KR101407719B1 - Multi-view image coding method and apparatus using variable GOP prediction structure, multi-view image decoding apparatus and recording medium storing program for performing the method thereof - Google Patents

Abstract

The present invention discloses a multi-view video encoding method and apparatus and a video decoding apparatus using a variable picture group prediction structure. According to another aspect of the present invention, there is provided a multi-view image encoding method including: receiving a screen group including a plurality of screens; Coding the pictures of the screen group according to a predetermined plurality of group prediction modes, and calculating a bit-distortion cost value; Determining an optimal group prediction mode among the group prediction modes considering the bit-distortion cost value; And generating multi-view image information encoded according to the determined group prediction mode. According to the present invention, it is possible to improve the coding efficiency of multi-view images and to perform image coding with optimized temporal and spatial prediction structures.

Multi-view image, GOP, bit-distortion cost value, prediction structure

Description

TECHNICAL FIELD [0001] The present invention relates to a multi-view image coding method and an apparatus, an image decoding apparatus, and a recording medium on which a program for performing the method is recorded. apparatus and recording medium storing program for < RTI ID = 0.0 >

The present invention relates to a multi-view image encoding method and apparatus, and an image decoding apparatus, and more particularly, to a method and apparatus for improving the efficiency of multi-view image encoding through variable adjustment of a picture prediction structure in a time direction and a view direction .

A multi-view image is an image of the same three-dimensional scene captured using two or more cameras. In particular, a multi-view image is a scene in which a variety of geometric calibrated images are synthesized It is possible to provide an image of a viewpoint.

Panoramic images, which are an example of multi-view images, are being studied extensively in aerospace / aviation, computer vision, and computer graphics, and include aerial photograph analysis, image change detection, video compression, video indexing, camera resolution, and FOV of view) to simple image editing. In computer vision, the depth and disparity information of an object in the image are extracted using several images acquired at different points of view. In computer graphics, Point images to generate a realistic image at a virtual point of view.

The multi-view image processing technique is used for a surveillance system using an omnidirectional camera, a three-dimensional virtual viewpoint used in a game, or a viewpoint switching for arbitrarily selecting an image input from a plurality of camera images. In addition, such multi-view video images can be extended to various multimedia services using interactive contents or realistic contents in combination with network technology.

Conventionally, a formalized prediction structure is used for multi-view image coding. However, the multi-view image coding is different from the single-view image coding because the coding efficiency is dependent on the prediction structure and the spatial and temporal characteristics of the multi-view images are different from each other. Therefore, there is a certain limit in improving the coding efficiency when using the conventional prediction structure as in the conventional art.

The present invention relates to a multi-view image encoding method and a multi-view image encoding method, in which a prediction structure of images existing in adjacent space-time is variably adjusted according to temporal and spatial correlation for efficiently encoding multi-view images including a plurality of spatially adjacent scenes, And a multi-view image decoding apparatus.

According to another aspect of the present invention, there is provided a multi-view image encoding method comprising: inputting a screen group including a plurality of screens; Encoding the pictures of the picture group according to a predetermined plurality of group prediction modes and calculating a bit-distortion cost value of each of the encoded pictures; Determining one group prediction mode among the group prediction modes as an optimal group prediction mode considering the bit-distortion cost value; And generating multi-view image information encoded according to the determined group prediction mode.

According to another aspect of the present invention, there is provided a multi-view image encoding apparatus comprising: a buffer for receiving a screen group including a plurality of screens and storing an input screen group; A prediction mode encoding unit that encodes the pictures of the picture group according to a plurality of predetermined group prediction modes; A bit-distortion value calculation unit for calculating a bit-distortion cost value of each of the encoded pictures; A prediction mode determiner for determining one of the group prediction modes as an optimal group prediction mode considering the bit-distortion cost value; And an encoding unit for generating multi-view image information encoded according to the determined group prediction mode.

According to another aspect of the present invention, there is provided a multi-view image encoding apparatus comprising: an input unit receiving a screen group including a plurality of screens; A prediction mode adjuster for adjusting a group prediction mode related to coding of pictures included in the picture group; A bit-distortion value calculation unit for calculating a bit-distortion cost value of each of the coded pictures according to the group prediction mode adjusted by the prediction mode adjustment unit; And a prediction mode deciding unit for deciding one of the group prediction modes as an optimal group prediction mode considering the bit-distortion cost value, wherein the decoding unit decodes the multi- Point video encoding apparatus for generating information.

According to another aspect of the present invention, there is provided a multi-view image decoding apparatus comprising: a prediction mode decoding unit for decoding prediction mode information from bitstream information of a coded picture group; An entropy decoding unit for performing entropy decoding on the bitstream information; An inverse quantization unit for performing inverse quantization on residual component information reconstructed by the entropy decoding unit; A motion compensation unit for generating a motion compensated picture using the restored motion information through the entropy decoding unit; And a picture rearrangement unit for generating a reconstructed picture using the inversely quantized residual component information from the inverse quantization unit and the motion compensated pictures from the motion compensation unit and rearranging the reconstructed pictures.

The present invention also provides a computer-readable recording medium for performing the above-described multi-view image encoding method on a computer.

According to the present invention, the concept of a view group in the view direction (VGOP) and a view group in the time direction (TGOP) is introduced as a mid-level concept between a sequence and a picture in encoding of a multi- The coding efficiency of the multi-view image is improved by adjusting the prediction structure of the screen group variably in consideration of the distortion cost value, and the image coding can be optimized with the temporal / spatial prediction structure.

Hereinafter, a multi-view image encoding / decoding apparatus and method using the variable picture group prediction structure according to the present invention and a recording medium on which a program for performing the method are recorded will be described in detail with reference to the drawings.

1 is a schematic diagram showing a multi-view image transmission system according to the present invention. The multi-viewpoint image transmission system shown in FIG. 1 includes a plurality of multi-viewpoint cameras 12, 14, 16 and 18, an encoding device 20, an Internet 30, a decryption device 40 and a user terminal 50 do.

The multi-viewpoint camera captures an object to be photographed and transmits it to a group of multi-view image encoding devices 20 through digital or analog transmission lines. The multi-view image encoding apparatus 20 of the present embodiment encodes an image using a variable GOP prediction structure in units of screen picture groups instead of a fixed prediction structure. Here, the screen group refers to the screen group (VGOP) in the view direction of the screen and the screen group (TGOP) in the temporal direction, and the variable GOP prediction structure defines the I, P and B screen structures for the VGOP and TGOP Which means they are different. A detailed description of the multi-view image encoding apparatus will be described later. The compressed data in the multi-view video encoding device 20 is transmitted to the multi-view video decoding device 40 via the Internet or another data communication network. The multi-view image decoding apparatus 40 decodes the transferred data and outputs the reconstructed image through the output means provided in the user terminal 50. [

2 is a block diagram illustrating a multi-view image encoding apparatus according to an exemplary embodiment of the present invention. 2 includes a buffer 102, a downsampling unit 104, a GOP prediction mode encoding unit 106, a bit-distortion cost value calculation unit 108, a GOP prediction mode A determination unit 110 and an encoding unit 112. [

The buffer 102 receives a screen group obtained from the multi-view image acquisition apparatus. In the present invention, the screen group is used as a concept of a group of GOP (GGOP) including a screen group (VGOP) in the view direction and a screen group (TGOP) in the time direction. A GGOP is an intermediate level between a sequence and a picture. One GGOP is composed of a plurality of screens, and a plurality of GGOPs can be gathered to form a single sequence.

The downsampling unit 104 downsamples the pictures belonging to the input picture group stored in the buffer. Since the multi-view image encoding apparatus 20 of FIG. 2 includes the encoding unit according to the GOP prediction mode, the down-sampled image information is encoded, thereby reducing the amount of calculation required for encoding according to the prediction mode and the time required for the calculation. However, if the result of the encoding for each prediction mode performed after receiving the downsampled image information is used as the final encoding result, there is a problem that the quality of the reconstructed image is degraded. Therefore, in this embodiment, (112).

The GOP prediction mode-based encoding unit 106 encodes the downsampled picture group according to a plurality of predetermined group prediction modes. The GOP prediction mode-based encoding unit 106 includes a first encoding unit for each prediction mode and a second encoding unit for each prediction mode. The first encoding unit for each prediction mode encodes the anchor pictures in the picture group according to the group prediction mode in the view direction, and the second encoding unit encodes the non-anchor pictures according to the group prediction mode in the time direction.

The bit-distortion cost value calculation unit 108 calculates the bit-distortion cost value of each of the encoded images using the encoded image information according to the GOP prediction mode encoding unit and the original image information. The bit-distortion cost value calculation section 108 includes a first bit-distortion cost value calculation section and a second bit-distortion cost value calculation section. The first bit-distortion cost calculation unit calculates the bit-distortion cost value of each of the anchor pictures coded through the first encoding unit for each prediction mode, and the second bit-distortion cost calculation unit calculates the bit- And calculates a bit-distortion cost value according to each encoded non-anchor picture.

FIG. 3 shows an example of a prediction structure for encoding a multi-view image having eight viewpoints. The prediction structure of FIG. 3 is composed of a view direction (y-axis, S0, S1, ..., S7) and a time direction (x-axis, T0, T1 ...). In Fig. 3, the arrow indicates the reference relationship between two frames. If A- > B, B means that A is coded with reference to A. As shown in FIG. 3, screens (T0, T8, ..) of a time zone having a reference relationship only in a view direction are referred to as an anchor screen, and screens other than an anchor screen are referred to as a non-anchor screen. The prediction structure of multi-view image coding can be classified into a prediction structure for a temporal direction having a spatial correlation and a temporal prediction structure having a temporal correlation.

Regarding the prediction structure for the view direction, a conventional GOP prediction structure such as IBPBBBPP is conventionally used and the same prediction structure is used regardless of time. The multi-view image coding method of the present invention is different from the conventional prediction structure in that it variably adjusts the prediction structure of the view direction and the time direction. 3 shows a VGOP representing a screen group in the view direction and a TGOP representing a screen group in the time direction. First, the coding method of the VGOP, which is a screen group in the view direction, is preferable to locate the position of the I viewpoint close to the central viewpoint through the equation (1). In Equation (1), Num _view means the number of viewpoints.

[Equation 1]

Num_view/2

I _view =

Num _view / 2

If I _view is centered according to Equation 1, the basic prediction structure is ... PPIPP ... , And the extended prediction structure depends on how many B screens are to be placed between the P screens. For example, PBPBIBPBP ... The form of a B is inserted in the form of ... PBBPBBIBBP ... Is a form in which two Bs are inserted. Thus, various prediction structures centering on the I picture can exist. In the present invention, the above-described various types of prediction structures are defined as a group prediction mode. The group prediction modes for the picture group (VGOP) in various view direction directions are defined in advance, and the optimal group prediction modes are selected from the viewpoint of the bit-distortion cost value as shown in Equation (2) Point video encoding in accordance with the present invention. Then, information on the selected optimal group prediction mode is encoded and transmitted to the header of the screen group (VGOP) in the view direction, and the decoder decodes the information using the information.

&Quot; (2) "

C = D +? R

Here, D is a distortion when coding is performed according to an arbitrary group prediction mode, R is a rate when coding is performed according to the group prediction mode, C is a bit-distortion cost value, and? Can be a value defined in H.264 / AVC as a weight for distortion and bits. The bit-distortion cost value can be calculated for each picture coded according to the group prediction mode using Equation 2. If the bit-distortion cost values of the pictures belonging to the corresponding VGOP are summed, Can be found.

In the present invention, the coding according to the group prediction mode of the temporal picture group (TGOP) is performed after determining the group prediction mode of the viewing direction group. In general, the prediction structure of TGOP has a hierarchical B screen structure. The structure of the TGOP is determined according to the defined GOP length so as to facilitate random access of the sequence. In the case of FIG. 3, the GOP length is 8. However, even if the GOP length is determined, various hierarchical B screen structures are possible. You can define two screens of the B1 level or you can shift them to the other. Since the coding efficiency varies depending on the structure of the hierarchical B picture, in the case of the TGOP as well as the VGOP, the group prediction modes in various time directions are predefined in advance, and the group prediction mode having the optimum coding efficiency for each TGOP is selected . Here, the optimal coding efficiency can be determined from the viewpoint of bit-distortion.

The GOP prediction mode determination unit 110 determines one group prediction mode among the group prediction modes considering the bit-distortion cost value. The GOP prediction mode determination unit 110 includes a first prediction mode determination unit and a second prediction mode determination unit. The first prediction mode decision unit decides a group prediction mode that minimizes the sum of the bit-distortion cost values calculated according to the second bit-distortion calculation unit among the group prediction modes in the temporal direction, The group prediction mode in the view direction that minimizes the sum of the bit-distortion cost values among the group prediction modes in the view direction is determined.

The encoding unit 112 encodes the image of the picture group according to the selected group prediction mode according to the GOP prediction mode deciding unit 110. [ In the present embodiment, since the prediction mode-based encoding of the downsampled image information is performed, it is preferable to perform image encoding through a separate encoding unit. However, unlike the present embodiment, when the downsampling unit is not provided, the GOP prediction mode determination unit may perform the encoding using the encoding result according to the group prediction mode in the view direction. Here, the encoding result means the encoded image information and the identification information for the group prediction mode.

4 is a block diagram illustrating a multi-view image encoding apparatus according to another embodiment of the present invention. The multi-view image encoding apparatus 20 'shown in FIG. 4 includes a buffer 152, a picture rearrangement unit 154, a DCT 156, a quantization unit Q 158, an inverse quantization unit Q ^-1 , and 160, an IDCT 162, an intra predictor 164, a motion compensator 168, a motion estimator 170, a GOP prediction mode adjuster 172, A cost value calculation unit 174, a GOP prediction mode determination unit 176, an entropy encoding unit 178, and a bitstream generation unit 180.

The multi-view image encoding apparatus of the present embodiment further includes a GOP prediction mode adjusting unit 172, a bit-distortion cost value calculating unit 174 and a GOP prediction mode determining unit 176 in the conventional image encoding apparatus, And outputs an encoding result value according to an optimal group prediction mode from the viewpoint of bit-distortion.

For example, when the length of the view group (VGOP) in the view direction is 8 and the four group prediction modes (IBPBPBPP, PPBIBPBP, PBBIBBPP and PPBBIBBP) for the VGOP are present, the image encoding apparatus of this embodiment has one view direction Group, and outputs the best group prediction mode among the four group prediction modes as a final encoding result. Hereinafter, the components constituting the image encoding apparatus will be described.

The buffer 152 receives and temporarily stores multi-view image information obtained from an image acquisition device such as a camcorder, a digital camera, and the like. The picture rearranging unit 154 accesses the buffer in accordance with the rearrangement procedure to be described later and provides the data of the picture to be encoded to the motion estimation unit and the subtractor.

First, a forward path will be described in detail. The target picture to be coded transmitted from the picture rearranging unit 154 is input to a subtractor. The subtractor generates a difference value matrix of the reconstructed reference picture and the target picture through the motion compensation unit 168 and transmits the generated difference value matrix to the DCT unit 156. [ The DCT unit 156 calculates a DCT coefficient by performing discrete cosine transform on the difference value matrix. The quantization unit (Q) 158 quantizes the DCT coefficients generated in the DCT unit. The DCT coefficients quantized by the quantization unit 158 are transmitted to the entropy encoding unit 178, and entropy-encoded by a method such as CAVLC or CAVAC. The entropy-encoded data is transmitted to the external network through the bitstream generator 180.

The following describes the reconstruction path in detail. In this embodiment, the quantized data through the quantization unit 158 is input to an inverse quantization unit (Q ^-1 , 160) and an IDCT 162 and a summer. The intraprediction unit 164 generates an I-picture using the intra-picture prediction algorithm and transfers the generated I-picture to the GOP prediction mode adjuster 172. In the case of coding in the inter mode, the reconstructed picture is stored in the picture storage unit 166, and then the stored picture is transferred to the motion compensation unit 168 and the motion compensation unit 168.

The motion predicting unit 170 predicts the motion of the target picture input from the picture rearranging unit 154 and transmits the motion vector for the block of the target picture to the entropy coding unit 178. [ The motion vector generated in the motion estimating unit 170 is transmitted to the motion compensating unit 168. The motion compensating unit 168 compensates the motion compensated prediction using the information about the reconstructed picture from the picture storing unit and the motion vector, Create a screen. The difference between the generated prediction picture and the target picture input from the picture rearrangement unit is calculated by the subtractor and transmitted to the DCT unit 156 as described above. The data of the predicted picture in the motion compensation unit is also input to an adder. The input data is added to the reconstructed difference value matrix through the IDCT and stored in the picture storage unit 166.

The GOP prediction mode adjuster 172 adjusts whether the target picture is coded into an I picture or a P or B picture according to a plurality of predetermined group prediction modes, And transmits the screen to the subtractor.

The bit-distortion cost value calculation unit 174 calculates a bit-distortion cost value of each of the pictures coded according to the plurality of group prediction modes. The bit-distortion cost value calculation unit calculates the bit-distortion cost value according to the current group prediction mode in consideration of the reconstructed image information, original image information, and bit rate generated through the reconstruction path.

The GOP prediction mode determination unit 176 adds the bit-distortion cost values of each of the eight pictures to each of the four group prediction modes, determines the group prediction mode that minimizes the summed value as the optimal group prediction mode do. In addition, the GOP prediction mode determination unit 176 transfers the identification information of the determined group prediction mode to the entropy encoding unit 178.

5 is a flowchart illustrating a multi-view image encoding method according to an embodiment of the present invention. The multi-view image encoding method shown in FIG. 5 includes the following steps performed in the image encoding apparatus of FIG.

In step 210, the buffer 102 receives image information in units of picture groups (GOPs). Here, the screen group includes a screen group in the view direction and a screen group in the time direction as a screen group of the multi-view image.

In step 220, the down-sampling unit 104 performs down-sampling on each of the screens included in the input screen group.

In step 230, the GOP prediction mode encoding unit 106 encodes the anchor pictures of the screen group in the view direction among the screen groups according to the group prediction mode in the view direction.

In operation 240, the bit-distortion cost calculator 108 calculates a bit-distortion cost value of each of the encoded anchor pictures in operation 230.

In step 250, the GOP prediction mode determination unit 110 determines the group prediction mode that minimizes the sum of the bit-distortion cost values among the group prediction modes in the view direction as the optimal group prediction mode.

In step 260, the GOP prediction mode encoding unit 106 encodes the non-anchor pictures in accordance with temporal group prediction modes related to the group prediction mode determined in step 250, among group temporal prediction modes for non- As shown in FIG. For example, when the prediction mode for the anchor picture group is determined to be PPBIBPBP, the prediction mode encoding at the time S0 is performed only for the group prediction modes starting from P among the group prediction modes in the temporal direction.

In operation 270, the bit-distortion cost calculator 108 calculates a bit-distortion cost value according to each non-anchor picture encoded in operation 260. [

In step 280, the GOP prediction mode determination unit 110 determines the group prediction mode that minimizes the sum of the bit-distortion cost values among the group prediction modes in the time direction to be the optimal group prediction mode.

In step 290, the encoding unit 112 encodes the multi-view images stored in the buffer 102 according to the optimal group prediction mode determined by the GOP prediction mode deciding unit 110, and outputs the encoded image information do. Unlike the multi-view image encoding method of the present embodiment, it is also possible to implement a coding method so as not to include the down-sampling step. In this case, since the encoding result generated by the encoding unit for each GOP prediction mode, that is, the encoded image information, can be utilized, the encoding unit 112 encodes the information for identifying the determined group prediction mode and the already- To perform coding.

6 is a block diagram illustrating a multi-view image decoding apparatus according to an embodiment of the present invention. 6 includes a GOP prediction mode decoding unit 302, an entropy decoding unit 304, an inverse quantization unit (Q ^-1 , 306), an IDCT 308, (310) and a screen rearrangement unit (312).

The GOP prediction mode decoding unit 302 reconstructs the identification information for the group prediction mode from the bitstream of the coded picture group. The entropy decoding unit 304 performs entropy decoding on the input bitstream according to the identification information of the restored group prediction mode. The inverse quantization unit (Q ^-1 , 306) dequantizes the entropy-decoded residual component information, and the inverse discrete cosine transform unit (IDCT) 308 performs inverse operation of the discrete cosine transform to convert the frequency component into a pixel component . The motion compensation unit 310 performs motion compensation using the entropy-decoded motion information to generate motion compensated reconstructed image information. The restored image information is added to the residual component information from the IDCT unit 308, and finally the reconstructed image information is transmitted to the screen rearrangement unit 312. [ The picture rearranging unit 312 receives the reconstructed picture information from the adder, and rearranges the picture according to the reproduction time order.

Meanwhile, the image encoding and decoding method of the present invention can be implemented by a computer-readable code on a computer-readable recording medium. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored.

Examples of the computer-readable recording medium include a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device and the like, and also a carrier wave (for example, transmission via the Internet) . In addition, the computer readable recording medium may be distributed over networked computer systems so that computer readable code can be stored and executed in a distributed manner. Further, functional programs, codes, and code segments for implementing the present invention can be easily deduced by programmers in the technical field to which the present invention belongs.

The present invention has been described above with reference to preferred embodiments. It will be understood by those skilled in the art that the present invention may be embodied in various other forms without departing from the spirit or essential characteristics thereof. Therefore, the above-described embodiments should be considered in an illustrative rather than a restrictive sense. The scope of the present invention is indicated by the appended claims rather than by the foregoing description, and all differences within the scope of equivalents thereof should be construed as being included in the present invention.

The present invention relates to a variable-length multi-view image coding method and apparatus, and an image decoding apparatus for a prediction structure for coding a view group in a view direction (VGOP) and a view group in a time direction (TGOP) The present invention is useful for application to various multimedia service systems using an image encoding system as well as interactive contents and realistic contents.

1 is a schematic diagram showing a multi-view image transmission system according to the present invention.

2 is a block diagram illustrating a multi-view image encoding apparatus according to an exemplary embodiment of the present invention.

FIG. 3 shows an example of a prediction structure for multi-view image encoding with eight viewpoints.

4 is a block diagram illustrating a multi-view image encoding apparatus according to another embodiment of the present invention.

5 is a flowchart illustrating a multi-view image encoding method according to an embodiment of the present invention.

6 is a block diagram illustrating a multi-view image decoding apparatus according to an embodiment of the present invention.

Claims (17)

In the multi-view image coding method, a) receiving a screen group including a plurality of screens; b) encoding the pictures of the picture group according to a plurality of predetermined group prediction modes, and calculating a bit-distortion cost value of each of the encoded pictures; c) determining one group prediction mode among the group prediction modes considering the bit-distortion cost value; And d) generating multi-view image information encoded according to the determined group prediction mode. The method according to claim 1, In the step a), the picture group includes a view group (VGOP) and a temporal group (TGOP). In the step b), the group prediction modes include a plurality And a plurality of group prediction modes for group-prediction mode and temporal-group-based picture group coding. 3. The method of claim 2, wherein step b) b1) encoding the anchor pictures among the picture groups in the view direction according to the group prediction mode in the view direction; And b2) calculating a bit-distortion cost value of each of the anchor pictures coded through the step b1). The method of claim 3, The determination of one group prediction mode in step c) Determining a group prediction mode that minimizes a sum of the bit-distortion cost values calculated in the step b2) among the group prediction modes in the view direction, The step of generating the multi-view image information in step d) Wherein the image information encoded in accordance with the group prediction mode in the view direction selected in step c) and the identification information on the determined group prediction mode are used from the coding information generated as a result of step b1) Image encoding method. 3. The method of claim 2, wherein step b) b1) downsampling the anchor pictures among the picture groups in the view direction; b2) encoding the downsampled anchor pictures according to the group prediction modes in the view direction; And b3) calculating a bit-distortion cost value according to each anchor picture coded through the step b2). 3. The method of claim 2, wherein step b) b1) encoding the anchor pictures among the picture groups in the view direction according to the group prediction modes in the view direction; b2) calculating a bit-distortion cost value of each of the anchor pictures coded through the step b1), and calculating a bit-distortion cost value of a group prediction mode in a view direction that minimizes a sum of the bit- Determining a mode; b3) selecting time direction group prediction modes related to the group prediction mode in the view direction determined in step b2) among the group prediction modes in the temporal direction, and encoding non-anchor pictures according to the selected group prediction modes ; b4) calculating a bit-distortion cost value according to each encoded non-anchor picture. The method according to claim 6, Wherein the step c) comprises: a step of minimizing a sum of the bit-distortion cost values of the non-anchor picture calculated in step b4) among the temporal group prediction modes related to the group prediction mode determined in step b2) Determining a group prediction mode of a direction, The generating of the multi-view image information in the step d) may include using the encoded image information generated as a result of the step b3) and the identification information on the group prediction mode determined in the step b2) or c) A multi-view video encoding method characterized by: 3. The method of claim 2, Wherein the view direction group prediction modes include at least one group prediction mode selected from " ... PPIPP ... ", "... PBIBP ... ", and" ... PBBIBBP ... ". A computer-readable recording medium on which a program for performing the multi-view image encoding method of any one of claims 1 to 8 is recorded on a computer. In a multi-view video encoding apparatus, A buffer for receiving a screen group including a plurality of screens and storing data for the input screen group; A prediction mode encoding unit that encodes the pictures of the picture group according to a plurality of predetermined group prediction modes; A bit-distortion value calculation unit for calculating a bit-distortion cost value of each of the encoded pictures; A prediction mode determining unit for determining one group prediction mode among the group prediction modes considering the bit-distortion cost value; And And an encoding unit for generating multi-view image information encoded according to the determined group prediction mode. 11. The method of claim 10, Wherein the screen group includes screens in a view direction and screens in a time direction, Wherein the group prediction modes include a plurality of group prediction modes for the screen group coding in the view direction and a plurality of group prediction modes for the screen group coding in the temporal direction. 12. The method of claim 11, Wherein the encoding unit for each prediction mode encodes the anchor pictures included in the picture group according to the group prediction mode in the view direction, and the bit-distortion cost value calculation unit calculates the bit-distortion cost value of each of the anchor pictures, - calculating a distortion cost value, Wherein the prediction mode determination unit determines a group prediction mode that minimizes a sum of the bit-distortion cost values among the group prediction modes in the view direction. 12. The method of claim 11, Wherein the encoding unit for each prediction mode includes a first encoding unit for each prediction mode for encoding the anchor pictures included in the picture group according to the group prediction mode in the view direction and a second encoding unit for encoding non-anchor pictures according to the group prediction mode in the time direction And a second encoding unit for each prediction mode, Wherein the bit-distortion cost calculator comprises a first bit-distortion calculator for calculating a bit-distortion cost value of each of the anchor pictures coded by the first encoder for each prediction mode, and a second bit- And a second bit-distortion calculation unit for calculating a bit-distortion cost value according to each non-anchor picture, Wherein the prediction mode determination unit includes a first prediction mode determination unit for determining one group prediction mode that minimizes a sum of bit-distortion cost values among the group prediction modes in the view direction, And a second prediction mode deciding unit for deciding a group prediction mode that minimizes a sum of bit-distortion cost values. 12. The method of claim 11, And a down-sampling unit for down-sampling screens of the input screen group, Wherein the encoding unit for each prediction mode encodes the downsampled pictures according to the group prediction modes. 15. The method of claim 14, Wherein the prediction mode determination unit determines one group prediction mode that minimizes a sum of bit-distortion cost values among the group prediction modes, Wherein the encoding unit generates encoded multi-view image information according to the determined group prediction mode. In a multi-view video encoding apparatus, A buffer for receiving a screen group including a plurality of screens and storing data for the input screen group; A prediction mode controller for adjusting a group prediction mode for coding pictures included in the picture group; A bit-distortion value calculation unit for calculating a bit-distortion cost value of each of the coded pictures according to the group prediction mode adjusted by the prediction mode adjustment unit; And And a prediction mode determining unit for determining one group prediction mode among the group prediction modes considering the bit-distortion cost value, wherein the multi-view image coding unit generates multi-view image information encoded according to the determined group prediction mode, Encoding apparatus. delete

Images