entropy coded data, the motion vector prediction mode, and the coded motion vector, and outputs a bitstream.

The encoding apparatus according to the embodiment of the present invention described above may further include an intra predictor, a deblocking filter, and the like based on the H.264 / AVC standard. In addition, the redundant data encoder 140 and the redundant data decoder 150 may not only perform DCT transform and quantization (inverse DCT transform and inverse quantization) of redundant data, but also specific pictures (based on the H.264 / AVC standard). For example, a DCT transform and quantization (inverse DCT transform and inverse quantization) operation on an I picture may be further performed. Furthermore, the encoding apparatus according to an embodiment of the present invention may further perform encoding of image data based on the H.264 / AVC standard as well as motion vector encoding of the present invention.

Hereinafter, a motion vector encoding method according to an aspect of the present invention will be described based on the operations of the controller 180, the motion estimator 110, the motion compensator 120, and the reference picture memory 130.

3 is a flowchart illustrating a procedure of a method of encoding motion vector information according to an embodiment of the present invention. Referring to FIG. 3, in operation 210, a block of a coding unit, for example, a macroblock, is first received.

Next, in step 220, prediction through motion estimation and motion compensation is performed to select a motion vector prediction mode (MVPM) having an optimal cost. In other words, the controller 180 may determine the motion vector prediction modes (MVPMs) illustrated in FIG. 1, that is, SKIP, P16 × 16, P16 × 8, P8 × 16, P8 × 8, P8 × 4, P4 × 8, and P4. The x4 mode is sequentially provided to the motion estimation unit 110, and information about a position where a reference picture to be referred to for motion estimation is stored is provided. At the same time, the operation of the motion estimation unit 110 is instructed. Then, the motion estimation unit 110 searches for a motion vector for each mode, and provides the motion compensation unit 120 with an index for a motion vector prediction mode (MVPM), an estimated motion vector, and a reference picture. In response, the motion compensator 120 generates a reconstructed macroblock using a motion vector and a reference picture. The reconstructed macroblock sequentially passes through the subtractor 125, the redundant data encoder 140, and the redundant data decoder 150, thereby generating redundant data. The reconstructed macroblock and surplus data are transferred to the adder 155, combined, and a reconstructed first macroblock is generated. As a result, a first macroblock for each of the SKIP, P16 × 16, P16 × 8, P8 × 16, P8 × 8, P8 × 4, P4 × 8, and P4 × 4 modes is generated. The controller 180 finally calculates a cost between the first macroblock of each of the modes and the macroblock of the original. Then, the cost having the smallest value among the calculated costs is set as the first cost.

Next, the controller 180 sets a reference picture to be applied to the encoding method of the present invention (step 230) and instructs to transfer the index of the reference picture to the motion compensation unit 120, as previously defined. .

Preferably, the reference picture as previously defined may be the most recently stored picture (eg, a picture having a reference index of 0) among pictures stored in the reference picture memory 130.

Operation 240 may be performed through the controller 180 and the motion estimator 110. The controller 180 instructs the motion estimator 110 to calculate the subunit predicted motion vector PMV. Then, the motion estimation unit 110 calculates a predicted motion vector (PMV) for the subunits included in the macroblock by referring to neighboring subunits (for example, a block of 8 × 8 pixel units).

For example, referring to FIG. 4, when the current macroblock includes four subunits 21, 22, 23, and 24 of 8 × 8 pixel units, in step 240, subunits X ₁ , X ₂ , X ₃ Compute a predicted motion vector (PMV) for X ₄ (21,22,23,24). That is, the motion vectors of the left A block 25, the upper B block 26, and the right upper C block 27 which are neighboring subunits neighboring the first X ₁ block 21 of the macroblock to be encoded are MV, respectively. _A , MV _B , and MV _C , the value PMV _X1 of the predicted motion vector PMV of the X ₁ block 21 is obtained through the calculation of Equation 1 below. The calculation of the predictive motion vector PMV is performed independently of each component value (x component and y component) of the vector.

The neighboring blocks for X ₂ , which is the second subunit of FIG. 4, are based on the same criteria. The left block is the X ₁ block 21, the upper block is the C block 27, and the right upper block is the E block 29. ) When the motion vectors of the blocks are MV _X1 , MV _C , and MV _E , respectively, the predicted motion vector PMV _X2 of the X ₂ block 22 is obtained through the calculation of Equation 2 below. In addition, the prediction motion vector PMV _X3 is calculated according to Equation 3 below with respect to the X ₃ block 23 which is a subunit.

Although the present invention exemplarily illustrates that the motion estimation unit 110 calculates a predicted motion vector (PMV) using motion vectors of blocks located on the left side, the upper side, and the upper right side of the current subunit. However, the present invention is not limited to this. For example, if the motion vector of the block located on the upper right side cannot be obtained, such as the X ₄ block 24, the block located on the upper left side (ie, the X ₁ block) is an alternative to using the motion vector of the block located on the upper right side. It is also possible to use the motion vector of (21). In this case, the predicted motion vector PMV _X2 of the X ₄ block 24 may be obtained through the calculation of Equation 4 below.

Further, although the present invention illustrates that the motion estimation unit 110 calculates a median value of sub-units located in the vicinity, the present invention is not limited thereto. For example, the prediction may be performed through various methods such as an average value calculation and a motion vector competition.

In operation 250, the motion compensator 120 first generates a second macroblock using the predicted motion vectors PMV _X1 , PMV _X2 , PMV _X3 , and PMV _X4 received from the motion estimator 110. In this case, the index for the reference picture required for the motion compensation is received from the controller 180 as previously determined.

In operation 260, the controller 180 calculates a cost (second cost) between the second macroblock and the original macroblock. In this case, the calculated cost is provided to the controller 180.

In step 270, the controller 180 compares the first cost generated in the step 220 and the second cost generated in step 260.

If the second cost value indicates a value smaller than the first cost (270-No), the controller 180 determines that a predetermined condition is satisfied, and the motion vector encoding mode (MVCM) is performed. Is determined as the sub-unit motion vector skip mode. The controller 180 provides only the identifier indicating the motion vector encoding mode (MVCM) (that is, the sub-unit motion vector skipping mode) to the multiplexer 190 without information on a separate motion vector (step 300). Accordingly, the multiplexer 190 generates an encoded stream including only an identifier indicating a sub-motion vector skip mode without information on the motion vector (step 310).

Preferably, the predetermined condition may be a condition in which the predicted motion vector PMV in subunits has the same value as the vector searched in subunits. Correspondingly, the encoding method according to the embodiment of the present invention checks whether the subunit prediction motion vector (PMV) has the same value as the sub-unit searched vector between steps 270 and 300 (280). Step) may be further included (see FIG. 5).

More preferably, the predetermined condition may include the two conditions, and may be a condition in which all of the reference pictures of the sub-prediction motion vectors are the same reference picture. Correspondingly, the encoding method according to the embodiment of the present invention further includes a step (step 290) of checking between step 280 and step 300 whether all reference pictures of the sub-prediction motion vectors are made of the same reference picture. (See FIG. 6).

Steps 280 and 290 may be performed through the controller 180.

On the other hand, if the first cost is relatively less than the second cost in step 220, the controller 180 determines that a predetermined condition is not satisfied. The controller 180 determines a motion vector coding mode (MVCM) as a motion vector prediction mode corresponding to the first cost, and determines the motion vector coding mode (MVCM) (that is, a motion vector prediction mode according to the first cost). In step 320, an identifier corresponding to (a) is generated.

Next, the controller 180 provides the multiplexer 190 with an identifier indicating a motion vector encoding mode (MVCM), and at the same time, the motion vector encoder 170 motion vector corresponding to the first cost and the motion vector. It is instructed to encode and output the information about the reference picture of (step 330), and instructs the redundant data encoder 140 to encode and output the excess data of the reconstructed macroblock based on the motion vector (340). step). Then, the multiplexer 190 generates an encoded stream by combining an identifier indicating the motion vector coding mode (MVCM), a motion vector, information on a reference picture of the motion vector, and surplus data (step 350).

For example, an identifier indicating the motion vector encoding mode (MVCM) may be defined as shown in Table 1 below.

mb_type code No. of bits Name of mb_type 0 "One" One P16 × 16 One "010" 3 P8 × 8 Skip MV 2 "011" 3 P16 × 8 3 "00100" 5 P8 × 16 4 "00101" 5 P8 × 8 5 "00110" 5 P8 × 8 ref0

If mb_type is "0", there is data related to the encoded motion vector, and this indicates that the motion vector encoding mode (MVCM) is a P16x16 prediction mode. In this case, the controller 180 sets the identifier of the motion vector encoding mode (MVCM) to “1” in step 320, and in step 330 to 350, the motion vector, information about the reference picture of the motion vector, and the surplus The data is combined to generate an encoded stream.

If mb_type is "2", there is data related to the encoded motion vector, and this indicates that the motion vector encoding mode (MVCM) is a P16x8 prediction mode. In this case, the controller 180 sets the identifier of the motion vector encoding mode (MVCM) to "011" in step 320, and through steps 330 to 350, information about the motion vector, the reference picture of the motion vector, and the surplus The data is combined to generate an encoded stream.

If mb_type is "3", there is data related to the encoded motion vector, and this indicates that the motion vector encoding mode (MVCM) is a P8x16 prediction mode. In this case, the controller 180 sets the identifier of the motion vector encoding mode (MVCM) to "00100" in step 320, and in step 330 to 350, the motion vector, information on the reference picture of the motion vector, and the surplus The data is combined to generate an encoded stream.

If mb_type is "4", there is data related to the encoded motion vector, and this indicates that the motion vector encoding mode (MVCM) is a P8x8 prediction mode. In this case, the controller 180 sets the identifier of the motion vector encoding mode (MVCM) to "00101" in step 320, and in step 330 to 350, the motion vector, information on the reference picture of the motion vector, and excess The data is combined to generate an encoded stream.

If mb_type is "5", there is data related to the encoded motion vector, where the motion vector encoding mode (MVCM) is P8x8 mode, and the sub-pictures included in the macroblock are stored in the reference picture memory 130. Among the stored pictures, this indicates that the most recently stored picture (eg, a picture having a reference index of 0). In this case, the controller 180 sets the identifier of the motion vector encoding mode (MVCM) to "00110" in step 320, and in step 330 to 350 only the motion vector and the surplus data without information on the reference picture of the motion vector. Combination generates an encoded stream.

On the other hand, when mb_type is "1", it indicates that the motion vector encoding mode (MVCM) is a P8 × 8 unit skip mode. In this case, the controller 180 sets the identifier of the motion vector encoding mode (MVCM) to “010” in step 300, and encodes information about a separate motion vector, reference picture information, and surplus data in step 310. The encoding stream is generated by including only the identifier of the motion vector encoding mode (MVCM), that is, "010".

Therefore, when the motion vector encoding mode (MVCM) is a P8x8 unit skipped mode, encoding is possible only through an identifier of the 3-bit motion vector encoding mode (MVCM), and motion is more precisely set by setting the motion vector in subunits. You can encode a vector.

After all, according to the method of encoding a motion vector of the present invention, when a motion vector of a subunit satisfies a predetermined condition, only one motion vector encoding mode (MVCM) is used without having to encode and transmit motion vector information. Coding is possible. Therefore, the motion vector can be encoded in an accurate unit while minimizing the number of bits of the identifier indicating the encoding mode.

7 is a block diagram illustrating a video decoding apparatus to which a motion information decoding apparatus according to an embodiment of the present invention is applied. Referring to FIG. 7, the video decoding apparatus includes a demultiplexer 410, a motion vector decoder 420, a reference picture memory 430, a motion predictor 440, a motion compensator 450, and an entropy decoder 460. And a redundant data decoding unit 470, an adding unit 475, and a control unit 480.

The demultiplexer 410 parses the encoded bitstream, first obtains an identifier indicating a motion vector encoding mode (MVCM), and provides the identifier to the controller 480. The demultiplexer 410 does not acquire information about a motion vector when the motion vector encoding mode MVCM is a sub-motion vector skip mode. On the other hand, when the motion vector encoding mode (MVCM) is not a sub-motion vector skipping mode, the demultiplexer 410 is a motion vector encoded from a data stream received after the motion vector encoding mode (MVCM), and a reference picture of the motion vector. Obtains information about and encoded redundant data. Information about the encoded motion vector and the reference picture of the motion vector is provided to the motion vector decoder 420, and the encoded surplus data is provided to the entropy decoder 460.

The motion vector decoder 420 decodes the data received from the demultiplexer 410 to restore the motion vector and information on the reference picture of the motion vector.

The motion predictor 440 calculates the predicted motion vector by using the motion vector of the peripheral subunit. The calculation of the predictive motion vector is preferably performed in the same manner as the method of calculating the predictive motion vector in the encoding process. For example, the calculation of the prediction motion vector may be performed through the calculation of Equations 1 to 4 above.

The motion compensator 450 receives a reconstruction unit of the motion vector corresponding to the subunit according to the motion vector encoding mode (MVCM) from the controller 480. The motion vector provided from the motion vector decoder 420 is applied to the reference picture in consideration of the motion vector coding mode (MVCM), and the image information is restored in a unit corresponding to the coding unit.

The entropy decoder 460 receives the compressed bitstream to perform entropy decoding to generate quantized coefficients, and the redundant data decoder 470 performs inverse quantization and inverse transformation on the quantized coefficients to restore the redundant data. do.

The adder 475 finally restores the image information by applying the reconstructed surplus data to the image information reconstructed through motion compensation.

The controller 480 controls the overall operations of the image decoding apparatus by collectively controlling the operations of the functional units. In particular, the controller 480 receives the identifier of the motion vector coding mode (MVCM) from the demultiplexer 410 and checks whether the sub-vector is a motion vector skipping mode.

If the identifier of the motion vector encoding mode (MVCM) is the sub-unit motion vector skipping mode, the controller 480 provides the motion compensation unit 450 with an index and a motion vector for the reference picture according to a predetermined condition. Control the restoration of information.

If the identifier of the motion vector encoding mode (MVCM) is not the sub-unit motion vector skipping mode, the controller 480 controls the demultiplexer 410 so that the motion vector corresponding to the identifier of the motion vector encoding mode (MVCM), Information about the reference picture of the motion vector is provided to the motion vector decoder 420, and surplus data is provided to the entropy decoder 460. The motion vector decoder 420 controls the operation of the encoded motion vector and the reference picture of the motion vector, and controls the operations of the entropy decoder 460 and the redundant data decoder 470. To recover the encoded redundant data. In addition, the motion vector and the reference picture of the decoding unit are provided to the motion compensator 450 to control the reconstruction of the image information.

The decoding apparatus according to the embodiment of the present invention described above may further include an intra compensation unit, a deblocking filter, and the like based on the H.264 / AVC standard. In addition, the redundant data decoder 470, based on the H.264 / AVC standard, in addition to the inverse DCT transform and inverse quantization operation of the redundant data, the inverse DCT transform and inverse quantization operation for a specific picture (eg, I picture) You can also do more. Furthermore, the decoding apparatus according to an embodiment of the present invention may further perform reconstruction of image data encoded based on the H.264 / AVC standard as well as motion vector decoding of the present invention.

Hereinafter, a video reconstruction process to which the motion vector decoding method according to the present invention is applied will be described with reference to the above-described components.

8 is a flowchart illustrating a procedure of a method of decoding motion vector information according to an embodiment of the present invention. Referring to FIG. 8, in operation 510, the demultiplexer 410 receives a bitstream, obtains an identifier indicating a motion vector encoding mode (MVCM), and provides it to the controller 480.

Operation 520 is performed by the controller 480. In operation 520, the controller 480 checks an identifier indicating a motion vector encoding mode (MVCM) and determines whether an identifier indicating a motion vector encoding mode (MVCM) indicates a sub-unit motion vector skipping mode.

On the other hand, the sub-unit motion vector omission mode is determined by a condition previously recognized by the encoder and the decoder in advance. Therefore, in the decoding process, the image of the coding unit is reconstructed by reflecting a predetermined condition. Here, the predetermined condition may be a condition in which the prediction motion vector of the sub-unit is the same value as the vector searched by the sub-unit, and the reference pictures of the prediction motion vectors of the sub-unit are all the same reference picture.

If the identifier indicating the motion vector coding mode (MVCM) indicates the sub-unit motion vector skipping mode (step 520), the motion vector and the reference picture index of the coding unit are restored in consideration of the predetermined condition. The motion compensation is performed by applying the reconstructed motion vector to a reference picture. To this end, the controller 480 controls operations of the motion compensator 450 and the reference picture memory 430 to perform steps 531, 533, and 535. In detail, the controller 480 controls the reference picture memory 430 to provide a reference picture corresponding to the predetermined condition to the motion compensator 450 (operation 531). When the controller 480 instructs the operation of the motion predictor 440, the motion predictor 440 calculates the predicted motion vector by using the motion vector of the neighboring subunit, and then moves to the motion compensator 450. (533). In operation 535, the motion compensator 450 reconstructs an image by generating a motion compensation block of a coding unit by combining the predicted motion vector and the reference picture.

On the other hand, if the identifier indicating the motion vector coding mode (MVCM) does not indicate the sub-unit motion vector skipping mode (step 520-no), the controller 480 performs the motion vector to perform steps 541, 543, and 545. The operations of the decoder 420, the reference picture memory 430, the motion compensator 450, the entropy decoder 460, the redundant data decoder 470, and the adder 475 are controlled.

In detail, in step 541, the controller 480 instructs the operation of the motion vector decoder 420, and the motion vector decoder 420 of the coded motion vector and the motion vector received from the demultiplexer 410. After restoring the information indicating the reference picture, the information is provided to the reference picture memory 430 and the motion compensator 450.

Next, in step 543, the controller 480 instructs the operation of the motion compensator 450 in consideration of a subunit of a size corresponding to the motion vector encoding mode MVCM, and the motion compensator 450 reconstructs the restored motion. The motion vector and the reference picture are combined to perform motion compensation of a coding unit, and to obtain image information.

Next, in step 545, the controller 480 controls the operations of the entropy decoder 460, the redundant data decoder 470, and the adder 475. In response, the entropy decoding unit 460 receives and restores the entropy-encoded surplus data from the demultiplexer 410, and the surplus data decoding unit 470 recovers the restored data to obtain and output the surplus data. The adder 475 combines the surplus data and the motion compensated image information to finally restore the image information of the coding unit.

According to an embodiment of the present invention, as a predetermined condition, a prediction motion vector (PMV) of a subunit of a coding unit is applied to a predetermined reference picture, and the result of the reconstruction is equal to or smaller than the image information of the coding unit. It is illustrated having a cost.

The reference picture in one embodiment of the present invention may be a picture most recently stored in the frame memory. In this case, the predetermined condition may be set such that all reference pictures, which are the reference of the prediction motion vector PMV, are the same as the pictures immediately before the current picture CP in time. Therefore, in the encoding process, if the condition is satisfied, only the information on the satisfaction of the condition is encoded, and in the decoding process, the reference picture required for reconstructing the coding unit itself without information on a separate reference picture is considered in consideration of the condition. Can be obtained.

Also, the reference picture may be any one picture stored in the reference picture memory. In this case, the predetermined condition may be set to be the same as any picture of a picture that is already encoded or decoded as a reference of the prediction motion vector (PMV) of the coding unit. Therefore, in the encoding process, when the condition is satisfied, the information on the satisfaction of the condition and the position of the reference picture are periodically encoded for each predetermined unit (eg, slice unit, picture unit, GOP unit, sequence unit, and the like) and decoded. In consideration of the above condition, the process may be applied to reconstruct a plurality of coding units using a reference picture that is periodically reconstructed.

Furthermore, it is natural that these conditions can be variously changed for efficient encoding and decoding.

The motion vector encoding method according to the present invention can be embodied as computer readable codes on a computer readable recording medium. Computer-readable recording media include all kinds of recording devices that store data that can be read by a computer system. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical disk, and the like, and may also include those implemented in the form of carrier waves (eg, transmission over the Internet). do. The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

Although the present invention has been described above by means of limited embodiments and drawings, the present invention is not limited thereto, and various modifications and changes may be made by those skilled in the art to which the present invention pertains.

Claims (25)

In the method of encoding information about a motion vector of a coding unit, Setting a subunit obtained by dividing a coding unit into at least two regions, and calculating a predictive motion vector (PMV) for a plurality of subunits; Checking whether the plurality of prediction motion vectors (PMVs) satisfy a predetermined condition; When the plurality of prediction motion vectors (PMVs) satisfy the predetermined condition, generating encoded data of the motion vector, including information indicating that the sub-unit motion vector is omitted. A method of encoding motion vector information. The method of claim 1, Estimating a motion vector according to each mode based on a plurality of motion vector prediction modes, Restoring image information of a coding unit by applying the estimated motion vector to a reference picture; Calculating surplus data between the reconstructed image information and original image information, encoding and decoding the surplus data; Generating the reconstructed image information by applying the encoded and decoded surplus data to the reconstructed image; Calculating a cost between the reconstructed image information and the original image information; Selecting a mode having a minimum cost among a plurality of motion vector prediction modes and setting the minimum cost as a first cost; Applying a predictive motion vector (PMV) to a reference picture, reconstructing image information of a coding unit, and calculating a second cost between the reconstructed image information and the original image information, The predetermined condition is A method of encoding motion vector information, wherein the second cost is a condition having a relatively smaller value than the first cost. The method of claim 2, If the predetermined condition is not satisfied, Motion vector information comprising the step of generating encoded data for the motion vector, including the motion vector estimated for the mode having the lowest cost, information about a reference picture, and residual data Encoding method. The method of claim 1, wherein the predetermined condition is A method of encoding motion vector information, characterized in that the sub-unit predicted motion vector (PMV) and the sub-unit searched motion vector have the same value. The method according to claim 4, wherein the predetermined condition is The motion vector information encoding method according to claim 1, wherein the reference pictures of the motion vectors of the subunits included in the coding unit are all made of the same reference picture. 6. The method of claim 5, wherein the reference picture is a picture that is neighbored in time with a picture currently being encoded. The method of claim 5, The reference picture is any one selected from pictures stored in the frame memory. And the encoded data includes information indicating at least one reference picture. The method according to any one of claims 1 to 7, The process of calculating the predicted motion vector (PMV), Obtaining a motion vector for subunits located nearby, And calculating a predicted motion vector (PMV) of the sub-units by calculating a median value of the motion vectors of the sub-units located in the periphery. A method for recovering information about a motion vector from a bitstream including an identification code indicating a unit for obtaining a motion vector, the method comprising: Parsing the identification code included in an input bitstream to obtain a motion vector coding mode (MVCM); When the mode is a mode indicating omission of the sub-unit motion vector, a process of calculating a predictive motion vector (PMV) according to a condition previously promised by an encoder and obtaining information on a reference picture and, And restoring a motion compensation block by applying the predicted motion vector (PMV) to the reference picture. The method of claim 9, When the motion vector encoding mode (MVCM) is not a mode indicating omission of the subunit motion vector, Restoring the encoded motion vector and the encoded residual data and confirming information about the reference picture; And reconstructing image information of a coding unit by applying the motion vector and the surplus data to a reference picture. The method of claim 9, wherein the process of calculating the predicted motion vector (PMV), Obtaining a motion vector for subunits located nearby, And calculating a median value of the motion vector for the sub-units located in the periphery, and setting the median value as the predictive motion vector (PMV). 10. The method of claim 9, wherein the reference picture is a picture neighboring in time with a picture currently being encoded. An apparatus for encoding information on a motion vector of a coding unit by using a motion estimation unit and a motion compensation unit, Instructs the search of the motion vector and the operation of the sub-predictive motion vector (PMV) for a plurality of motion vector prediction modes (MVPM), calculates the cost of the reconstructed coding unit, and A controller for controlling the operation of the motion estimator and the motion compensator; In response to the motion vector prediction mode (MVPM) input by the controller, the motion vector of each of the motion vector prediction modes (MVPM) is searched and provided to the motion compensator, and the sub-units are searched using neighboring motion vectors. A motion estimation unit for calculating a predicted motion vector (PMV) and then providing the motion compensation unit to the motion compensation unit; It includes a motion compensation unit for restoring the image information of the coding unit by applying the motion vector received from the motion estimation unit to the reference picture, The control unit, When the predetermined condition is satisfied, the motion information encoding apparatus includes information indicating that the motion vector for the coding unit is omitted in the encoded data. The method of claim 13, wherein the control unit, If the predetermined condition is not satisfied, Motion information, characterized by generating coded data for a motion vector, including a motion vector prediction mode (MVPM) having a minimum cost, a motion vector searched based on the mode (MVPM), and residual data. Encoding device. The method of claim 13, wherein the predetermined condition is The cost for the subunit predicted motion vector (PMV) has a value that is relatively smaller than the minimum cost computed based on the motion vector prediction mode (MVPM), The motion information encoding apparatus of claim 1, wherein the prediction motion vector (PMV) of the sub-unit and the motion vector obtained through the search of the sub-unit are made of the same value. The method of claim 15, wherein the predetermined condition is A motion information encoding apparatus, characterized in that all reference pictures of sub-unit motion vectors are composed of the same reference picture. 17. The motion information encoding apparatus of claim 16, wherein the reference picture is a picture neighboring in time with a picture currently being encoded. The method of claim 14, The reference picture is any one selected from pictures stored in the frame memory. And the coded data includes information indicating at least one reference picture. The method of claim 13, A surplus data encoder for encoding surplus data between original image information and reconstructed image information in a coding unit; A redundant data decoding unit for restoring the encoded redundant data is further included. The control unit, And calculating a cost using the number of bits of surplus data provided from the surplus data encoder and a distortion value provided from the surplus data decoder. The method according to any one of claims 13 to 19, The motion estimator, Obtain a motion vector for the surrounding subunits, And a predicted motion vector of the subunits by calculating a median value of the motion vectors of the subunits located in the periphery. An apparatus for recovering information about a motion vector from a bitstream, A controller for acquiring a motion vector coding mode (MVCM) from an input bitstream, controlling a reconstruction of a motion vector corresponding to the mode (MVCM) and reconstruction of image information of a coding unit; A motion vector decoder configured to reconstruct motion information including an encoded motion vector and a reference picture with reference to the mode (MVCM) and a predetermined condition provided from the controller; A motion vector predictor configured to calculate a predicted motion vector (PMV) of a sub-unit by referring to the mode (MVCM) and an encoder previously promised from the controller; A motion compensator for restoring image information by applying the motion vector to the reference picture; A surplus data decoding unit for restoring the encoded surplus data; And an adder which adds the image information output from the motion compensator and the surplus data from the surplus data decoder. The method of claim 21, When the mode (MVCM) is a mode indicating omission of the subunit motion vector, And the control unit controls the motion vector predictor according to a condition previously promised by an encoder and obtains information about a reference picture. The method of claim 21, wherein the motion vector predictor, Obtain a motion vector for the surrounding subunits, And calculating the median of the obtained motion vectors and setting the median as the predicted motion vector (PMV) of the sub-unit. 24. The motion information decoding apparatus of claim 21 or 23, wherein the reference picture is a picture neighboring in time with a picture currently being encoded. The method of claim 21, If the mode is not a mode indicating omission of the subunit motion vector, The control unit controls the motion vector decoder and the redundant data decoder according to a condition previously agreed with an encoder. The motion vector decoder obtains a motion vector and a reference picture from input data, The redundant data decoding unit obtains redundant data with reference to the input bitstream.