CN105025300A - Method and device for encoding/decoding images - Google Patents

Abstract

A method and device for encoding/decoding an image: determining a motion vector of a current block based on a motion vector of at least one block encoded/decoded before the current block is encoded/decoded, and predicting based on a first direction according to the determined motion vector , performing predictive encoding/decoding on the current block by one of second direction prediction and bidirectional prediction.

Description

Method and device for encoding/decoding images

This application was submitted to the China Intellectual Property Office with an application date of January 19, 2011, an application number of 201180014824.9, and an invention title of "Method for encoding/decoding images using the motion vector of the previous block as the motion vector of the current block" and equipment" application.

technical field

The present invention relates to a method and device for encoding/decoding images, and more particularly, to a method and device for encoding/decoding images based on inter-frame prediction.

Background technique

According to video compression standards such as MPEG-1, MPEG-2, and MPEG-4 H.264/MPEG-4 AVC (Advanced Video Coding), images are encoded by dividing them into blocks of a predetermined size. Next, each block of a predetermined size is predictively encoded by using inter prediction or intra prediction.

For inter prediction, motion estimation is performed by searching for at least one reference frame of a block that is the same as or similar to a current block, a motion vector generated by performing motion estimation is encoded together with pixel values, and the encoded result is then inserted into a bitstream.

Contents of the invention

Technical solutions

One or more aspects of the present invention provide an image encoding method and device, an image decoding method and device, and a computer-readable recording medium recorded with a program for executing the image encoding method and/or the image decoding method , wherein, in the image encoding method and device and the image decoding method and device, the motion vector of the current block is determined based on the motion vector of at least one previous block, and the image is encoded/decoded based on the determined motion vector.

Beneficial effect

According to the present invention, encoding is performed using a motion vector of at least one previous block as a motion vector of a current block, thereby increasing the hit ratio of encoding modes and the accuracy of prediction. Therefore, images can be encoded and decoded with a higher compression rate.

Description of drawings

FIG. 1 is a block diagram of an image encoding device according to an embodiment of the present invention.

FIG. 2 is a block diagram of an image decoding device according to an embodiment of the present invention.

FIG. 3 illustrates a layered coding unit according to an embodiment of the present invention.

FIG. 4 is a block diagram of a coding unit-based image encoder according to an embodiment of the present invention.

FIG. 5 is a block diagram of a coding unit based image decoder according to an embodiment of the present invention.

FIG. 6 illustrates a maximum coding unit, a sub coding unit, and a prediction unit according to an embodiment of the present invention.

FIG. 7 illustrates coding units and transformation units according to an embodiment of the present invention.

8a and 8b illustrate split shapes of coding units, prediction units, and transformation units according to an embodiment of the present invention.

FIG. 9 is a block diagram of an image encoding device according to another embodiment of the present invention.

FIG. 10 is a diagram illustrating a method of predicting a block included in a bidirectional predictive image according to an embodiment of the present invention.

11a and 11b are diagrams illustrating a method of determining a motion vector of a current block based on a motion vector of a previously encoded block according to an embodiment of the present invention.

12a and 12b are diagrams illustrating a method of determining a motion vector of a current block based on a motion vector of a previously encoded block according to an embodiment of the present invention.

13a and 13b are diagrams illustrating a method of determining a motion vector of a current block based on a motion vector of a previously encoded block according to an embodiment of the present invention.

FIG. 14 is a block diagram of an image decoding device according to another embodiment of the present invention.

FIG. 15 is a flowchart illustrating an image encoding method according to an embodiment of the present invention.

FIG. 16 is a flowchart illustrating an image decoding method according to an embodiment of the present invention.

best mode

According to an aspect of the present invention, there is provided an image decoding method of determining a motion vector of a current block based on a motion vector of at least one block decoded before the current block is decoded and decoding the current block based on the determined motion vector, The method includes: decoding information on a prediction direction to be used for decoding the current block in a first direction, a second direction, and bidirectionally, and information on pixel values of the current block; As a result of decoding the information, determine the prediction direction in which the current block will be predicted, and determine at least one motion vector used to predict the current block according to the determined prediction direction; based on the at least one motion vector and the information about the pixel value. As a result of decoding, restore the current block. The first direction is the direction from the current image to the previous image, and the second direction is the direction from the current image to the subsequent image.

According to another aspect of the present invention, there is provided an image encoding method of determining a motion vector of a current block based on a motion vector of at least one block encoded before the current block is encoded and encoding the current block based on the determined motion vector , the method includes: determining a first direction motion vector and a second direction motion vector of the current block based on the motion vector of the at least one block; based on the first direction motion vector and the second direction motion vector, predicting from the first direction, Determining a prediction method to be used for encoding the current block in the second direction prediction and bidirectional prediction; information about the prediction direction used in the determined prediction method and about pixel values of the current block generated based on the prediction method information is encoded. The first direction is the direction from the current image to the previous image, and the second direction is the direction from the current image to the subsequent image.

According to another aspect of the present invention, there is provided an image decoding apparatus that determines a motion vector of a current block based on a motion vector of at least one block decoded before the current block is decoded and decodes the current block based on the determined motion vector , the apparatus comprising: a decoder for decoding information about a prediction direction to be used for decoding a current block in a first direction, a second direction, and in both directions and information about pixel values of the current block; a motion vector A determiner, based on a result of decoding the information about the prediction direction, determines a prediction direction in which the current block will be predicted, and determines at least one motion vector for predicting the current block in accordance with the determined prediction direction; the restoration unit, based on the At least one motion vector and a result of decoding information about pixel values, recover the current block. The first direction is the direction from the current image to the previous image, and the second direction is the direction from the current image to the subsequent image.

According to another aspect of the present invention, there is provided an image encoding apparatus for determining a motion vector of a current block based on a motion vector of at least one block encoded before the current block is encoded and encoding the current block based on the determined motion vector , the device includes: a motion vector determiner, based on the motion vector of the at least one block, determines a first direction motion vector and a second direction motion vector of the current block; an encoder, based on the first direction motion vector and the second direction motion Vector, determining the prediction method to be used to encode the current block from among the first direction prediction, the second direction prediction and the bidirectional prediction, the information on the prediction direction used in the determined prediction method and the information on the prediction based on the method to encode information about the pixel values of the current block. The first direction is the direction from the current image to the previous image, and the second direction is the direction from the current image to the subsequent image.

According to another aspect of the present invention, there is provided a computer-readable recording medium recorded with a program for executing the image encoding method and/or the image decoding method.

Detailed ways

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

FIG. 1 is a block diagram of an image encoding device 100 according to an embodiment of the present invention.

Referring to FIG. 1 , an image encoding apparatus 100 includes a maximum coding unit divider 110 , a coded depth determiner 120 , an image data encoder 130 , and an encoding information encoder 140 .

The maximum coding unit splitter 110 may split a current frame or a slice based on a maximum coding unit that is a coding unit having a maximum size. That is, the maximum coding unit divider 110 may divide the current frame or slice into at least one maximum coding unit.

According to an embodiment of the present invention, a coding unit may be represented using a maximum coding unit and a depth. As described above, the maximum coding unit indicates the coding unit having the largest size among the coding units of the current frame, and the depth indicates the degree to which the coding unit is hierarchically reduced. As the depth increases, the coding unit decreases from a maximum coding unit to a minimum coding unit, where the depth of the maximum coding unit is defined as the minimum depth and the depth of the minimum coding unit is defined as the maximum depth. Since the size of a CU decreases from the largest CU as the depth increases, a sub-coding unit of the k-th depth may include multiple sub-coding units of the k+n-th depth (k and n are integers equal to or larger than 1) .

Encoding an image in a larger coding unit may result in a higher image compression rate according to an increase in the size of a frame to be encoded. However, if a larger coding unit is fixed, it may not be possible to efficiently encode an image by reflecting continuously changing image characteristics.

For example, when a smooth area such as the sea or sky is encoded, the larger the coding unit is, the more the compression ratio can be improved. However, when a complex area such as a person or a building is encoded, the smaller the coding unit, the more the compression rate can be improved.

Therefore, according to the embodiments of the present invention, different maximum image coding units and different maximum depths are set for each frame or slice. Since the maximum depth represents the maximum number of times the coding unit may be reduced, the size of each minimum coding unit included in the maximum image coding unit may be variously set according to the maximum depth.

The coded depth determiner 120 determines a maximum depth. The maximum depth may be determined based on a calculation of a rate-distortion (RD) penalty. The maximum depth may be determined differently for each frame or slice or for each maximum coding unit. The determined maximum depth is provided to the encoding information encoder 140 , and the image data according to the maximum coding unit is provided to the image data encoder 130 .

The maximum depth represents a coding unit having a minimum size (ie, a minimum coding unit) that may be included in the maximum coding unit. In other words, the maximum coding unit may be divided into sub-coding units having different sizes according to different depths (see FIGS. 8a and 8b for details). Sub coding units having different sizes included in the maximum coding unit may be predicted or transformed based on processing units having different sizes. The transform may be a discrete cosine transform (DCT) or a Karhunen Loever transform (KLT), in which pixel values in the spatial domain are transformed into frequency domain coefficients.

In other words, the image encoding apparatus 100 may perform a plurality of processing operations for image encoding based on processing units having various sizes and various shapes. To encode image data, multiple processing operations are performed, such as prediction, transform, and entropy coding, wherein processing units of the same size or different sizes may be used for each operation.

For example, the image encoding apparatus 100 may select a processing unit different from a coding unit to predict the coding unit.

When the size of the coding unit is 2Nx2N (where N is a positive integer), the processing unit for prediction may be 2Nx2N, 2NxN, Nx2N, and NxN. In other words, motion prediction may be performed based on a processing unit having a shape that bisects at least one of a height and a width of a coding unit. Hereinafter, a processing unit serving as a basis for prediction is defined as a "prediction unit".

The prediction mode may be at least one of an intra mode, an inter mode, and a skip mode, and a specific prediction mode may be performed only on a prediction unit having a specific size or a specific shape. For example, intra mode may be performed only on prediction units of sizes 2Nx2N and NxN having a square shape. Also, skip mode may be performed only on prediction units of size 2Nx2N. If a plurality of prediction units exists in a coding unit, after prediction is performed on each prediction unit, a prediction mode having the smallest encoding error may be selected.

Alternatively, the image encoding apparatus 100 may perform transformation on image data based on a processing unit having a different size from a coding unit. For transformation in a coding unit, transformation may be performed based on a processing unit having the same size as the coding unit or smaller than the coding unit. Hereinafter, a processing unit serving as a basis for transformation is defined as a "transformation unit".

The coded depth determiner 120 may determine sub coding units included in the maximum coding unit by using Lagrange multiplier-based RD optimization. In other words, the coded depth determiner 120 may determine what shape a plurality of sub coding units split from the maximum coding unit having different sizes according to depths of the plurality of sub coding units have. The image data encoder 130 outputs a bitstream by encoding the maximum coding unit based on the split shape determined by the coded depth determiner 120 .

The encoding information encoder 140 encodes information about an encoding mode of the maximum coding unit determined by the coded depth determiner 120 . In other words, the encoding information encoder 140 outputs a bitstream by encoding information about a split shape of a maximum coding unit, information about a maximum depth, and information about an encoding mode of a sub coding unit of each depth. The information about the encoding mode of the sub-coding unit may include information about a prediction unit of the sub-coding unit, information about a prediction mode of each prediction unit, and information about a transformation unit of the sub-coding unit.

The information on the split shape of the maximum coding unit may be information indicating whether each coding unit is split, for example, flag information. For example, when the maximum coding unit is split and encoded, information indicating whether the maximum coding unit is split is encoded. Also, when a sub coding unit split from the maximum coding unit is split and encoded, information indicating whether the sub coding unit is split is encoded.

Since sub coding units having different sizes exist in each maximum coding unit and information on encoding modes is determined for each sub coding unit, information on at least one encoding mode may be determined for one maximum coding unit.

The image encoding apparatus 100 may generate sub coding units by halving the height and width of the maximum coding unit according to an increase in depth. That is, when the size of the coding unit of the k-th depth is 2N×2N, the size of the coding unit of the (k+1)-th depth is N×N.

Accordingly, the image encoding apparatus 100 may determine an optimal split shape for each maximum coding unit based on the size and maximum depth of the maximum coding unit in consideration of image characteristics. By variably adjusting the size of the maximum coding unit in consideration of image characteristics and encoding the image by dividing the maximum coding unit into sub-coding units of different depths, images having different resolutions may be more efficiently encoded.

FIG. 2 is a block diagram of an image decoding device 200 according to an embodiment.

Referring to FIG. 2 , an image decoding device 200 includes an image data acquisition unit 210 , an encoding information extractor 220 and an image data decoder 230 .

The image data acquiring unit 210 acquires image data according to the maximum coding unit by parsing the bitstream received by the image decoding apparatus 200 and outputs the image data to the image data decoder 230 . The image data obtaining unit 210 may extract information about a maximum coding unit of a current frame or slice from a header of the current frame or slice. In other words, the image data obtaining unit 210 divides the maximum coding unit so that the image data decoder 230 can decode the image data according to the maximum coding unit.

The encoding information extractor 220 extracts information about a maximum coding unit, a maximum depth, a split shape of a maximum coding unit, and an encoding mode of a sub coding unit from a header of a current frame by parsing a bitstream received by the image decoding apparatus 200 . Information about the split shape of the maximum coding unit and information about the encoding mode are provided to the image data decoder 230 .

The information about the split shape of the maximum coding unit may include information about sub coding units having different sizes according to depths and included in the maximum coding unit. As described above, the information on the split shape may be information indicating whether each coding unit is split, for example, flag information. The information about encoding modes may include information about prediction units according to sub coding units, information about prediction modes, and information about transformation units.

The image data decoder 230 restores the current frame by decoding the image data of each maximum coding unit based on the information extracted by the encoding information extractor 220 .

The image data decoder 230 may decode sub coding units included in the maximum coding unit based on the information about the split shape of the maximum coding unit. The decoding process may include prediction processing including intra prediction and motion compensation, and inverse transform processing.

The image data decoder 230 may perform intra prediction or inter prediction based on the information about the prediction unit of each sub coding unit and the information about the prediction mode, thereby predicting each sub coding unit. The image data decoder 230 may also perform inverse transformation on each sub-coding unit based on information about transformation units of the sub-coding unit.

FIG. 3 illustrates a layered coding unit according to an embodiment of the present invention.

Referring to FIG. 3 , hierarchical coding units may include coding units whose width×height is 64×64, 32×32, 16×16, 8×8, and 4×4. In addition to these coding units with a perfectly square shape, there may also be coding unit.

Referring to FIG. 3 , for image data 310 whose resolution is 1920×1080, the size of the maximum coding unit is set to 64×64, and the maximum depth is set to 2.

For the image data 320 whose resolution is 1920×1080, the size of the maximum coding unit is set to 64×64, and the maximum depth is set to 3. For the image data 330 whose resolution is 352×288, the size of the maximum coding unit is set to 16×16, and the maximum depth is set to 2.

When the resolution is high or the amount of data is large, the maximum size of the coding unit may be relatively large to improve the compression ratio and accurately reflect image characteristics. Accordingly, for the image data 310 and 320 having a higher resolution than the image data 330 , 64×64 may be selected as the size of the maximum coding unit.

The maximum depth indicates the total number of layers in the hierarchical coding unit. Since the maximum depth of the image data 310 is 2, the coding unit 315 of the image data 310 may include a maximum coding unit having a major axis size of 64 and sub coding units having a major axis size of 32 and 16 according to an increase in depth.

On the other hand, since the maximum depth of the image data 330 is 1, the coding unit 335 of the image data 330 may include a maximum coding unit with a major axis size of 16 and coding units with major axis sizes of 8 and 4 according to an increase in depth.

However, since the maximum depth of the image data 320 is 3, the coding unit 325 of the image data 320 may include a maximum coding unit with a major axis size of 64 and subcodes with a major axis size of 32, 16, 8, and 4 according to an increase in depth. coding unit. Since an image is encoded based on smaller sub-coding units with increasing depth, the embodiment may be applicable to encoding an image including more fine scenes.

FIG. 4 is a block diagram of a coding unit-based image encoder 400 according to an embodiment of the present invention.

The intra prediction unit 410 performs intra prediction on the prediction unit of the intra mode in the current frame 405, and the motion estimator 420 and the motion compensator 425 perform inter prediction on the prediction unit of the inter mode using the current frame 405 and the reference frame 495 and motion compensation.

A residual value is generated based on the prediction unit output from the intra prediction unit 410 , the motion estimator 420 , and the motion compensator 425 , and the generated residual value is output as a quantized transform coefficient through the transformer 430 and the quantizer 440 .

The quantized transform coefficients are restored to residual values by an inverse quantizer 460 and an inverse transformer 470, and the restored residual values are post-processed by a deblocking unit 480 and a loop filtering unit 490, and output as a reference frame 495 . The quantized transform coefficients may be output as a bitstream 455 by an entropy encoder 450 .

In order to perform encoding based on the image encoding method according to an embodiment of the present invention, the components of the image encoder 400 (ie, intra prediction unit 410, motion estimator 420, motion compensator 425, transformer 430, quantizer 440, entropy encoding 450, dequantizer 460, inverse transformer 470, deblocking unit 480, and loop filtering unit 490) perform an image encoding process based on a maximum coding unit, sub coding units according to depths, prediction units, and transformation units.

FIG. 5 is a block diagram of a coding unit-based image decoder 500 according to an embodiment of the present invention.

Referring to FIG. 5, a bit stream 505 passes through a parser 510, whereby coded image data to be decoded and coded information necessary for decoding are parsed. The encoded image data is output as dequantized data through the entropy decoder 520 and the dequantizer 530 , and restored as a residual value through the inverse transformer 540 . The residual value is restored by adding the residual value to an intra prediction result of the intra prediction unit 550 or a motion compensation result of the motion compensator 560 according to coding units. The restored coding unit is used for prediction of a next coding unit or a next frame through the deblocking unit 570 and the loop filtering unit 580 .

In order to perform decoding based on the image decoding method according to the embodiment of the present invention, the components of the image decoder 500 (i.e., parser 510, entropy decoder 520, inverse quantizer 530, inverse transformer 540, intra prediction unit 550, motion The compensator 560, the deblocking unit 570, and the loop filtering unit 580) perform an image decoding process based on a maximum coding unit, sub coding units according to depths, prediction units, and transformation units.

In particular, the intra prediction unit 550 and the motion compensator 560 determine a prediction unit and a prediction mode by sub-coding units by considering a maximum coding unit and a depth, and the inverse transformer 540 performs inverse transformation by considering a size of a transformation unit.

FIG. 6 illustrates a maximum coding unit, a sub coding unit, and a prediction unit according to an embodiment of the present invention.

The image encoding device 100 shown in FIG. 1 and the image decoding device 200 shown in FIG. 2 use hierarchical coding units to perform encoding and decoding by considering image characteristics. The maximum coding unit and the maximum depth may be adaptively set according to image characteristics, or differently set according to user's needs.

In FIG. 6 , the hierarchical coding unit structure 600 has a maximum coding unit 610 whose height and width are 64 and a maximum depth is 4. Referring to FIG. The depth increases along the vertical axis of the hierarchical coding unit structure 600, and as the depth increases, the height and width of the sub coding units 620 to 650 decrease. A maximum coding unit 610 and prediction units of sub coding units 620 to 650 are displayed along a horizontal axis of the hierarchical coding unit structure 600 .

The depth of the maximum coding unit 610 is 0, and the coding unit size (ie, height and width) is 64×64. The depth increases along the vertical axis, and there is a sub-coding unit 620 with a size of 32×32 and a depth of 1, a sub-coding unit 630 with a size of 16×16 and a depth of 2, and a sub-coding unit with a size of 8×8 and a depth of 3. A coding unit 640, and a sub-coding unit 650 having a size of 4×4 and a depth of 4. The sub coding unit 650 whose size is 4×4 and whose depth is 4 is a minimum coding unit.

Referring to FIG. 6 , examples of prediction units are shown along the horizontal axis according to each depth. That is, the prediction unit of the maximum coding unit 610 whose depth is 0 may be a prediction unit having a size equal to that of the coding unit 610 (ie, 64×64), or having a size smaller than the size of the coding unit 610 whose size is 64×64. The prediction unit 612 is 64×32, the prediction unit 614 is 32×64 in size, or the prediction unit 616 is 32×32 in size.

A prediction unit of a coding unit 620 having a depth of 1 and a size of 32×32 may be a prediction unit having a size equal to that of the coding unit 620 (ie, 32×32), or having a size smaller than the size of the coding unit 620 having a size of 32×32. A prediction unit 622 is 32×16, a prediction unit 624 is 16×32 in size, or a prediction unit 626 is 16×16 in size.

The prediction unit of the coding unit 630 whose depth is 2 and whose size is 16×16 may be a prediction unit having a size equal to that of the coding unit 630 (that is, 16×16), or having a size smaller than the size of the coding unit 630 whose size is 16×16. It is a 16×8 prediction unit 632 , a prediction unit 634 with a size of 8×16 or a prediction unit 636 with a size of 8×8.

A prediction unit of a coding unit 640 whose depth is 3 and whose size is 8×8 may be a prediction unit having a size equal to that of the coding unit 640 (that is, 8×8), or having a size smaller than the size of the coding unit 640 whose size is 8×8. is an 8×4 prediction unit 642 , a 4×8 prediction unit 644 , or a 4×4 prediction unit 646 .

Finally, a coding unit 650 having a depth of 4 and a size of 4×4 is a coding unit of a maximum depth. The prediction unit of the coding unit 650 may be a prediction unit having a size equal to that of the prediction unit 650 . However, even if a coding unit has the maximum depth, the size of the coding unit does not always need to be equal to the size of a prediction unit of the coding unit. Similar to the coding units 610 to 640 , the coding unit 650 may also be predicted by dividing the coding unit 650 into sub-coding units that are prediction units.

FIG. 7 illustrates coding units and transformation units according to an embodiment of the present invention.

The image encoding device 100 shown in FIG. 1 and the image decoding device 200 shown in FIG. 2 perform a maximum coding unit by itself or by dividing the maximum coding unit into at least one sub-coding unit whose size is equal to or smaller than the size of the maximum coding unit. to perform encoding and decoding. In the encoding and decoding process, the size of the transformation unit may be selected in such a manner as to maximize the compression rate regardless of the coding unit and the prediction unit. For example, referring to FIG. 7 , when a size of a current coding unit 710 is 64×64, transformation may be performed using a transformation unit 720 having a size of 32×32.

8a and 8b illustrate split shapes of coding units, prediction units, and transformation units according to an embodiment of the present invention.

In FIG. 8A , the left diagram illustrates a split shape selected by the image encoding apparatus 100 shown in FIG. 1 for encoding a maximum coding unit 810 . The image encoding apparatus 100 divides the maximum coding unit 810 into various shapes, performs encoding, and selects an optimal split shape by comparing encoding results of various split shapes with each other based on an R-D cost. When it is optimal to encode the maximum coding unit 810 as it is, as shown in FIGS. 8a and 8b , the maximum coding unit 810 may be encoded without dividing the maximum coding unit 810 .

In the left diagram of FIG. 8A , the maximum coding unit 810 is encoded by splitting the maximum coding unit 810 whose depth is 0 into sub-coding units whose depth is equal to or greater than 1. Referring to FIG. That is, the maximum coding unit 810 is divided into four depth-1 sub-coding units, and all or some depth-1 sub-coding units are divided into depth-2 sub-coding units.

A sub-coding unit located at an upper right side and a sub-coding unit located at a lower left side among sub-coding units having a depth of 1 are divided into sub-coding units having a depth equal to or greater than 2. Some sub-coding units having a depth equal to or greater than 2 may be divided into sub-coding units having a depth equal to or greater than 3.

In FIG. 8A , the right diagram illustrates split shapes of prediction units of a maximum coding unit 810 .

In the right diagram in FIG. 8A , the prediction unit 860 for the maximum coding unit 810 may be split differently from the maximum coding unit 810 . In other words, the prediction unit for each sub-coding unit may be smaller than the corresponding sub-coding unit.

For example, a prediction unit for a sub coding unit 854 located on a lower right side among sub coding units whose depth is 1 may be smaller than the sub coding unit 854 . Also, prediction units for some of the sub-coding units 814, 816, 850, and 852 of the sub-coding units 814, 816, 818, 828, 850, and 852 whose depth is 2 may be smaller than those of the sub-coding units 814, 816, 850, and 852, respectively. 852.

Also, prediction units for the sub coding units 822, 832, and 848 whose depth is 3 may be smaller than the sub coding units 822, 832, and 848, respectively. The prediction unit may have a shape in which each sub coding unit is bisected in a height or width direction or a shape in which each sub coding unit is quartered in a height or width direction.

Fig. 8b shows prediction units and transform units according to an embodiment of the present invention.

In FIG. 8B , a left diagram illustrates split shapes of prediction units for the maximum coding unit 810 shown in FIG. 8A , and a right diagram illustrates split shapes of transformation units of the largest coding unit 810 .

In the right diagram of FIG. 8B , the split shape of the transformation unit 870 may be set differently from the prediction unit 860 .

For example, even if a prediction unit for the coding unit 854 is selected in a shape in which the height of the coding unit 854 whose depth is 1 is halved, a transformation unit may be selected in the same size as the coding unit 854 . Similarly, even if the prediction units for the coding units 814 and 850 are selected in a shape in which the height of each of the coding units 814 and 850 whose depth is 2 is bisected, it is possible to use Each of the original sizes is the same size as the selected transform unit.

A transformation unit may be selected with a smaller size than a prediction unit. For example, when the prediction unit for the coding unit 852 is selected in a shape in which the width of the coding unit 852 having a depth of 2 is bisected, it may be selected in a shape in which the coding unit 852 is quartered in the direction of height and width. transform unit.

FIG. 9 is a block diagram of an image encoding device 900 according to another embodiment of the present invention. Referring to FIG. 9 , an image encoding apparatus 900 may be included in the image encoding apparatus 100 of FIG. 1 or the image encoder 400 of FIG. 4 to encode a current block based on inter prediction.

Referring to FIG. 9 , an image encoding apparatus 900 includes a motion vector determiner 910 and an encoder 920 .

In some encoding modes, the motion vector of the current block is determined based on the motion vector of at least one block that has been previously encoded, and the current block is encoded based on the determined motion vector. Examples of these coding modes are direct prediction mode and skip mode. In the direct prediction mode and the skip mode, the motion vector of the current block is determined based on information that has been previously encoded, and the motion vector of the current block is additionally encoded as information on the current block.

However, in direct prediction mode, a residual block generated by subtracting a prediction block generated using a motion vector from the current block is encoded as information on pixel values, however, in skip mode, the prediction block is regarded as the same as The current block is the same, and only flag information indicating that encoding is performed in skip mode is encoded as information on pixel values.

Although both the direct prediction mode and the skip mode are encoding modes based on inter prediction, since motion vectors are not separately encoded, the compression rate is high. However, in the direct prediction mode and the skip mode, since the current block is encoded by predicting the current block only in a specific direction, the hit rate and prediction accuracy of each of these modes may be low. Accordingly, when encoding a current block included in a bidirectionally predicted image according to a direct prediction mode or a skip mode, the image encoding apparatus 900 encodes the current block by performing prediction in various directions, which will be described in detail below.

FIG. 10 is a diagram illustrating a method of predicting a block included in a bidirectional predictive image according to an embodiment of the present invention.

Referring to FIG. 10 , a current block 1010 included in a current image 1000 as a bidirectional predictive image is predictively encoded by referring to at least one of a previous image before the current image 1000 and a subsequent image 1030 after the current image 1000 . Detect at least one block corresponding to the current block 1010 (for example, the previous block 1022 and the subsequent block 1032) from at least one of the previous image 1020 and the subsequent image 1030, and then predictively encode the current block 1010 based on the detected block 1022 or 1032 .

In other words, if the direction from the current image 1000 to the previous image 1020 is the first direction (L0 direction), the current block 1010 may be predicted based on the previous block 1022 detected from the previous image 1020 in the first direction (hereinafter, called "first-direction prediction"). Likewise, if the direction from the current image 1000 to the subsequent image 1030 is the second direction (L1 direction), the current block 1010 may be predicted based on the subsequent block 1032 detected from the subsequent image 1030 according to the second direction (hereinafter referred to as "Second Direction Forecast"). Also, both first-direction prediction and second-direction prediction may be performed on the current block 1010, that is, the current block 1010 may be bi-directionally predicted.

However, only when the current block 1010 is bi-predicted, the direct prediction mode according to the related art may be performed. In other words, when the current block 1010 is predictively encoded based on the first-direction prediction or the second-direction prediction, the current block 1010 cannot be encoded in the direct prediction mode. Also, only when the current block 1010 is bi-predicted, the skip mode according to the related art may be performed. In the skip mode according to the prior art, the motion vector of the current block 1010 determined based on the motion vectors of the previous blocks adjacent to the current block 1010 is the motion vector mv_L0 in the first direction, and the prediction block generated according to the first direction prediction Regarded as the same as the current block 1010, flag information indicating that encoding is performed according to a skip mode is encoded.

That is, according to the prior art, although first-direction prediction, second-direction prediction, and bidirectional prediction can be performed on a block included in a bidirectional predictive image, the direct prediction mode can be used only when prediction is performed in a specific direction and skip mode. Therefore, the use of the direct prediction mode and the skip mode, which lead to an improvement in the compression ratio, is limited.

To solve this problem, the image encoding apparatus 900 according to the current embodiment performs encoding according to an image encoding method in which a motion vector of the current block 1010 is determined based on previous encoding information, prediction may be performed in various methods without encoding the motion vector.

The motion vector determiner 910 determines a first direction motion vector and a second direction motion vector of the current block 1010 to perform first direction prediction, second direction prediction and bidirectional prediction. The first direction motion vector and the second direction motion vector are determined based on the motion vector of at least one block that has been previously coded before the current block 1010 is coded.

The first direction motion vector and the second direction motion vector may be determined using a method of determining motion vectors according to the prior art. For example, if the current block 1010 is encoded based on the direct prediction mode, the motion vector can be determined using the method of determining the motion vector according to the direct prediction mode based on H.264/AVC, wherein, in the direct prediction mode, the current block 1010 The information about the motion vector is not encoded, but the information about the residual block is encoded. If the current block 1010 is encoded based on the skip mode, the motion vector can be determined using the method of determining the motion vector according to the skip mode based on H.264/AVC, wherein, in the skip mode, the information on the motion vector and Information about the residual block is encoded.

Direct prediction modes based on H.264/AVC can be classified into temporal direct prediction modes and spatial direct prediction modes. Therefore, the motion vector determiner 910 may determine the motion vector by using the method of determining the motion vector in the H.264/AVC-based temporal direct prediction mode or by using the method of determining the motion vector in the H.264/AVC-based spatial direct prediction mode. A first directional motion vector and a second directional motion vector are determined. These methods will now be described in detail with reference to Figures 11a and 11b.

11a is a diagram illustrating a method of determining a motion vector of a current block based on a motion vector of a previously encoded block according to an embodiment of the present invention.

Referring to FIG. 11a, when encoding a current block 1010 included in a current image 1110 based on a temporal direct prediction mode, a motion vector mv_colA of a subsequent block 1120 may be used to generate a motion vector of the current block 1010, wherein the subsequent block 1120 is included in In a subsequent image 1114 after the current image 1110 and juxtaposed with the current block 1010 .

In other words, if the subsequent image 1114 is an anchor image, the first direction motion vector mv_LOA and the second direction motion vector mv_LOA of the current block 1010 may be determined using the motion vector mv_colA of the subsequent block 1120 included in the anchor image 1114 mv_L1A. can be determined by scaling the motion vector mv_colA of the subsequent block 1120 based on the temporal distance between the current image 1110 and the previous image 1112 preceding the current image 1110 and the temporal distance between the anchor image 1114 and the previous image 1112 The first direction motion vector mv_L0A and the second direction motion vector mv_L1A of the current block 1010 .

For example, if the motion vector mv_colA of the subsequent block 1120 is generated for the block 1122 detected from the previous image 1112, the following equation may be used to generate the first direction motion vector mv_L0A and the second direction motion vector mv_L1A of the current block 1110:

mv_LOA=(t1/t2)×mv_colA

mv_L1A=mv_L0A-mv_colA,

Wherein, "mv_LOA" and "mv_L1A" represent the first-direction motion vector and the second-direction motion vector of the current block 1010, respectively.

11b is a diagram illustrating a method of determining a motion vector of a current block based on a motion vector of a previously encoded block according to another embodiment of the present invention.

Referring to FIG. 11b , the first and second direction motion vectors of the current block 1010 may be determined based on motion vectors mv_A, mv_B, and mv_C of blocks 1130 , 1132 , and 1134 adjacent to the current block 1100 . A first-direction motion vector and a second-direction motion vector of the current block 1010 may be determined based on motion vectors of blocks located to the left, upper, and uppermost right of the current block 1010 .

When the current block 1010 is encoded according to the spatial direct prediction mode, the median value among at least one first-direction motion vector among the motion vectors mv_A, mv_B, and mv_C is determined as the first-direction motion vector of the current block 1010, the motion vector A median value among at least one second-direction motion vector among mv_A, mv_B, and mv_C is determined as the second-direction motion vector of the current block 1010 . The first direction motion vector is determined by calculating the median value between the at least one first direction motion vector by setting the at least one second direction motion vector to "0", by setting the at least one first direction motion vector The motion vector is set to "0" to calculate a median value between at least one motion vector in the second direction to determine the motion vector in the second direction.

Among the reference pictures determined by the motion vectors mv_A, mv_B and mv_C, the picture corresponding to the smallest first-direction reference index is determined as the first-direction reference picture, and among the reference pictures determined by the motion vectors mv_A, mv_B and mv_C, the picture corresponding to the smallest first-direction reference index is determined as the first-direction reference picture. The picture corresponding to the two-direction reference index is determined as the second-direction reference picture.

When the current block 1010 is encoded according to the skip mode, as shown in FIG. The median value of can be determined as the motion vector of the current block 1010 .

If the image coding apparatus 900 of FIG. 9 determines the motion vector of the current block as described above with reference to FIG. 11a or 11b and encodes the current block, the decoding end determines the motion vector of the current block as described above with reference to FIG. 11a or 11b and restore the current image. In other words, the image encoding end and the image decoding end may implicitly share the same method for determining the motion vector, and determine the motion vector of the current block according to the shared method.

However, according to another embodiment of the present invention, the image encoding end may generate a plurality of motion vector candidates based on motion vectors of a plurality of previous blocks encoded before the current block is encoded, specifying a motion vector among the plurality of motion vector candidates The information of the motion vector of the current block is encoded explicitly, and the encoded result is inserted into the bitstream. In this case, the image decoding end may determine the motion vector of the current block based on the result of the encoding inserted into the bitstream. Such a method will now be described with reference to Figures 12a and 12b.

12a and 12b are diagrams illustrating a method of determining a motion vector of a current block based on a motion vector of a previously encoded block according to an embodiment of the present invention.

Referring to FIG. 12a, the motion vector determiner 910 of FIG. 9 may generate a plurality of motion vector candidates based on motion vectors of previously encoded blocks adjacent to the current block. The plurality of motion vector candidates may be the motion vector of the leftmost block _a0 among the blocks adjacent to the upper side of the current block, the motion vector of the uppermost block _b0 adjacent to the left side of the current block, and the motion vector of the uppermost block b0 adjacent to the left side of the current block, The motion vector of block c adjacent to the upper right side of , the motion vector of block d adjacent to the upper left side of the current block, and the motion vector of block e adjacent to the lower right side of the current block.

One of the first direction motion vectors among the motion vectors of the leftmost block a ₀ , the uppermost block b ₀ , the block c, and the block d may be determined as the first direction motion vector of the current block. One of the second direction motion vectors among the motion vectors of the leftmost block a ₀ , the uppermost block b ₀ , the block c, and the block d may be determined as the second direction motion vector of the current block. For example, if the motion vectors of the leftmost block a ₀ and the uppermost block b ₀ among the leftmost block a ₀ , uppermost block b ₀ , block c, and block d are the first direction motion vector, then the leftmost block One of the motion vectors of a ₀ and the uppermost block b ₀ may be determined as the first direction motion vector of the current block, and one of the motion vectors of the blocks c and d may be determined as the second direction motion vector of the current block.

Various methods can be used to determine one of the plurality of motion vector candidates as the motion vector of the current block, however, the motion vector of the current block can generally be selected in such a way that the current block can be more precisely selected based on the motion vector predicted. If the current block is more accurately predicted when the motion vector of the leftmost block a ₀ is used among the motion vectors of the leftmost block a ₀ and the uppermost block b ₀ used as the first direction motion vector, the leftmost The motion vector of block _a0 may be determined as the first direction motion vector of the current block. Predicted blocks obtained by predicting the current block using motion vectors of the plurality of motion vector candidates, respectively, may be subtracted from the current block to generate residual blocks, each of which has a sum of absolute values (SAD) may be calculated, and then, a motion vector having the smallest SAD among the plurality of motion vector candidates may be determined as the motion vector of the current block.

Referring to FIG. 12b, motion vectors of all blocks adjacent to a current block may be used as a plurality of motion vector candidates. In other words, not only motion vectors of blocks adjacent to the upper side of the current block (including the leftmost block a ₀ ) but also motion vectors of blocks adjacent to the left side of the current block (including the uppermost block b ₀ ) can be are used as the plurality of motion vector candidates.

As described above with reference to FIG. 12b, one of the first-direction motion vectors and one of the second-direction motion vectors among the plurality of motion vector candidates of all blocks adjacent to the current block can be determined as the first-direction motion of the current block, respectively. vector and the motion vector in the second direction.

Therefore, referring to FIGS. 11a, 11b, 12a, and 12b, the motion vector determiner 910 may determine one of a plurality of motion vector candidates as the motion vector of the current block by using the following equation:

C={median(mv_a0,mv_b0,mv_c),mv_a0,mv_a1...,mv_aN,mv_b0,mv_b1,...,mv_bN,mv_c,mv_d,mv_e,mv_L0A,mv_L1A},

Wherein, "mv_LOA" and "mv_L1A" may represent motion vectors obtained according to the method in Fig. 11a, and "median(mv_a0, mv_b0, mv_c)" may represent motion vectors obtained according to the method in Fig. 11b. Other motion vectors are described above with reference to Figures 12a and 12b.

The first directional motion vector of the current block may be determined from the first directional motion vector included in the C set described above, and the second directional motion vector of the current block may be determined from the second directional motion vector included in the C set.

According to another embodiment of the present invention, the total number of motion vector candidates may be determined to be small, and the motion vector of the current block may be determined from the motion vector candidates based on the following equation:

C={median(mv_a',mv_b',mv_c'),mv_a',mv_b',mv_c',mv_L0A,mv_L1A},

Among them, "mv_x" indicates the motion vector of block x, "median()" indicates the median value, and "mv_a" indicates the first valid motion vector among the motion vectors mv_a0, mv_a1, . . . , mv_aN. For example, if block a0 is coded using inter prediction, the motion vector mv_a0 of block a0 is invalid, so mv_a'=mv_a1. If the motion vector mv_a1 of block a1 is also invalid, then mv_a'=mv_a2.

Likewise, "mv_b" indicates the first effective motion vector among the motion vectors mv_b0, mv_b1, ..., mv_bN, and "mv_c" indicates the first effective motion vector among the motion vectors mv_c, mv_d, ..., mv_e.

"mv_LOA" and "mv_L1A" may represent motion vectors obtained according to the method of FIG. 11a. According to another embodiment of the present invention, instead of motion vectors mv_L0A and mv_L1A, motion vectors generated using temporal distances between adjacent images may be used as motion vector candidates. This embodiment will be described in detail with reference to Figures 13a and 13b.

13a and 13b are diagrams illustrating a method of determining a motion vector of a current block based on a motion vector of a previously encoded block according to an embodiment of the present invention.

Referring to FIG. 13a, temporal distance-based motion vector candidates may be determined differently from the temporal direct prediction mode according to the related art.

A motion vector candidate of the current block 1010 included in the current image 1110 may be generated from a motion vector of a block 1320 included in the previous image 1112 preceding the current image 1110 and collocated with the current block 1010 . For example, if the motion vector mv_colB of the block 1320 collocated with the current block 1010 is generated for the block 1322 detected from the subsequent image 1114, the following equations may be used to generate the motion vector candidates mv_LOB and mv_L1B of the current block 1010:

mv_L1B=(t3/t4)×mv_colB

mv_L0B=mv_L1B-mv_colB,

Wherein, “mv_LOB” represents a first-direction motion vector candidate of the current block 1010 , and “mv_L1B” represents a second-direction motion vector candidate of the current block 1010 .

Referring to FIG. 13b , a motion vector candidate mv_LOC of a current block 1010 included in a current image 1110 may be determined using a motion vector mv_colC of a block 1330 included in a previous image 1112 and collocated with the current block 1010 . For example, when the motion vector mv_colC of the block 1330 collocated with the current block 1010 is generated for the block 1332 detected from another previous image 1116, the following equation may be used to generate the motion vector candidate mv_LOC of the current block 1010:

mv_LOC=(t6/t5)×mv_colC.

The motion vector candidates mv_L0B and mv_L1B described above with reference to FIGS. 13a and 13b may be included in the above-mentioned C set of motion vector candidates of the current block 1010 described above. That is, motion vector candidates mv_LOB and mv_L1B may be included in C-set instead of motion vector candidates mv_LOA and mv_L1A, or motion vector candidates mv_LOB and mv_L1B may be additionally included in C-set.

Referring back to FIG. 9 , if the motion vector determiner 910 determines a motion vector of a current block based on a motion vector of a previously encoded block, the encoder 920 encodes information on the current block.

If the first direction motion vector and the second direction motion vector of the current block are determined, the encoder 920 performs first direction prediction, second direction prediction and bidirectional prediction based on the first direction motion vector and the second direction motion vector. Then, a prediction direction in which the current block can be most accurately predicted is determined. A predicted block obtained by performing first-direction prediction, second-direction prediction, and bidirectional prediction may be compared with a current block, and a direction that may generate a residual block having the smallest SAD may be determined as a prediction direction. Otherwise, the prediction direction may be determined by performing first-direction prediction, second-direction prediction, and bidirectional prediction, encoding the performed results, and comparing the encoded results with each other based on the RD cost.

After the prediction direction is determined, information on the prediction direction and information on pixel values are encoded. Information about the prediction direction can be encoded as sequence parameters, slice parameters or block parameters. Predetermined bins may be allocated to the first direction, the second direction, and the bidirectional, bins corresponding to a prediction direction in which the current block is encoded may be entropy encoded, and then a result of the entropy encoding may be inserted into a bitstream.

Information on the residual block (eg, direct prediction mode) may be additionally encoded as information on pixel values. A residual block may be obtained by subtracting a prediction block generated by performing prediction in a determined prediction direction from a current block, frequency domain coefficients may be generated by performing transformation on the residual block, and then the frequency domain coefficients may be quantized and entropy coding. The transform may be a DCT transform or a KLT transform.

Also, as with information on pixel values, only a skip mode according to the related art and flag information indicating that a predicted block and a current block are identical may be encoded. That is, since a prediction block obtained by performing prediction in a determined prediction direction is considered to be the same as a current block, only skip mode and flag information may be encoded.

A reference image for predicting a current block may be inferred based on a reference index of a previously coded block. As described above with reference to FIG. 11b , a reference index of a current block and a reference image corresponding to the current block may be determined based on reference indices of previously encoded blocks. Also, a reference image corresponding to the motion vector determined by the motion vector determiner 910 may be determined as a reference image of the current block. For example, as described above with reference to FIG. 11a, if the motion vector mv_LOA is determined as the first direction motion vector of the current block, the previous image 1112 according to the motion vector mv_LOA may be automatically determined as the reference image of the current block.

If the motion vector determiner 910 explicitly determines the first direction motion vector and the second direction motion vector among multiple motion vector candidates as described above with reference to FIGS. 12a, 12b, 13a and 13b, the encoder 920 can Information indicating at least one of the motion vector in the first direction and the motion vector in the second direction is encoded.

If the prediction direction is determined to be the first direction, information indicating only the first direction motion vector among the plurality of motion vector candidates is encoded. If the prediction direction is determined to be the second direction, only information indicating the second direction motion vector may be encoded. However, if the prediction direction is determined to be bidirectionally performed, information indicating a first-direction motion vector and a second-direction motion vector among the plurality of motion vector candidates may be encoded.

Binary numbers may be assigned to all motion vectors included in the above-mentioned set C, respectively, and binary numbers assigned to a first-direction motion vector and/or a second-direction motion vector determined to be a motion vector of a current block may be encoded. However, since one of the first directional motion vectors in the C set is determined as the first directional motion vector of the current block as described above, the first directional motion vector assigned to the current block among the binary numbers assigned to the first directional motion vectors in the C set A binary number of a motion vector in one direction may be encoded as information indicating a motion vector in a first direction. Also, the binary number assigned to the second-direction motion vector of the current block among the binary numbers assigned to the second-direction motion vector in the C set may be encoded as information indicating the second-direction motion vector.

FIG. 14 is a block diagram of an image decoding device 1400 according to another embodiment of the present invention. The image decoding apparatus 1400 may be included in the image decoding apparatus 200 of FIG. 2 or the image decoding apparatus 500 of FIG. 5 , and may decode a current block by performing inter prediction.

Referring to FIG. 14 , an image decoding device 1400 includes a decoder 1410, a motion vector determiner 1420, and a restoration unit 1430.

The decoder 1410 receives a bitstream including data on a current block, and decodes information on a prediction direction in which the current block is predicted. The image decoding apparatus 1400 determines a motion vector of a current block based on a motion vector of at least one block that has been previously decoded before the current block is decoded, and performs an operation on the current block based on the determined motion vector in a decoding mode for restoring the current block. to decode. Therefore, the information on the prediction direction is decoded before the motion vector of the current block is determined. The information on the prediction direction may be, for example, sequence parameters, slice parameters or block parameters included in the bitstream.

In an encoding/decoding mode (for example, direct prediction mode or skip mode) in which a motion vector of a current block is determined based on a motion vector of a previously encoded/decoded block, the current block is encoded/decoded based only on prediction performed in a specific direction . However, in an encoding/decoding direction according to an embodiment of the present invention, a current block may be encoded/decoded in a direct prediction mode or a skip mode based on a result of performing prediction in a first direction, a second direction, and a bidirectional direction. For this, the image encoding apparatus 900 of FIG. 9 encodes information on a prediction direction, and inserts the encoded result into a bitstream. The decoder 1410 decodes information on a prediction direction inserted into a bitstream by using the image encoding device 900 .

The decoder 1410 may decode information indicating a motion vector of a current block among a plurality of motion vector candidates together with information about a prediction direction. If the motion vector of the current block is implicitly determined based on the method of determining the motion vector shared between the encoding side and the decoding side as described above, the current block is determined based on the shared method and information indicating that the motion vector of the current block does not need to be decoded The total number of one motion vector for the block. However, when a plurality of motion vector candidates are generated based on motion vectors of a plurality of previously encoded blocks as described above and a motion vector of a current block is determined from the plurality of motion vector candidates, the current one of the plurality of motion vector candidates is indicated. The information of the motion vector of the block can be decoded.

Also, the decoder 1410 decodes information on pixel values of the current block. In this case, information about the residual block of the current block or flag information indicating that the current block is the same as the prediction block is decoded. The decoder 1410 may restore the residual block by performing entropy decoding, inverse quantization, or inverse transformation on information about the residual block.

The motion vector determiner 1420 determines the motion vector of the current block based on the result of decoding the information on the prediction direction by using the decoder 1410, thereby predicting the current block in a predetermined direction. If it is determined to perform prediction in the first direction based on the result of the decoding, the first direction motion vector is determined as the motion vector of the current block. If it is determined to perform prediction in the second direction based on the result of the decoding, the second direction motion vector is determined as the motion vector of the current block. If it is determined that prediction is bidirectionally performed based on a result of the decoding, a first-direction motion vector and a second-direction motion vector are determined as motion vectors of the current block.

When the motion vector of the current block is to be implicitly determined according to the method of determining the motion vector shared between the encoding side and the decoding side, the motion vector of the current block is determined according to the shared method. This method is described above with reference to Figures 11a and 11b.

However, when the motion vector of the current block is to be determined from a plurality of motion vector candidates, the motion vector of the current block is determined based on information indicating the motion vector of the current block decoded by the decoder 1410 among the plurality of motion vector candidates . The plurality of motion vector candidates is generated using a motion vector of at least one block that has been decoded before the current block is decoded, and the motion vector of the current block is determined from the plurality of motion vector candidates.

If the current block is predicted only in the first direction, the first direction motion vector is determined from the plurality of motion vector candidates based on the information indicating the first direction motion vector decoded using the decoder 1410 . If the current block is predicted only in the second direction, the second direction motion vector is determined from the plurality of motion vector candidates based on the information indicating the second direction motion vector decoded using the decoder 1410 . If the current block is bidirectionally predicted, based on information indicating a motion vector in the first direction and information indicating a motion vector in the second direction decoded using the decoder 1410, the plurality of motion vector candidates determine a motion vector in the first direction and a motion vector in the second direction. Direction motion vector.

The restoration unit 1430 restores the current block based on the information on the pixel value decoded using the decoder 1410 and the motion vector of the current block determined using the motion vector determiner 1420 .

A prediction block is generated by predicting a current block in a first direction, a second direction, or bidirectionally according to a motion vector determined using the motion vector determiner 1420 . If a residual block is restored by decoding information on pixel values using the decoder 1410, the restored residual block is combined with a prediction block to restore a current block. If the decoder 1410 decodes the flag information indicating that the predicted block is the same as the current block, the predicted block may be directly used as the restored current block.

A reference image for predicting a current block may be referenced based on a reference index of a previously decoded block. As described above with reference to FIG. 11b , a reference index of a current block and a reference image corresponding to the current block may be determined based on reference indices of previously decoded blocks. Also, a reference image corresponding to the motion vector determined by the motion vector determiner 1420 may be determined as a reference image of the current block. For example, as described above with reference to FIG. 11a, if the motion vector mv_LOA is determined as the first direction motion vector of the current block, the previous image 1112 according to the motion vector mv_LOA may be automatically determined as the reference image of the current block.

FIG. 15 is a flowchart illustrating an image encoding method according to an embodiment of the present invention.

Referring to FIG. 15 , in operation 1510, the image encoding apparatus determines a motion vector of a current block. For example, in operation 1510, a first direction motion vector and a second direction motion vector of a bidirectionally predicted current block included in a current picture are determined. As described above with reference to FIGS. 11 a and 11 b , a motion vector of a current block may be implicitly determined according to a method of determining a motion vector shared between an encoding end and a decoding end. In addition, as described above with reference to FIGS. 12a, 12b, 13a, and 13b, a plurality of motion vector candidates are generated, and the first direction motion vector and the second direction motion vector may be determined from the plurality of motion vector candidates.

In operation 1520, the image encoding apparatus determines a prediction method from among the first direction prediction, the second direction prediction, and the bidirectional prediction to encode the current block. The prediction method is determined from the first direction prediction using the first direction motion vector, the second direction prediction using the second direction motion vector, and the bidirectional prediction using both the first direction motion vector and the second direction motion vector to predict the current block to encode.

A prediction method that more accurately predicts a current block among first-direction prediction, second-direction prediction, and bidirectional prediction may be used to encode the current block. As described above, a prediction method that obtains the minimum SAD between a prediction block and a current block may be used to encode the current block.

In operation 1530, the image encoding apparatus encodes information about a prediction direction used in the prediction method determined in operation 1520. Bins may be assigned to the first direction, the second direction, and bidirectional, respectively, and bins assigned to a prediction direction for encoding a current block may be encoded. The encoded binary numbers may be entropy encoded according to a predetermined entropy encoding method. Information about the prediction direction can be encoded as sequence parameters, slice parameters or block parameters.

Information about the pixel values of the current block is also encoded. In this case, information on a residual value of a residual block obtained by subtracting a prediction block from a current block may be encoded. The residual block can be transformed, quantized and entropy coded. Also, as described above, only flag information indicating that the predicted block is the same as the current block may be encoded.

Information defining a motion vector among a plurality of motion vector candidates for encoding the current block may also be encoded. If the plurality of motion vector candidates are generated based on the motion vector of at least one block encoded before the current block is encoded, and one of the plurality of motion vector candidates is selected and used as the motion vector of the current block, indicating Information of the selected motion vector may be encoded. If the first direction prediction is determined as the prediction method in operation 1520, only information about the first direction motion vector may be encoded. If the second direction prediction is determined as the prediction method in operation 1520, only information about the second direction motion vector may be encoded. If bidirectional prediction is determined as a prediction method in operation 1520, only information on both the first-direction motion vector and the second-direction motion vector is encoded.

FIG. 16 is a flowchart illustrating an image decoding method according to an embodiment of the present invention.

Referring to FIG. 16 , in operation 1610, the image decoding apparatus decodes information about a prediction direction to be used to decode a current block among a first direction, a second direction, and a bidirectional direction. In this case, information on the prediction direction included in the bitstream may be decoded. The binary number assigned to the prediction direction among the binary numbers respectively assigned to the first direction, the second direction, and the bidirectional direction may be decoded. The information on the prediction direction included in the bitstream can be decoded as sequence parameters, slice parameters or block parameters.

In operation 1610, the image decoding apparatus also encodes information on pixel values of the current block. Information about a residual value of a residual block of a current block may be encoded. Information on the residual block may be entropy decoded, dequantized, and inversely transformed to restore the residual block. Also, information indicating that the current block is the same as the predicted block may be decoded.

When the motion vector of the current block is determined from a plurality of motion vector candidates as described above, information defining a motion vector to be used for decoding the current block among the plurality of motion vector candidates may be decoded. If a result of decoding the information on the prediction direction shows that the prediction direction is the first direction, the information indicating the first direction motion vector of the current block among the plurality of motion vector candidates may be decoded. If a result of decoding the information on the prediction direction shows that the prediction direction is the second direction, the information indicating the second direction motion vector of the current block among the plurality of motion vector candidates may be decoded. If the result of decoding the information on the prediction direction shows that the prediction direction is bidirectional, the information indicating the motion vector in the first direction among the plurality of motion vector candidates and the information indicating the motion vector in the second direction among the plurality of motion vector candidates may be compared. The information of the directional motion vector is decoded.

In operation 1620, the image decoding apparatus determines a prediction direction in which a current block is to be predicted based on the information on the prediction direction decoded in operation 1610. A prediction direction in which a current block is to be predicted is determined based on a binary number corresponding to a prediction direction, information of which is decoded in operation 1610 . If a prediction direction in which the current block is to be predicted is determined, at least one motion vector for predicting the current block in the determined prediction direction is determined.

As described above, the motion vector of the current block may be implicitly determined according to the method of determining the motion vector shared between the encoding end and the decoding end. If the motion vector of the current block is explicitly selected from a plurality of motion vector candidates, the plurality of motion vector candidates are generated based on at least one block motion vector decoded before the current block is decoded, and the motion vector is selected from the plurality of motion vector candidates. The motion vector of the current block is determined from the vector candidates. The motion vector of the current block may be determined based on the information indicative of the motion vector of the current block among the plurality of motion vector candidates decoded in operation 1610 . At least one of the first direction motion vector and the second direction motion vector may be determined based on the information decoded in operation 1610 .

In operation 1630, the image decoding apparatus restores a current block. A prediction block of the current block is determined based on the motion vector of the current block determined in operation 1620, and the current block is restored based on the prediction block.

A prediction block of the current block may be generated by performing one of first direction prediction, second direction prediction, and bidirectional prediction based on the first direction motion vector and/or the second direction motion vector determined in operation 1620 .

If the residual block is restored in operation 1610, the prediction block is combined with the restored residual block to restore the current block. If information indicating that the predicted block is the same as the current block is decoded, the predicted block is used as the current block.

Although the present invention has been specifically shown and described with reference to the exemplary embodiments of the invention and the accompanying drawings, those skilled in the art will understand that the exemplary embodiments can be modified without departing from the principle and spirit of the present invention. Instead, the scope of the invention is defined in the claims and their equivalents. Also, a system according to an embodiment of the present invention can be realized as computer readable codes in a computer readable recording medium.

For example, an image encoding device, an image decoding device, a motion vector encoding device, and a motion vector decoding device according to exemplary embodiments of the present invention may be included in the same as those shown in FIGS. 1, 2, 4, 5, 9, and 14. Shown is a bus to which units of the device are connected and at least one processor connected to the bus. Each of the devices may further include a memory connected to the bus to store commands, received messages or generated messages, and connected to the at least one processor for performing the above operations.

The computer-readable recording medium may be any recording device capable of storing data read by a computer system, for example, read-only memory (ROM), random-access memory (RAM), compact disk (CD)-ROM, magnetic tape, floppy disk, Optical data storage devices, etc. The computer-readable recording medium can be distributed in computer systems inline through a network, and the present invention can be stored and implemented as computer-readable codes in the distributed system.

Claims (1)

1. a picture decoding method, comprising:

Obtain about first direction, second direction and two-way among the information of prediction direction for decoding to current block;

Based on obtain the information about prediction direction, determine to current block carry out predict by according to prediction direction;

Determine at least one motion vector current block predicted according to the prediction direction determined;

According to the prediction direction determined, based at least one motion vector of the current block determined, recover current block.