KR20070055903A - Method and apparatus for prediction of adaptive motion vector in multi-viewpoint image - Google Patents

Abstract

The present invention relates to a method and apparatus for adaptively predicting a motion vector for encoding multi-viewpoint video data photographed by a plurality of cameras having different viewpoints. A block detector which detects only blocks that refer to the same block as the block to which the current block refers temporally or spatially among adjacent blocks of the block; And an adaptive predictor that obtains a predicted motion vector value of the current block based on the motion vector of the block detected by the block detector to adaptively predict the motion vector to improve the compression ratio.

Multiview image, H.264, motion vector prediction

Description

Apparatus and method for adaptive motion vector prediction in multiview images {Method and Apparatus for prediction of adaptive motion vector in multi-viewpoint image}

FIG. 1A illustrates an internal configuration diagram of an adaptive motion vector prediction apparatus in a multiview video.

1 (b) shows an internal detailed view of the adaptive motion vector prediction apparatus in a multiview video.

2 (a) to (c) show a motion vector prediction method in the H.264 standard.

3 (a) and (b) illustrate a motion vector prediction method and a calculation process when a reference frame of a current block and one adjacent block is a previous spatial frame at the same time.

4 (a) and (b) illustrate a motion vector prediction method and a calculation process when a reference frame of a current block and one neighboring block is the same spatial frame at a previous time.

5 (a) and (b) illustrate a motion vector prediction method and a calculation process when a reference frame of a current block and two adjacent blocks is a previous spatial frame at the same time.

6 (a) and 6 (b) illustrate a motion vector prediction method and a calculation process when a reference frame of a current block and two adjacent blocks is the same spatial frame at a previous time.

7 (a) and (b) illustrate a motion vector prediction method and a calculation process when reference frames of all blocks adjacent to the current block are previous spatial frames of the same time.

8 (a) and (b) illustrate a motion vector prediction method and a calculation process when a reference frame of a current block and two adjacent blocks is the same spatial frame at a previous time.

9 (a) and 9 (b) show a motion vector prediction method and a calculation process when a reference frame of a current block is a coded frame of a previous space and all reference frames of adjacent blocks are frames encoded at a previous time.

10 (a) and (b) show a motion vector prediction method and a calculation process when a reference frame of a current block is a frame encoded at a previous time and all the reference frames of adjacent blocks are encoded frames in a different space from the current space. Illustrated.

11 illustrates an encoder based on an adaptive motion prediction technique as a preferred embodiment of the present invention.

12 illustrates a decoder based on an adaptive motion prediction technique as a preferred embodiment of the present invention.

FIG. 13 is a flowchart illustrating a motion vector prediction process in a multiview image, according to an embodiment of the present invention.

14 is a flowchart illustrating a motion vector prediction process of a multiview image in an encoder according to an embodiment of the present invention.

15 is a flowchart illustrating a motion vector prediction process of a multiview image in a decoder according to an embodiment of the present invention.

The present invention relates to an apparatus and method for adaptive motion vector prediction in multi-viewpoint video data captured by a plurality of cameras having different viewpoints, and more particularly, to a motion vector for a block having a predetermined size. The present invention relates to a method and apparatus for increasing the compression ratio by adaptively predicting a motion vector based on three neighboring blocks already encoded and decoded.

Currently, according to the H.264 standard used for encoding video data, one frame is divided into blocks having a predetermined size, and a motion search is performed to search for the most similar block by referring to an adjacent frame that has already been encoded.

However, even when the reference frames of the neighboring blocks are all different from the current reference frame, or even when the reference frames of the two blocks are the same as the current reference frame, the median prediction is used to use the value as the motion vector prediction value of the current block. It may be unsuitable for obtaining high compression rates.

In order to solve the above inconsistency, the present invention provides a multi-view video that increases the compression rate by adaptively using the motion vector prediction method according to the spatial frame and the temporal frame referenced by the neighboring block in predicting the motion vector value of the current block. The present invention relates to a method and apparatus for adaptive motion vector prediction.

In a preferred embodiment of the present invention, the adaptive motion vector prediction apparatus in the multi-view image is a block detector for detecting only blocks that refer to the same block as the block to which the current block refers to temporally or spatially among the adjacent blocks of the current block ; And an adaptive predictor for obtaining a predicted motion vector value of the current block based on the motion vector of the block detected by the block detector.

In the adaptive predicting unit of the present invention, when the number of detected blocks is one, the predicted motion vector value of the current block is a motion vector value of the detected block.

In the adaptive prediction unit of the preferred embodiment of the present invention, when the number of detected blocks is two, the predicted motion vector value of the current block is an average value of the motion vectors of the detected blocks.

In the adaptive prediction unit of the preferred embodiment of the present invention, when the number of detected blocks is three, the predicted motion vector value of the current block is an intermediate value of the motion vectors of the detected blocks.

When a block is not detected by the adaptive predictor of the present invention, the predicted motion vector value of the current block is (0,0).

In one preferred embodiment of the invention the motion vector is estimated based on the H.264 standard.

In another preferred embodiment of the present invention, the adaptive motion vector prediction encoder detects an adjacent block that refers to the same block as the block that the current block refers to temporally or spatially and based on the detected motion vector of the block. A motion vector predictor for obtaining a predicted motion vector value of the current block; And an entropy encoder that encodes a difference between the predicted motion vector value of the current block obtained by the motion vector predictor from the motion vector value of the current block.

In another preferred embodiment of the present invention, the decoder for decoding the blocks of the encoded multiview video bit stream by using the adaptive motion vector prediction method subtracts the prediction motion vector value of the predetermined block from the motion vector value of the predetermined block. A receiver configured to receive a multiview video bitstream including a motion vector difference value and motion vector information; An entropy decoding unit to decode blocks in the received bit stream using the motion vector information and the motion vector difference value; Among the blocks in the decoded bit stream, a neighboring block that refers to the same block as the block to which the decoded current block refers temporally or spatially is detected, and a predicted motion vector value of the current block based on the detected motion vector of the neighboring block. A motion vector predictor for obtaining a; And a motion vector reconstruction unit reconstructing the motion vector by adding the predicted motion vector value to the motion vector difference value.

In another preferred embodiment of the present invention, the adaptive motion vector prediction method in a multiview image comprises the steps of: (a) detecting a block to which the current block refers temporally or spatially; (b) detecting a block referring to the block detected in step (a) of the block adjacent to the current block; And (c) obtaining a predicted motion vector value of the current block based on the motion vector of the block detected in step (b).

In another preferred embodiment of the present invention, the encoding method based on adaptive motion vector prediction comprises: (a) detecting an adjacent block that refers to the same block as the block to which the current block refers temporally or spatially; Obtaining a predicted motion vector value of the current block based on a motion vector of? and (b) encoding a difference between the predicted motion vector value of the current block obtained by the motion vector predictor from the motion vector value of the current block.

In another preferred embodiment of the present invention, a method of decoding blocks of an encoded multiview video bit stream based on adaptive motion vector prediction includes (a) a predicted motion vector value of the predetermined block from a motion vector value of a predetermined block. Receiving a multiview video bitstream including a motion vector differential value and motion vector information obtained by subtracting the motion vector; (b) decoding the blocks in the received multi-view video bit stream using the motion vector information and the motion vector difference value; (c) detecting a neighboring block of the blocks in the decoded bit stream that refer to the same block as the block that the decoded current block refers to temporally or spatially and predicting the current block based on the detected motion vector of the neighboring block; Obtaining a motion vector value; And (d) reconstructing the motion vector by adding the predicted motion vector value to the motion vector difference value.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings. It should be noted that the same elements among the drawings are denoted by the same reference numerals and symbols as much as possible even though they are shown in different drawings. In the following description of the present invention, if it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.

1 (a) and (b) show an internal configuration diagram of an adaptive motion vector prediction apparatus in a multiview video.

As shown in FIG. 1 (a), as an exemplary embodiment of the present invention, the adaptive motion vector prediction apparatus 100 in a multiview video includes a block detector 110 and an adaptive predictor 120.

The adaptive motion vector prediction apparatus 100 is based on obtaining a motion vector by the motion vector prediction method of the H.264 standard. In addition, the adaptive motion vector prediction apparatus 100 is based on the space and time referred to by a neighboring block to predict the motion vector of the current block. The motion vector prediction method is adaptively used.

In the multi-view video encoding, when encoding the current frame, the encoding efficiency is improved by referring to all frames of another space previously encoded or frames of another time encoded previously.

If the reference frame of the block to be currently encoded and the reference frame of adjacent blocks have different spaces or time, the block detector 110 predicts an incorrect value when the motion vector of the current block is predicted using the motion vector of the adjacent block. In order to prevent such a prediction error, only blocks that refer to the same block as the block to which the current block refers temporally or spatially among neighboring blocks of the current block are detected.

The adaptive predictor 120 obtains a predicted motion vector value of the current block based on the motion vector of the block detected by the block detector 110. That is, the predicted motion vector value of the current block is obtained based on the number of blocks detected by the block detector 110 and the motion vectors of the detected blocks. This will be described in detail with reference to FIG. 1B.

 As an embodiment of implementing an adaptive motion vector prediction apparatus in a multiview video, the block detectors 110 and 130 may perform a current block grasp 131, a temporal vector estimator 132, and a spatial vector estimator 133. (FIG. 1B).

The current block identifying unit 131 detects a block referenced by the current block, and determines whether the vector referring to the reference block is a time vector based on space or a space vector based on space.

The time vector estimator 132 refers to a block that refers to the same reference block as the reference block of the current block detected by the block detector 110 when the motion vector of the current block corresponds to a time vector based on a temporal difference from the reference block. Among the motion vectors, a block corresponding to the time vector is identified. 4, 6 and 8 as a specific example.

The spatial vector estimator 133 refers to a block referring to the same reference block as the reference block of the current block detected by the block detector 110 when the motion vector of the current block corresponds to a spatial vector based on a spatial difference from the reference block. The block in which the motion vector corresponds to the space vector is identified. 3, 5, and 7 as a specific embodiment.

The adaptive predictors 120 and 140 calculate predicted motion vector values based on the number of blocks detected by the temporal vector estimator 132 or the spatial vector estimator 133 and the motion vectors of the identified blocks.

When there is one adjacent block having the same reference block as the current block, the adaptive predictor 140 uses the time vector as a predicted motion vector value of the current block when the detected motion vector of the adjacent block is a time vector. When the detected motion vector of one adjacent block is a spatial vector, the spatial vector is used as a predicted motion vector value of the current block. This will be described in more detail with reference to FIGS. 3 (a), (b) and 4 (a) and (b).

When there are two adjacent blocks having the same reference block as the current block, the adaptive predictor 140 uses the detected motion vector average of the two blocks as the predicted motion vector value of the current block. More specifically, the motion vector of two blocks detected when the motion vector with the reference block of the current block is a spatial vector, and the motion of the two blocks detected when the motion vector with the reference block of the current block is a time vector. Each of the temporal vector means is obtained as the predicted motion vector value of the current block. This will be described in more detail with reference to FIGS. 5 (a), (b) and 6 (a) and (b).

When there are three adjacent blocks having the same reference block as the current block, that is, the entire adjacent blocks, the adaptive predictor 140 uses the median motion vector values of the three blocks as the predicted motion vector values of the current block. . More specifically, the median motion vector of the three blocks detected when the motion vector with the reference block of the current block is a spatial vector, and the motion vector of the three blocks detected when the motion vector with the reference block of the current block is a time vector. Each of the median motion time vector values is used as the predicted motion vector value of the current block. This will be described in more detail with reference to FIGS. 7 (a), (b) and 8 (a) and (b).

When there is no adjacent block having the same reference block as the current block, the adaptive predictor 140 may more specifically refer to the reference frame of the current block as an encoded frame of a previous space and all the reference frames of the adjacent block are previous time. And a reference frame of the current block is a frame encoded at a previous time, and a reference frame of a neighboring block is a coded frame of a different space from the current space. Use it as a value. This will be described in more detail with reference to FIGS. 9 (a), (b) and 10 (a) and (b).

2 (a) to (c) show a motion vector prediction method in the H.264 standard.

According to the H.264 standard used for encoding video data, a frame is divided into blocks having a predetermined size, and a motion search is performed to search for the most similar block by referring to an adjacent frame that has already been encoded. That is, a median value among the motion vectors of the three macroblocks on the left side, the upper side, and the upper right side of the current macroblock is determined as the predicted value of the motion vector.

After that, the current motion block is represented by the optimal motion vector obtained from the motion search to increase the compression ratio, and before the motion search is performed, the motion vector prediction is performed to minimize and encode the difference between the actual motion vector and the predicted motion vector.

More specifically, Fig. 2 (a) shows a motion vector median prediction method in the H.264 standard. If neighboring blocks A, B, and C are the same slice, the motion vector median of the blocks is used as the motion vector prediction value of the current block.

To do this, first check whether the surrounding blocks are equal to the current reference frame. If only one of the three neighboring blocks A, B, or C refers to the same reference frame, the motion vector of the block having the same reference frame is used as the motion vector prediction value of the current block without performing intermediate value prediction. Median prediction is performed on the motion vectors of three blocks A, B, and C and used as the motion vector prediction value of the current block.

2 (b) and 2 (c) show a motion vector prediction method when the block sizes are 16x8 and 8x16, respectively, in the H.264 standard.

In FIG. 2B, when the size of the current block E is 16 × 8 (220, 230), when the reference frame of the current block E1 220 and the reference frame of the block B are the same, the motion vector of the block B is converted to the current block E1 220. It is used as the motion vector prediction value of, otherwise the median prediction method is used.

In addition, when the reference frame of the current block E2 230 is the same as the reference frame of the block A, the motion vector of the block A is used as the motion vector prediction value of the current block E2. Otherwise, the median prediction method is used.

If the current block E has a size of 8x16 (240, 250) in FIG. 2 (c), when the reference frame of the block E3 240 is the same as the reference frame of the block A, the motion vector of the block A is changed to the current block E3. Use it as a motion vector prediction value, otherwise use the median prediction method.

If the reference frame of the current block E4 is the same as the reference frame of the block C, the motion vector of the block C is used as the motion vector prediction value of the E4 block. Otherwise, the median prediction method is used.

3 to 10 illustrate a method and method for obtaining a motion prediction vector from a multiview image based on the motion prediction method of the H.264 standard as an embodiment of the present invention.

Prior to the detailed description, the dotted lines of FIG. 3 to 10 denote temporal vectors based on temporal differences between the current block and the reference block, and the solid lines denote spatial vectors based on the spatial difference between the current block and the reference block. .

3 (a) and (b) illustrate a motion vector prediction method and a calculation process when a reference frame of a current block and one adjacent block is a previous spatial frame at the same time.

3 (a) shows that when the reference frames of the current block 301 and the adjacent block B 302 are previous spatial frames at the same time, the motion vector is a spatial vector, and in this case, the motion vector is moved to obtain a predicted motion vector value of the current block. A spatial motion vector of block B that references the same time frame in space is used.

3 (b) shows the difference between the actual spatial motion vector and the predicted spatial motion vector of the current block at the encoder side using the adaptive motion vector prediction method when the reference frame of the current block and the adjacent block B is the same time frame of the previous space. The calculation process is shown.

In this case, the optimal spatial motion vector DV = (4, 3) through motion search and the spatial motion vector PDV = (5, 3) (previous space) obtained by the predictive motion vector value of the current block by the adaptive predictor 140 The difference DVD = (−1, 0) between the spatial motion vectors of block B referencing the same time frame is used in the encoder.

4 (a) and (b) illustrate a motion vector prediction method and a calculation process when a reference frame of a current block and one neighboring block is the same spatial frame at a previous time.

In FIG. 4 (a), when the reference frame of the current block and one adjacent block is the same space frame of the previous time (402), block A (reference to the frame encoded in the previous space of the same space to obtain a motion vector prediction value) Using the motion vector of 401).

4 (b) shows the actual time motion vector and predicted time of the current block at the encoder side using the adaptive motion vector prediction method when the reference frame of the current block and the adjacent block A is a frame encoded in the same space of the previous time. Represents a calculation process for finding the difference between motion vectors.

In this case, the optimal temporal motion vector MV = (1, 4) through motion search and the temporal motion vector PMV = (0, 3) (previous time) obtained by the adaptive predictor 140 as the predicted motion vector value of the current block. The difference MVD = (1, 1) between the motion vectors of block A referring to the frame encoded by < RTI ID = 0.0 >

5 (a) and (b) illustrate a motion vector prediction method and a calculation process when a reference frame of a current block and two adjacent blocks is a previous spatial frame at the same time.

In FIG. 5A, when the reference frames of blocks A and B of the current block and adjacent blocks are the same time frame in the previous space (501, 502, 503), the spatial motion vector prediction value refers to the same time frame in the previous space. The average of the spatial motion vectors of blocks A and B (501, 503) is used as the value.

FIG. 5 (b) shows the actual spatial motion vector and the predicted spatial motion vector at the encoder side using the adaptive motion vector prediction method when the reference frames of blocks A and B of the current block and neighboring blocks are the same time frame of the previous space. It shows the process of finding the difference between them.

In this case, the optimal spatial motion vector DV = (4, 1) through motion search and the spatial motion vector PDV = (4, 1) (previous space) obtained by the predictive motion vector value of the current block by the adaptive predictor 140 We use the difference DVD = (0, 0) between the spatial motion vector average values of blocks A and B that refer to the same time frame.

6 (a) and 6 (b) illustrate a motion vector prediction method and a calculation process when a reference frame of a current block and two adjacent blocks is the same spatial frame at a previous time.

In FIG. 6 (a), when the reference frames of blocks B and C of the current block and the adjacent blocks are the same spatial frames of the previous time (601, 602, 603), the previously encoded frame of the current view is referred to to obtain a motion vector prediction value. The motion vector average of blocks B, C (602, 603) is used.

FIG. 6 (b) shows an example of an operation between an actual temporal motion vector and a predicted temporal motion vector at an encoder using an adaptive motion vector prediction method when reference frames of blocks B and C of the current block and neighboring blocks are the same spatial frame of a previous time. It shows the process of finding the difference.

In this case, the optimal time motion vector MV = (3, 3) through motion search and the motion time vector PMV = (2, 3) (previous time) obtained by the predictive motion vector value of the current block by the adaptive prediction unit 140. Difference MVD = (1, 0) between the motion vector average values of blocks B and C that refer to the same spatial frame.

7 (a) and (b) illustrate a motion vector prediction method and a calculation process when reference frames of all blocks adjacent to the current block are previous spatial frames of the same time.

In FIG. 7 (a), when the reference frames of all blocks adjacent to the current block are previous spatial frames of the same time (701, 702, 703, 704), the spaces of the neighboring blocks (702, 703, 704) are obtained to obtain a motion prediction value. Use the median motion vector.

FIG. 7B illustrates a process of obtaining a difference between an actual spatial motion vector and a predicted spatial motion vector when reference frames of all blocks adjacent to the current block are previous spatial frames at the same time.

In this case, the spatial motion obtained by the optimal motion vector DV = (8,3) through the motion search on the encoder side using the adaptive motion vector prediction method and the predicted motion vector value of the current block by the adaptive predictor 140 Difference between vector PDV = (6, 2) (median spatial motion vector of adjacent blocks) or PDV = (7, 3) (mean value of spatial motion vector of adjacent blocks) DVD = (1, 2) or DVD = (2 , 2).

8 (a) and (b) illustrate a motion vector prediction method and a calculation process when a reference frame of a current block and two adjacent blocks is the same spatial frame at a previous time.

In FIG. 8 (a), when the reference frames of all blocks adjacent to the current block are the same spatial frames of the previous time (801, 802, 803, 804), the neighboring blocks (802, 803, 804) of the neighboring blocks to obtain a motion prediction value are obtained. Use the median motion vector.

FIG. 8 (b) shows a process of obtaining a difference between the actual motion vector and the predicted motion vector when the reference frames of all blocks adjacent to the current block are the same spatial frames of the previous time.

In this case, the motion obtained by using the optimal motion vector MV = (4, 3) through the motion search on the encoder side using the adaptive motion vector prediction method and the predicted motion vector value of the current block by the adaptive predictor 140 Difference between time vector PMV = (3, 1) (median motion time vector of neighboring blocks) or PMV = (2, 1) (mean value of motion time vector of neighboring blocks) MVD = (1, 2) or MVD = ( 2, 2).

9 (a) and 9 (b) show a motion vector prediction method and a calculation process when a reference frame of a current block is a coded frame of a previous space and all reference frames of adjacent blocks are frames encoded at a previous time.

In FIG. 9 (a), when a reference frame of a current block is an encoded frame of a previous space and all of the reference frames of adjacent blocks are frames encoded at a previous time, a neighboring block is required to perform displacement motion vector prediction of the current block. Since no information is available, the displacement motion vector prediction value uses the vector (0,0) as its value.

FIG. 9 (b) shows an optimal motion vector DV through motion search at an encoder using an adaptive motion vector prediction method when the previous frames are encoded frames and all reference frames of adjacent blocks are frames encoded at a previous time. = Difference DVD = (3, 3) between the spatial motion vector PDV = (0,0) ((0,0) vector) obtained by the predictive motion vector value of the current block by the adaptive predictor 140 ).

10 (a) and (b) show a motion vector prediction method and a calculation process when a reference frame of a current block is a frame encoded at a previous time and all the reference frames of adjacent blocks are encoded frames in a different space from the current space. Illustrated.

In FIG. 10 (a), when the reference frame of the current block is a frame encoded at a previous time and all the reference frames of the neighboring blocks are encoded frames in a different space from the current space, the motion vector prediction of the current block is performed in the neighboring block. The motion vector prediction value uses the (0,0) vector as its value because it cannot obtain the information necessary to do so.

FIG. 10 (b) shows that the frame is encoded at the previous time and the reference frames of the neighboring blocks are all encoded frames in a space different from the current space. The difference between the motion vector MV = (2, 1) and the vector PMV = (0,0) ((0,0) vector) obtained by the adaptive predictor 140 as the predicted motion vector value of the current block MVD = (2, 1) is used in the encoder.

11 illustrates an encoder based on an adaptive motion prediction technique as a preferred embodiment of the present invention.

An encoder based on the adaptive motion prediction technique includes a motion vector predictor 1110 and an entropy encoder 1120.

The motion vector predictor 1110 detects an adjacent block that refers to the same block as the block that the current block refers to temporally or spatially, and obtains a predicted motion vector value of the current block based on the detected motion vector of the block. In addition, the motion vector predictor 1110 includes a block detector 1111 and an adaptive predictor 1112, and a description thereof will be similar to the related description in FIG. 1.

The entropy encoder 1120 predicts the current block obtained by the adaptive motion vector prediction of the present invention from the motion vector predictor 1110 in the motion vector (MV) s1122 of the current block found through the motion prediction. The motion vector diffrerence (MVD) of the motion vector value PMV (Prediction Motion Vector) s1121 is encoded, and the difference values obtained in FIGS. 3 (b) to 10 (b) are encoded.

In the encoder, the entropy encoder 1120 obtains a predicted motion vector value using the adaptive motion vector prediction method of the present invention, and as the prediction degree becomes more accurate, the difference between the predicted motion vector value of the current block in the current vector MVD = MV-PMV) is reduced to reduce the number of bits to be used for encoding.

12 illustrates a decoder based on an adaptive motion prediction technique as a preferred embodiment of the present invention.

A decoder for decoding the blocks of the encoded multiview video bit stream using the adaptive motion vector prediction method includes a receiver (not shown), an entropy decoder 1210, a motion vector predictor 1230, and a motion vector reconstructor 1240. ).

The receiver may generate a multi-view video bitstream generated by an encoder including motion vector information and a motion vector difference value that is a difference (MVD = MV-PMV) of the predicted motion vector value of the predetermined block from the motion vector of the predetermined block in the encoder. Receive.

The entropy decoding unit 1210 receives a multiview video bitstream generated by an encoder and first decodes an intra frame using the motion vector information and the motion vector difference value of blocks in the received bit stream. Entropy decodes Inter frames.

The motion vector predictor 1230 detects an adjacent block that refers to the same block as the block that the decoded current block refers to temporally or spatially among the blocks in the bit stream decoded by the entropy decoding unit 1210, and detects the detected neighboring block. The predicted motion vector value of the current block is obtained based on the motion vector of. The technical idea of obtaining the predicted motion vector value in the motion vector predictor 1230 is similar to the technical idea described with reference to FIG. 1, and a detailed description thereof will be described with reference to FIG. 1.

The motion vector reconstruction unit 1240 entropy decoding unit 1210 entropy decodes the inter frame to add the motion vector difference MVD and the predicted motion vector value PMV of the block obtained by the motion vector prediction unit 1230. Restore the motion vector.

FIG. 13 is a flowchart illustrating a motion vector prediction process in a multiview image, according to an embodiment of the present invention.

In order to predict the motion vector in the multiview image, an adjacent block referring to the reference block of the current block is detected (S1310). It is determined whether the motion vector of the current block is a time-based vector or a space-based vector (S1320). Among the detected blocks, blocks in which the motion vector is the same as the vector type of the motion vector of the current block are detected (S1330 and S1340).

Embodiments when the motion vector of the current block is a time-based vector are shown in FIGS. 4, 6, 8, and 10, and embodiments of the space-based vector of the current block are shown in 3, 5, 7, and 9 It is.

The predicted motion vector value is adaptively predicted based on the detected time-based or space-based motion vector. A predicted motion vector value of the current block is detected based on the type of the motion vector and the number of detected blocks. For this, refer to the description of the adaptive predictor 140 of FIG. 1.

14 is a flowchart illustrating a motion vector prediction process of a multiview image in an encoder according to an embodiment of the present invention.

In order to encode a motion vector based on an adaptive prediction method in a multiview image, a block referenced by a current block in time or space is detected. Thereafter, a block referring to the block referenced by the current block among the blocks adjacent to the current block is detected (S1410). Herein, an adjacent block of the current block means a block on the left side, upper side and right upper side of the current block.

Thereafter, a predicted motion vector value of the current block is obtained based on the detected motion vector of the block (S1430). In this case, the method of obtaining the predicted motion vector value of the current block is similar to the method of obtaining the predicted motion vector value in the adaptive predictor 140 of FIG. 1 in the technical concept.

Thereafter, a motion vector difference value that is a difference (MVD = MV-PMV) of the predicted motion vector value of the current block from the motion vector of the current block is obtained, and the difference value is encoded (S1430 and S1440).

15 is a flowchart illustrating a motion vector prediction process of a multiview image in a decoder according to an embodiment of the present invention.

In order to decode the blocks of the encoded multiview video bitstream based on adaptive motion vector prediction, the encoded multiview video bitstream is received (S1510). In this case, the multi-view video bitstream includes a motion vector difference value and motion vector information obtained by subtracting the predicted motion vector value of the predetermined block from the motion vector value of the predetermined block.

Thereafter, the blocks in the received multiview video bit stream are decoded using the motion vector information and the motion vector difference value (S1520).

Among the blocks in the next decoded bit stream, a neighboring block is detected that refers to the same block as the block to which the decoded current block refers temporally or spatially, and a predicted motion vector value of the current block is based on the detected motion vector of the neighboring block. In operation S1530, the motion vector is reconstructed by adding the motion vector difference MVD and the predicted motion vector value PMV of the block obtained by the motion vector predictor 1230 (S1540).

The invention can also be embodied as computer readable code on a computer readable recording medium. Computer-readable recording media include all types 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 data storage, and the like, which are also implemented in the form of carrier waves (for example, transmission over the Internet). It also includes. 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.

The best embodiments have been disclosed in the drawings and specification above. Although specific terms have been used herein, they are used only for the purpose of describing the present invention and are not used to limit the scope of the present invention as defined in the meaning or claims.

Therefore, those skilled in the art will understand that various modifications and equivalent other embodiments are possible from this. Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

In the present invention, in order to increase the accuracy of the motion vector prediction, the compression rate is increased by reducing the motion vector difference using only blocks related to the current frame among neighboring blocks.

In addition, since the motion vector prediction method of the present invention is determined according to the state of neighboring blocks, the motion vector prediction method does not have any influence on the existing syntax, and only slightly corrects the semantics of the motion vector prediction method in the encoding and decoding steps. There is an effect that can be achieved.

Accordingly, a higher compression rate can be obtained than the conventional coding method by increasing the efficiency of entropy encoding by reducing the difference between the actual motion vector and the predicted motion vector by performing more accurate motion vector prediction.

Claims (36)

A block detector which detects only blocks that refer to the same block as the block to which the current block refers temporally or spatially among adjacent blocks of the current block; And And an adaptive predictor for obtaining a predicted motion vector value of the current block based on the motion vector of the block detected by the block detector. The method of claim 1, wherein the adaptive prediction unit The predicted motion vector value of the current block is the motion vector value of the detected block when the number of detected blocks is one. The method of claim 1, wherein the adaptive prediction unit 2. The apparatus of claim 2, wherein the predicted motion vector value of the current block is an average value of the motion vectors of the detected blocks when the number of detected blocks is two. The method of claim 1, wherein the adaptive prediction unit 3. The apparatus of claim 3, wherein the predicted motion vector value of the current block is an intermediate value of the motion vectors of the detected blocks when the number of detected blocks is three. The method of claim 1, wherein the adaptive prediction unit The adaptive motion vector prediction apparatus of the multiview video, wherein the predicted motion vector value of the current block is (0,0) when the block is not detected. The method of claim 1, The motion vector is adaptive motion vector prediction apparatus in a multi-view video, characterized in that estimated based on the H.264 standard. A motion vector predictor detecting a neighboring block referring to the same block as the block to which the current block refers temporally or spatially and obtaining a predicted motion vector value of the current block based on the detected motion vector of the block; And an entropy encoder which encodes a difference between the predicted motion vector value of the current block obtained by the motion vector predictor from the motion vector value of the current block. The method of claim 7, wherein the motion vector predictor A block detector which detects only a block among the left block, the upper block, and the right upper block of the current block that refers to the same block as the block to which the current block refers temporally or spatially; And And an adaptive predictor for obtaining a predicted motion vector value of the current block based on the motion vector of the block detected by the block detector. The method of claim 7, wherein the motion vector predictor And the predicted motion vector value of the current block is the motion vector value of the detected block when the number of detected blocks is one. The method of claim 7, wherein the motion vector predictor The adaptive motion vector prediction encoder of claim 2, wherein the predicted motion vector value of the current block is an average value of the motion vectors of the detected blocks when the number of detected blocks is two. The method of claim 7, wherein the motion vector predictor And wherein the predicted motion vector value of the current block is the median of the motion vectors of the detected blocks when the number of detected blocks is three. The method of claim 7, wherein the motion vector predictor Adaptive motion vector prediction encoder, characterized in that the predicted motion vector value of the current block is (0,0) when no block is detected. A decoder for decoding blocks of an encoded multiview video bit stream using an adaptive motion vector prediction method, A receiver configured to receive a multiview video bitstream including a motion vector difference value and motion vector information obtained by subtracting the predicted motion vector value of the predetermined block from the motion vector value of a predetermined block; An entropy decoding unit to decode blocks in the received bit stream using the motion vector information and the motion vector difference value; Among the blocks in the decoded bit stream, a neighboring block that refers to the same block as the block to which the decoded current block refers temporally or spatially is detected, and a predicted motion vector value of the current block based on the detected motion vector of the neighboring block. A motion vector predictor for obtaining a; And And a motion vector reconstruction unit configured to reconstruct a motion vector by adding the predicted motion vector value to the motion vector difference value. The method of claim 13, wherein the motion vector predictor A block detector which detects only a block among the left block, the upper block, and the right upper block of the current block that refer to the same block as the block to which the current block refers temporally or spatially; And And an adaptive predictor for obtaining a predicted motion vector value of the current block based on the motion vector of the block detected by the block detector. The method of claim 13, wherein the motion vector predictor And if the number of detected blocks is one, the predicted motion vector value of the current block is the motion vector value of the detected block. The method of claim 13, wherein the motion vector predictor The adaptive motion vector prediction decoder of claim 2, wherein the predicted motion vector value of the current block is an average value of the motion vectors of the detected blocks when the number of detected blocks is two. The method of claim 13, wherein the motion vector predictor 3. The adaptive motion vector prediction decoder of claim 3, wherein the predicted motion vector value of the current block is an intermediate value of the motion vectors of the detected blocks when the number of detected blocks is three. The method of claim 13, wherein the motion vector predictor Adaptive motion vector prediction decoder, characterized in that the predicted motion vector value of the current block is (0,0) when no block is detected. (a) detecting a block to which the current block refers temporally or spatially; (b) detecting a block referring to the block detected in step (a) of the block adjacent to the current block; And (c) obtaining a predicted motion vector value of the current block based on the motion vector of the block detected in step (b). 20. The method of claim 19, wherein in step (c) The predictive motion vector value of the current block is a motion vector value of the detected block when the number of detected blocks is one. 20. The method of claim 19, wherein in step (c) The predicted motion vector value of the current block is the average value of the motion vectors of each of the two detected blocks when the number of detected blocks is two. 20. The method of claim 19, wherein in step (c) The predicted motion vector value of the current block is an intermediate value of the motion vectors of each of the detected three blocks when the number of detected blocks is three. 20. The method of claim 19, wherein in step (c) The predicted motion vector value of the current block is (0,0) when the block is not detected. The method of claim 19, The motion vector is adaptive motion vector prediction method in a multi-view video, characterized in that estimated based on the H.264 standard. (a) detecting a neighboring block that references the same block as the block that the current block refers to temporally or spatially and obtaining a predicted motion vector value of the current block based on the detected motion vector of the block; and (b) encoding a difference of a predicted motion vector value of the current block obtained by the motion vector predictor from the motion vector value of the current block. The method of claim 25, wherein step (a) (a1) detecting only a block of the left block, the upper block, and the right upper block of the current block that references the same block as the block to which the current block refers temporally or spatially; And (a2) calculating a predicted motion vector value of the current block based on the detected motion vector of the block; encoding method based on adaptive motion vector prediction. The method of claim 25, wherein step (b) The predicted motion vector value of the current block is the motion vector value of the detected block when the number of detected blocks is one. The method of claim 25, wherein step (b) The predictive motion vector value of the current block is an average value of the motion vectors of the detected blocks when the number of detected blocks is two. The method of claim 25, wherein step (b) The predictive motion vector value of the current block is an intermediate value of the motion vectors of the detected blocks when the number of detected blocks is three. The method of claim 25, wherein step (b) The prediction motion vector value of the current block is (0,0) when the block is not detected. A method for decoding blocks of an encoded multiview video bit stream based on adaptive motion vector prediction, comprising: (a) receiving a multi-view video bitstream including a motion vector difference value and motion vector information obtained by subtracting the predicted motion vector value of the predetermined block from the motion vector value of a predetermined block; (b) decoding the blocks in the received multi-view video bit stream using the motion vector information and the motion vector difference value; (c) detecting a neighboring block of the blocks in the decoded bit stream that refer to the same block as the block that the decoded current block refers to temporally or spatially and predicting the current block based on the detected motion vector of the neighboring block; Obtaining a motion vector value; And and (d) reconstructing the motion vector by adding the predicted motion vector value to the motion vector difference value. 32. The method of claim 31, wherein step (c) (c1) detecting only a block among the left block, the upper block, and the right upper block of the current block that refer to the same block as the block to which the current block refers in time or space; (c2) obtaining a predicted motion vector value of the current block based on the detected motion vector of the block. 32. The method of claim 31, wherein in step (c) The predictive motion vector value of the current block is a motion vector value of the detected block when the number of detected blocks is one. 32. The method of claim 31, wherein in step (c) The predictive motion vector value of the current block is an average value of motion vectors of the detected blocks when the number of detected blocks is two. 32. The method of claim 31, wherein in step (c) The predictive motion vector value of the current block is an intermediate value of the motion vectors of the detected blocks when the number of detected blocks is three. 32. The method of claim 31, wherein in step (c) The predicted motion vector value of the current block is (0,0) when the block is not detected.

Images