Linear prediction-based video encoding / decoding method, device and recording medium

Abstract

An image encoding / decoding method, device and recording medium of the present disclosure may comprise: a step of acquiring picture level information of the current picture; a step of listing temporal reference motion information of the current picture for each temporal reference motion information unit of the current picture so as to generate a temporal reference motion information field of the current picture; and an inter-prediction step of performing inter-prediction for each block unit of the current picture on the basis of the temporal reference motion information field of the current picture.

Description

Linear prediction-based video encoding / decoding method, device, and recording medium

[0001] The present disclosure may be a technology for improving the prediction accuracy of a linear prediction method among inter-screen prediction methods in a picture decoding process.

[0002] Video data, a vast volume of data, is being compressed to reduce its size for efficient storage and transmission. Various techniques are being studied to analyze the statistical characteristics of redundant signals and utilize these characteristics to apply them to video compression technologies. Furthermore, effective methods are being developed for transmitting the information required for the decoder to restore signals removed during the encoding process.

[0003] During the process of decoding a picture, inter-screen prediction can be performed on a prediction block-by-prediction basis. Inter-screen prediction can be performed by weighting and summing prediction signals generated using one or more pieces of motion information (motion vectors, reference picture indices, etc.) on a prediction block-by-prediction basis.

[0004] The present disclosure aims to improve the efficiency of video encoding / decoding through inter-screen prediction based on linear prediction of video.

[0005] The video encoding / decoding method, device, and recording medium of the present disclosure may include a step of obtaining picture level information of a current picture; a step of listing temporal reference motion information of the current picture in units of temporal reference motion information units of the current picture to generate a temporal reference motion information field of the current picture; and a step of performing inter-screen prediction in units of blocks of the current picture based on the temporal reference motion information field of the current picture.

[0006] In the video encoding / decoding method, device, and recording medium of the present disclosure, the picture level information may include a picture type, a picture quantization parameter, a reference picture list, and a picture level linear prediction mode.

[0007] In the video encoding / decoding method, device and recording medium of the present disclosure, the listing of the temporal reference motion information can be performed in a raster scan order.

[0008] In the video encoding / decoding method, device, and recording medium of the present disclosure, the step of generating the temporal reference motion information field comprises: a motion reference picture selection step of selecting a motion reference picture to refer to a motion information field for generating the temporal reference motion information field; a motion information field spatial resolution changing step of changing the spatial resolution of a motion information field of the motion reference picture; a motion information projection step of the motion reference picture, in which the motion information of the motion reference picture is designated as temporal reference motion information of a temporal reference motion information unit of the current picture including a projected pixel obtained by projecting the motion information onto the current picture, thereby generating a temporal reference motion information field of the current picture before modification;

[0009] It may include a temporal reference motion information modification step of scaling the motion vector of the temporal reference motion information field before modification and filling in the empty units of the first temporal reference motion information field to obtain the temporal reference motion information field of the current picture.

[0010] In the video encoding / decoding method, device, and recording medium of the present disclosure, the motion reference picture selection step is performed at the picture level, and a subsequent step of the motion reference picture selection step may be performed at a lower level rather than the picture level.

[0011] In the video encoding / decoding method, device, and recording medium of the present disclosure, the motion reference picture selection step can be performed by considering the POC (Picture Order Count) value of the current picture and the motion reference picture.

[0012] In the video encoding / decoding method, device, and recording medium of the present disclosure, the motion information field spatial resolution changing step can be performed by comparing the spatial resolution of the motion reference picture and the spatial resolution of the current picture.

[0013] In the video encoding / decoding method, device and recording medium of the present disclosure, in the motion information projection step of the motion reference picture, in response to the presence of two or more motion information in the temporal reference motion information unit, the motion information can be excluded from the generation of the pre-modification temporal reference motion information field regardless of the similarity of the motion information.

[0014] The present disclosure is expected to have the effect of improving the inter-screen prediction accuracy based on linear prediction of video.

[0015] FIG. 1 illustrates an embodiment of a video decoder of the present disclosure.

[0016] Figure 2 illustrates an embodiment of a motion information unit and a motion information field.

[0017] Figure 3 illustrates an embodiment of a temporal reference motion information unit and a temporal reference motion information field.

[0018] Figure 4 illustrates the execution process of the temporal reference motion information field generation unit.

[0019] Figure 5 illustrates one embodiment of the performance of the motion reference picture selection step.

[0020] Figure 6 illustrates a step of changing the spatial resolution of a motion information field.

[0021] Figure 7 illustrates one embodiment of performing a motion information field spatial resolution change step.

[0022] Figure 8 illustrates a step of deriving motion information with changed spatial resolution.

[0023] Figure 9 illustrates an example of a temporal reference motion information sampling position.

[0024] Figure 10 illustrates an example of motion information of multiple units within a unit group.

[0025] Figure 11 illustrates the process of performing the motion information projection step of the reference picture.

[0026] Figure 12 shows an example of motion information projection order 2.

[0027] Figure 13 illustrates an example of a case where the prediction quality reliability of a motion vector is low.

[0028] Figure 14 illustrates one embodiment of the step of deriving a motion vector projection position.

[0029] Figure 15 illustrates the process of performing the temporal reference motion information modification step.

[0030] Figure 16 illustrates an example 1 of performing a step of filling in a temporal reference motion information field.

[0031] Figure 17 illustrates an example 2 of performing a step of filling in a temporal reference motion information field.

[0032] Figure 18 illustrates the process of performing inter prediction on a block-by-block basis in the inter prediction unit.

[0033] Figure 19 illustrates an embodiment of a method for deriving a motion vector when the linear prediction mode is performed on the entire picture area.

[0034] Figure 20 illustrates an example of deriving a prediction signal using motion information.

[0035] Figure 21 illustrates the second temporal reference motion vector candidate location.

[0036] Figure 22 illustrates the process of performing the linear motion information derivation step.

[0037] Figure 23 illustrates an example of a result of performing a temporal reference motion information unit selection step.

[0038] Figure 24 shows an example of a linear motion valid section.

[0039] The present invention is susceptible to various modifications and embodiments. Specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, but rather to encompass all modifications, equivalents, and alternatives falling within the spirit and technical scope of the present invention. Throughout the description of each drawing, similar reference numerals have been used to designate similar components.

[0040] While terms such as "first" and "second" may be used to describe various components, these components should not be limited by these terms. These terms are used solely to distinguish one component from another. For example, without departing from the scope of the present invention, a first component may be referred to as a "second component," and similarly, a second component may also be referred to as a "first component." The term "and / or" includes a combination of multiple related items described herein or any of multiple related items described herein.

[0041] When a component is referred to as being "connected" or "connected" to another component, it should be understood that it may be directly connected or connected to that other component, but that there may be other components intervening. Conversely, when a component is referred to as being "directly connected" or "connected" to another component, it should be understood that there are no other components intervening.

[0042] The terminology used in this application is only used to describe specific embodiments and is not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this application, it should be understood that the terms "comprise" or "have" indicate the presence of a feature, number, step, operation, component, part, or combination thereof described in the specification, but do not preclude the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

[0043] Hereinafter, with reference to the attached drawings, preferred embodiments of the present invention will be described in more detail. Hereinafter, identical components in the drawings will be designated by the same reference numerals, and redundant descriptions of identical components will be omitted.

[0044]

[0045] FIG. 1 illustrates an embodiment of a video decoder of the present disclosure.

[0046] The picture level information setting section of Fig. 1 can store information used in the entire restoration process of the current picture in memory through a process of parsing or deriving the information.

[0047] The temporal reference motion information field generation unit of FIG. 1 can generate a temporal reference motion information field, which is a list of temporal reference motion information that can be used in a motion vector restoration process in units of prediction blocks within the current picture, and this process can be performed at the picture or large unit / block or coding unit / block or prediction unit / block level.

[0048] The entropy decoding unit of Fig. 1 can entropy decode information that can be used in the decoding process of various hierarchical units that may exist in a sequence, such as a sequence, a picture, a slice, a tile, a large unit, a coding unit, a prediction unit, and a transform unit. Entropy decoding can use various decoding methods, such as, for example, Exponential Golomb, Context-Adaptive Variable Length Coding (CAVLC), and Context-Adaptive Binary Arithmetic Coding (CABAC).

[0049] The inverse quantization unit of Fig. 1 can perform inverse quantization on the quantization coefficients of the transform block based on the quantization parameters.

[0050] The inverse transform unit of Fig. 1 can perform inverse transform on the inverse quantized transform coefficients of the transform block. The transform can be performed by various types of transform methods such as DCT, DST, and KLT, and the transform kernel type can be transmitted or derived by a set rule. The transform kernel can be determined based on at least one of information regarding the prediction mode (inter / intra prediction), block size / shape, intra prediction mode, component type (luminance / chrominance component), and segmentation type (QT, BT, TT, etc.).

[0051] The intra prediction unit of FIG. 1 can perform prediction using information restored within the same picture in units of prediction blocks. The restored information may be information on surrounding reference samples or information on surrounding reference blocks in units of prediction blocks. In addition, the availability of the surrounding reference sample information or the information on the surrounding reference blocks can be determined based on a prediction mode of the surrounding reference sample information or the information on the surrounding reference blocks. If the surrounding reference sample information or the information on the surrounding reference blocks is not available, it can be replaced and used with information on other surrounding reference samples or other surrounding reference blocks.

[0052] In intra prediction, the prediction mode may include a directional prediction mode that uses reference pixel information according to the prediction direction, and a non-directional mode that does not use directional information when performing prediction. The mode for predicting the luminance component and the mode for predicting the chrominance component may be different, and the chrominance component may be predicted using the intra prediction mode used to predict the luminance component or the predicted / restored luminance component.

[0053] The inter prediction unit of Fig. 1 can perform prediction using information restored within another picture in units of prediction blocks. In some cases, information restored within the same picture can be used together. The other picture may be at least one picture from among the previous picture or the next picture of the current picture including the current prediction block unit. In order to perform inter prediction, it is possible to determine whether the motion prediction method of the corresponding prediction block unit is Skip Mode, Merge Mode, AMVP Mode, or Current Picture Reference Mode based on the prediction block unit.

[0054] The filtering unit of Fig. 1 can perform filtering on restored pixel values. The filtering unit can perform at least one of a deblocking filter, an offset correction unit, or an ALF (Adaptive Loop Filter).

[0055] A deblocking filter can remove block distortion caused by boundaries between blocks in a reconstructed picture. To determine whether to perform deblocking, a deblocking filter can be applied to the current block based on the pixels contained in several columns or rows within the block. When applying a deblocking filter to a block, a strong filter or a weak filter can be applied depending on the required deblocking filtering strength. Furthermore, when applying a deblocking filter, horizontal and vertical filtering can be processed in parallel when performing vertical and horizontal filtering.

[0056] The offset correction unit can correct the offset from the original image on a pixel-by-pixel basis for an image that has undergone deblocking. To perform offset correction for a specific picture, the pixels contained in the image can be divided into a certain number of regions, the regions to be offset can be determined, and the offset can be applied to those regions. Alternatively, the offset can be applied by considering the edge information of each pixel.

[0057] Adaptive Loop Filtering (ALF) can be performed based on the comparison of the filtered restored image with the original image. After dividing the pixels included in the image into predetermined groups, a filter to be applied to each group can be determined, and filtering can be performed differentially for each group. Information regarding whether to apply ALF can be transmitted by luminance signal for each coding unit (CU), and the shape and filter coefficients of the ALF filter to be applied can vary depending on each block. Furthermore, an ALF filter of the same form (fixed form) can be applied regardless of the characteristics of the target block.

[0058]

[0059] The picture level information setting unit of Fig. 1 can parse information applicable to the entire area of ​​the picture currently being decoded from a bitstream or generate it according to a predetermined procedure. The picture level information can include at least one of a picture type, a picture quantization parameter, a reference picture list, or a picture level linear prediction mode.

[0060] A picture-level linear prediction mode, which is one of the picture-level information, can be parsed from the picture header. The linear prediction mode may be a prediction mode that performs unidirectional or bidirectional linear motion prediction for all areas of the current picture or for some prediction blocks using motion information in the temporal reference motion information field of the current picture.

[0061] Linear prediction mode may mean performing picture-level linear prediction, performing block-level linear prediction, or not performing linear prediction mode. Performing picture-level linear prediction may be a mode in which linear prediction is performed on the entire region within a picture at the picture level, performing block-level linear prediction may be a mode in which linear prediction mode is selectively performed at the block level, and not performing linear prediction mode may mean not performing linear prediction mode on the current picture.

[0062]

[0063] Figure 2 illustrates an embodiment of a motion information unit and a motion information field.

[0064] Figure 3 illustrates an embodiment of a temporal reference motion information unit and a temporal reference motion information field.

[0065] The temporal reference motion information field generation unit of Fig. 1 generates a temporal reference motion information unit ( × Temporal reference motion information existing in units of pixel size can be listed. The temporal reference motion information can be composed of a motion vector, a motion vector scale indicating the difference value of POC (Picture Order Count) between the start frame and the end frame of the motion, and can be derived from the motion information field of the restored picture. The width of the current picture , height When the temporal reference motion information field is There may be temporal reference motion information, and the temporal reference motion information may be stored in the raster scan order based on the position of the temporal reference motion information unit in the picture. The motion information field of the restored picture stores the motion information used by the restored picture in the restoration process in the motion information unit ( It may be listed in units of (pixel size). and may be the same or different, and may be the same or different. , , , Each may be a multiple of 4, for example. The motion information may be composed of a motion vector, a reference picture index, etc. The motion information field of the restored picture stored in the restored picture buffer at the current point in time may be stored in the motion information buffer. The restored picture buffer and the motion information buffer may be different buffers or may refer to the same buffer.

[0066]

[0067] Figure 4 illustrates the execution process of the temporal reference motion information field generation unit.

[0068] The temporal reference motion information field generation unit of Fig. 1 can be performed by the process of Fig. 4. Each step of Fig. 4 can be performed at the picture level, the large unit level, or the coding unit level. For example, the meaning that any step is performed at the picture / large unit / coding unit level can mean that the step is performed at the time when decoding of the current picture / large unit / coding unit starts, and the information generated thereby can be applied to the entire current picture / large unit / coding unit. The large unit can be a set of one or more coding tree units, or a coding tree unit divided horizontally or vertically. The width / height of the large unit can be, for example, an integer multiple of the width / height of the coding tree unit, or a value obtained by dividing the width / height of the coding tree unit by a power of 2. The shape of the large unit can be, for example, a square or a rectangle. The restored picture-level linear prediction mode syntax in the picture-level information setting section of Fig. 1 may mean performing picture-level linear prediction, performing block-level linear prediction, or not performing linear prediction mode. If the picture-level linear prediction mode syntax means performing picture-level linear prediction, all processes of Fig. 4 may be performed at the picture level. If the picture-level linear prediction mode syntax means performing block-level linear prediction, some of the steps of Fig. 4 may be performed at the picture level, and some of the steps may be performed at the large unit or coding unit level. For example, the motion reference picture selection step of Fig. 4 may be performed at the picture level, and the steps after the motion reference picture selection step of Fig. 4 may be performed at the large unit level. As another example, the motion reference picture selection step of Fig. 4 may be performed at the picture level, and the steps after the motion reference picture selection step of Fig. 4 may be performed at the coding unit level.

[0069] The motion reference picture selection step of FIG. 4 may be a step of selecting a picture to refer to a motion information field in order to generate a temporal reference motion information field of the current picture among the restored pictures stored in the DPB. The motion reference picture selection step may be performed at a picture level, a large unit level, or a coding unit level. A set maximum number of pictures (max_num_ref_picture) may be selected among the restored pictures stored in the DPB, and the selection process may be as follows. Each process may be omitted.

[0070] Process 1) A maximum number (max_num_backward_picture) of past pictures preceding the current picture in display order can be selected. Pictures with a smaller POC difference from the current picture can be selected first, and the number of pictures actually selected (num_backward_picture) can be less than or equal to max_num_backward_picture.

[0071] Step 2) A maximum number (max_num_forward_picture) of future pictures that are after the current feature in display order can be selected. Pictures with a smaller difference in POC from the current picture can be selected first, and the number of pictures actually selected (num_forward_picture) can be less than or equal to max_num_forward_picture. If (num_backward_picture + num_forward_picture) <= max_num_ref_picture, steps 3 and 4 can be omitted. If (num_backward_picture + num_forward_picture) > max_num_ref_picture, max_num_ref_picture pictures can be selected through step 3. Alternatively, step 2 can be terminated when (num_backward_picture + num_forward_picture) becomes equal to max_num_ref_picture, and steps 3 and 4 can be omitted. By reversing the order of steps 2 and 1, you can select from the future pictures first and then from the past pictures.

[0072] Step 3) The nth closest picture to the current picture from each of the past pictures and future pictures can be grouped and finally selected by performing Step 4 for each group, and when the number of finally selected pictures becomes equal to max_num_ref_picture while performing Step 4, the motion reference picture selection step of Fig. 4 can be terminated. Step 4 can be performed for each group in ascending or descending order of the n value.

[0073] Step 4) In the nth group, both past pictures and future pictures may exist, or only one of the two pictures may exist. If only one of the two pictures exists, the picture can be included in the final selected picture group. If both pictures exist, the order of inclusion in the final selected picture group can be determined through the following process. By referring to the motion information field of each picture, motions whose start picture and end picture are in the opposite direction with respect to the current picture can be searched, and the value with the smallest POC difference from the current picture can be derived among the end pictures of the motions. Among the past pictures and future pictures, the picture with the smaller minimum POC difference value derived through the above process can be preferentially included in the final selected picture group.

[0074] Fig. 5 illustrates one embodiment of the execution of the motion reference picture selection step. An example of the execution of the motion reference picture selection step of Fig. 4 may be as shown in Fig. 5.

[0075] Figure 6 illustrates a step of changing the spatial resolution of a motion information field.

[0076] Figure 7 illustrates one embodiment of performing a motion information field spatial resolution change step.

[0077] The motion information field spatial resolution change step of FIG. 4 can change the spatial resolution of the motion information field of the picture selected in the motion reference picture selection step of FIG. 4, and the execution process and result can be as shown in FIG. 6 and FIG. 7, respectively. The motion information field spatial resolution change step of FIG. 4 can be performed at the picture level, the large unit level, or the coding unit level.

[0078] When changing the spatial resolution of a motion information field at the picture level, the spatial resolution of the motion information field of the entire area of ​​the restored picture can be changed. Depending on the embodiment, the spatial resolution of the entire motion information field can be the same as the size of the restored picture, or can be upsampled or / and downsampled based on the size of the restored picture. For example, this can be performed as in Fig. 7-(a).

[0079] Alternatively, when changing the spatial resolution of a motion information field at a large unit level or a coding unit level, a subset of the motion information field whose spatial resolution is to be changed may be determined based on the position of the large unit or coding unit being processed in the current picture. Depending on the embodiment, the spatial resolution of a part of the motion information field may be the same as the size of the current coding unit, or may be upsampled and / or downsampled based on the size of the current coding unit. For example, FIG. 7-(b) may illustrate an example of downsampling a motion information field of a reconstructed picture at a coding unit level.

[0080] The step of determining whether to change the spatial resolution of the motion information field in FIG. 6 can determine whether to change the spatial resolution of the motion information field for each motion information reference picture. If the spatial resolution of the motion reference picture and the spatial resolution of the current picture are the same, the spatial resolution change of the motion information field may not be performed and the subsequent steps in FIG. 6 may be omitted. If the spatial resolution of the motion reference picture and the spatial resolution of the current picture are different, the spatial resolution change of the motion information field may be performed.

[0081] The step of determining the degree of change in the spatial resolution of the motion information field in Fig. 6 can derive the degree of change in the spatial resolution of each of the width and height of the motion information field. A scaling value that makes the spatial resolution of the motion reference picture the same as the spatial resolution of the current picture can be obtained and used for scaling the spatial resolution of the motion information field. The width of the motion reference picture , the height of the motion reference picture , the width of the current picture , the height of the current picture When the width of the motion information field is / ship, height / It can be changed to a ship.

[0082] The step of deriving motion information with changed spatial resolution of FIG. 6 can scale the spatial resolution of the entire or a subset of the motion information field using the spatial resolution scaling value determined in the step of determining the degree of spatial resolution change of the motion information field of FIG. 6. The target area of ​​the motion information field where the spatial resolution change is to be performed may be an area determined when the step of changing the spatial resolution of the motion information field of FIG. 4 is started.

[0083]

[0084] Figure 8 illustrates a step of deriving motion information with changed spatial resolution.

[0085] The step of deriving motion information with changed spatial resolution of Fig. 6 can be performed through the process of Fig. 8.

[0086] The integer unit motion information scaling step of Fig. 8 can scale the motion information of the integer unit based on the spatial resolution scaling value. Among the elements that constitute the motion information, the motion vector can be scaled, and the motion vector can be composed of an x-axis motion and a y-axis motion. As a process of performing scaling on the motion of each axis, for the x-axis motion / Multiply by and apply the y-axis motion / can be multiplied by

[0087] Figure 9 illustrates an example of a temporal reference motion information sampling position.

[0088] The unit sampling position derivation step of Fig. 8 can determine the position of the unit to be sampled according to the degree of spatial resolution scaling of the motion information field. (not overlapping in the motion information field) / )*( / ) groups of temporal reference motion information units, one sampling position can be derived within the area of ​​the temporal reference motion information unit group, and each sampling position can have the same interval in the horizontal direction and the vertical direction. Fig. 9 can show an example of a unit sampling position, and the unit sampling position can be an integer unit position or a decimal unit position within the unit group area. The unit sampling position can be a specific position that is fixed to the same extent by the encoder / decoder, or can be derived based on a sampled position when downsampling the current picture at the picture level.

[0089]

[0090] The motion information derivation step of the sampled unit position of Fig. 8 can derive the motion information of the unit sampling position using the motion information of the unit at the sampled position or the surrounding position.

[0091] If the unit sampling position is an integer unit position, the motion information of the integer unit can be used as the result. Alternatively, the result can be derived using the motion information of surrounding integer units including the integer unit.

[0092] If the unit sampling position is not an integer unit position, the resulting motion information can be derived by combining the motion information of two or more integer units around the sampling position.

[0093] When the unit sampling position is an integer unit position or not an integer unit position, a method for deriving motion information of the unit sampling position by combining motion information of two or more surrounding integer units may be as follows. The two or more surrounding integer units may be specified in the decoder / encoder, and may be, for example, temporal reference motion information units within the current temporal reference motion information unit group, or may also include temporal reference motion information units within an adjacent temporal reference motion information unit group.

[0094] Figure 10 illustrates an example of motion information of multiple units within a unit group.

[0095] Among the surrounding integer units, integer units having the same reference picture index can be grouped, and one group can be selected. For example, in Fig. 10, u0 and u2 can be one group, and u1 and u2 can be one group. The group with the largest group size can be selected, or the group whose reference picture index is closest to the current picture index can be selected, or the group having a reference picture index that is in the opposite time direction from the motion reference picture with respect to the current picture can be selected. The weighted average of the motion vectors and the reference picture index in the selected group can be designated as the motion vector and the reference picture index of the sampling position unit, respectively. The weighted average can have the same weights or can have a value inversely proportional to the distance from the sampling position.

[0096] When using two or more of the above groups, a specific group can be designated as a reference group, and the motion vectors of other groups can be used by projecting them onto the reference picture of the reference group, and the weighted average of the motion vectors of the multiple groups and the reference picture index of the reference group can be designated as the motion vector and reference picture index of the sampling position unit, respectively.

[0097] The motion information projection step of the motion reference picture of FIG. 4 projects each motion information stored in the motion information field of the motion reference picture stored in the motion reference picture list in the motion reference picture selection step of FIG. 4 onto the current picture to create a temporal reference motion information unit () including the projected pixel. × It can be specified as temporal reference motion information (pixel size).

[0098] Figure 11 illustrates the process of performing the motion information projection step of the reference picture.

[0099] The motion information projection order derivation step of Fig. 11 can derive the projection order of motion information of motion reference pictures in the motion reference picture list. The motion information projection order can be defined identically for encoding / decoding or can be transmitted as an index. The motion information projection order can be, for example, as follows.

[0100] The motion information projection order 1 may be in ascending order of motion information index among one or more motion information of a unit, starting from the motion reference picture with the earliest sequence number within the motion reference picture list, in the raster scan order based on the upper left pixel of the unit within the motion information field, and in the order of motion information index.

[0101] Figure 12 shows an example of motion information projection order 2.

[0102] With respect to the motion information projection order 2, all motion information of motion reference pictures in the motion reference picture list can be grouped into two groups. The first group can be a group of motion information in which the picture to which the motion information belongs and the reference picture in the motion information exist in both directions based on the current picture. Motion information that does not belong to the first group can be included in the second group. The motion information projection order 2 can perform a projection process by starting from the first group and ordering the motion information included in the group using the method of the motion information projection order rule 1. FIG. 12 can represent a projection order determined by the motion information projection order rule 2.

[0103] The motion information projection order 3 can be determined in the following order among one or more motion information of a unit, starting from the motion reference picture with the earliest sequence number within the motion reference picture list, in the raster scan order based on the upper left pixel of the unit within the motion information field, and among one or more motion information of a unit. Among the one or more motion information, the order can be determined by using the case where the start and end pictures of the vector are on opposite sides of the current picture in the display order when projecting the motion as the first priority condition, and the closeness of the temporal distance between the end picture of the vector and the current picture as the second priority condition.

[0104] Figure 13 illustrates an example of a case where the prediction quality reliability of a motion vector is low.

[0105] The first decision step of whether to use motion information of Fig. 11 can primarily determine whether to use the corresponding motion information by judging the prediction quality reliability of the motion information. The prediction quality reliability can be, for example, the pixel value similarity of the start n×m region and the end n×m region of the motion vector, which is one of the components of the motion information. The similarity can be, for example, the average value of the pixel value difference between the two regions. If the similarity is below a threshold, the subsequent process of Fig. 11 for the corresponding motion information can be omitted, and if it is above the threshold, the subsequent process of Fig. 11 can be performed. Alternatively, when two or more motion information exist in the temporal reference motion information unit to which the motion information currently being processed belongs, and the start and end pictures of the motion information currently being processed exist in the same temporal direction with respect to the current picture, and the start and end pictures of the remaining motion information exist in the opposite temporal direction with respect to the current picture, it can be decided not to use the motion information currently being processed. Or, when there are two or more motion information in the temporal reference motion information unit to which the motion information currently being processed belongs, it may be decided to exclude both motion information regardless of the similarity. Or, when there are two or more motion information in the temporal reference motion information unit to which the motion information currently being processed belongs, at least one motion information may be used regardless of the similarity, and / or at least one motion information may be excluded regardless of the similarity to perform the first decision on whether to use the motion information.

[0106] Figure 14 illustrates one embodiment of the step of deriving a motion vector projection position.

[0107] The motion vector projection position derivation step of Fig. 11 can derive which reference motion unit of the current picture the motion information is projected to by projecting the motion vector to the current picture using the motion vector and POC difference of the motion information. After determining the POC order relationship between the current picture, the motion reference picture, and the reference picture of the motion reference picture, projection can be performed using a determined method depending on each order relationship, and Fig. 14 can show an example of projection performance depending on the case.

[0108] POC of the current picture , POC of motion reference picture , the POC of the reference picture of the motion reference picture When this is said, the above causal relationship case can be classified into one of the following three cases.

[0109] Case 1: < < or < <

[0110] Case 2: < < or < <

[0111] Case 3: < < or < <

[0112] The second decision step of whether to use motion information of Fig. 11 can secondarily decide whether to use the corresponding motion information based on the projection position derived in the motion vector projection position derivation step of Fig. 11. Examples of multiple conditions for deciding whether to use the motion information may be as follows. Zero or more of the following conditions may be equally judged by the encoder / decoder, and if even one of the specified conditions is not satisfied, it may be decided not to use the corresponding motion information.

[0113] As a first condition, the current motion information may not be used if the projection position is outside the boundary of the current picture.

[0114] As a second condition, if motion information for which projection has already been performed in the previous order is stored in the motion information unit at the moment when motion information is to be stored, the current motion information may not be used unconditionally. Alternatively, the following condition may be determined, and if the condition is not satisfied, the current motion information may not be used, and if the condition is satisfied, the existing motion information may be replaced with the current motion information. The following conditions may be any one of: i) if the previously stored motion information is included in group k (see Fig. 12) and the current motion information is included in group l (satisfying k>=l); ii) if the difference between the start and end picture POCs of the current motion information is smaller than the difference between the start and end picture POCs of the previously stored motion information; and iii) if the position where the current motion information is projected is closer to the center position of the reference motion unit than the position where the previously stored motion information is projected.

[0115] The motion information modification and storage step of Fig. 11 may partially modify and then store motion information in a reference motion unit including a position where the motion information is projected. The motion information of a motion reference picture may be composed of a motion vector and a reference picture index of the motion reference picture.

[0116] The above motion reference picture index can be converted into a POC difference between the motion reference picture and the reference picture of the motion reference picture and stored in the reference motion field of the current picture.

[0117] The scale of the above motion vector can be scaled to a POC distance of 1 and stored, and the POC difference value can be stored as 1 or the POC difference value can not be stored.

[0118] The direction of the above motion vector can be unified and stored in the past direction or the future direction. The direction of the motion vector can be two, and it is a motion vector in the future direction (Fig. 14). < ) Conversely, the motion vector in the past direction (Fig. 14) > ) can be. If the direction of the motion vector is unified to the future direction, the sign of the motion vector in the past direction can be stored as an inverse value, and if it is unified to the past direction, the sign of the motion vector in the future direction can be stored as an inverse value.

[0119]

[0120] Figure 15 illustrates the process of performing the temporal reference motion information modification step.

[0121] The temporal reference motion information modification step of FIG. 4 can perform a process of scaling the motion vector of the temporal reference motion information field and filling the empty units of the temporal reference motion information field with motion information by the process of FIG. 15 when the picture level linear prediction mode syntax means performing picture level linear prediction.

[0122] The temporal reference motion information scaling step of FIG. 15 can scale the reference motion vector in each temporal reference motion information in the temporal reference motion information field of the current picture and modify the reference motion vector scale value. As a method of scaling the reference motion vector, (the difference between the POCs of two reference pictures of the current picture) / (the POC difference between the motion reference picture from which the reference motion vector is obtained and the reference picture of the motion reference picture) can be multiplied. The reference motion vector scale value can be replaced with a previously stored value (the difference between the POCs of two reference pictures of the current picture).

[0123] The step of filling in the temporal reference motion information field of Fig. 15 can predict and store motion information in a unit where motion information does not exist in the temporal reference motion information field. This step may be performed only when the temporal reference motion information field generation unit of Fig. 1 is performed at the picture level, or it may always be performed regardless of the level at which it is performed. The motion information prediction method can use a method identically designated for the encoder / decoder, and an example of its performance can be as follows.

[0124] Figure 16 illustrates an example 1 of performing a step of filling in a temporal reference motion information field.

[0125] Example 1 of the execution is to iterate over the units in which temporal reference motion information exists in the temporal reference motion information field in the raster scan order, and to find the corresponding unit and this It may be possible to copy motion information to the empty unit below. p can be a real number greater than or equal to 1, can be a real number greater than or equal to 1. The same process can be repeated for units filled with motion information during the traversal process, and the process can be repeated until there are no more empty units. Figure 16 shows p = 1, =1 can be expressed.

[0126] Figure 17 illustrates an example 2 of performing a step of filling in a temporal reference motion information field.

[0127] Example 2 of the implementation may be to predict temporal reference motion information for each empty unit by traversing the empty units in the temporal reference motion information field in raster scan order. As a method of predicting temporal reference motion information of an empty unit, the empty unit and this The reference motion vectors of units that have temporal reference motion information can be weighted based on the distance to the current empty unit. Figure 17 shows p = 1, =2 can be expressed.

[0128] After execution example 1 is performed, execution example 2 may be performed, or after execution example 2 is performed, execution example 1 may be performed, or only execution example 1 may be performed, or only execution example 2 may be performed.

[0129]

[0130] Figure 18 illustrates the process of performing inter prediction on a block-by-block basis in the inter prediction unit.

[0131] The inter prediction unit of Fig. 1 can generate an inter prediction signal in units of prediction blocks. A prediction block can be defined as one of the following. If the linear prediction mode restored in the picture level information setting unit of Fig. 1 is a picture-wide execution mode, the prediction block can be identical to a temporal reference motion information unit. If the linear prediction mode is a block-based execution mode or does not use a linear prediction mode, the prediction block can be defined according to the coding block partitioning or prediction block partitioning of the picture. A process for generating a prediction signal in units of prediction blocks can be as shown in Fig. 18, which can represent a process for restoring motion information of a prediction block and generating a prediction signal using the motion information. Each step of Fig. 18 can be omitted and the order can be changed.

[0132] A prediction signal can be generated for each prediction block by the process of FIG. 18 while traversing all prediction blocks in a picture in a raster scan order, a depth-first search order, or an order dependent on the processing order of units (large coding units, coding tree units, tiles, slices, etc.) of a higher layer than the prediction block.

[0133] If the linear prediction mode of the picture level parsed in the picture level information setting unit of Fig. 1 is picture level linear prediction performance, the linear motion information derivation step of Fig. 18 and the prediction signal generation step of Fig. 18 can be performed to generate a prediction signal of the current prediction block. If the linear prediction mode of the picture level is block level linear prediction performance-capable, the syntax indicating whether to perform the linear prediction mode on the current prediction block is parsed in the block level linear prediction mode restoration step of Fig. 18, and if it means true, each step of Fig. 18 can be performed to generate a prediction signal, and if it means false, each step except the linear motion information derivation step of Fig. 18 can be performed to generate a prediction signal.

[0134] The block level linear prediction mode restoration step of Fig. 18 can restore whether the linear prediction mode of the current prediction block is performed by parsing it from the bitstream. Only when the linear prediction mode of the picture level parsed in the picture level information setting section of Fig. 1 is block level linear prediction performable, the current step is performed, and a syntax indicating either block level linear prediction performable or block level linear prediction not performable can be parsed.

[0135] The motion information count restoration step of FIG. 18 can restore the number of motion information to be used for generating a prediction signal of the current prediction block. If the current prediction block performs block-level linear prediction, the number of motion information to be restored can be derived as a specific number of 0 or more. The motion information corresponding to the above number can be restored through the reference picture index restoration step of FIG. 18, the motion coding mode restoration step of FIG. 18, the spatiotemporal reference motion vector derivation step of FIG. 18, and the residual motion vector restoration and summation step of FIG. 18. If the current prediction block does not perform block-level linear prediction, the number of motion information can be parsed from the bitstream.

[0136] The reference picture index restoration step of FIG. 18 can restore the reference picture index of each motion information of the current prediction block. If the current prediction block performs block-level linear prediction, the current step can be omitted. If the current prediction block is not in linear prediction mode, the reference picture index can be parsed from the bitstream as many times as the number of motion information.

[0137] The motion coding mode restoration step of FIG. 18 can restore the motion reference mode of the current prediction block, and the motion reference mode can restore the motion reference mode of each motion information individually or in a fused form. The motion reference mode may include a mode that uses motion information of a previously restored prediction block to restore motion information (motion merge mode), a mode that uses motion information of a previously restored prediction block and differential motion information together (motion reference mode), etc. If the current prediction block is not a linear prediction mode, the motion reference mode can be parsed from the bitstream.

[0138] Figure 21 illustrates the second temporal reference motion vector candidate location.

[0139] The spatiotemporal reference motion vector derivation step of FIG. 18 can construct a spatiotemporal reference motion vector list with motion information of previously reconstructed prediction blocks and select spatiotemporal reference motion vectors from the list. The spatiotemporal reference motion vector list can be filled with motion information (spatial reference motion vector) candidates of prediction blocks that are spatially adjacent or close to the current prediction block within the same picture, motion information (first temporal reference motion vector) candidates of an area in a spatial position similar to the current prediction block within the reconstructed picture, motion information (second temporal reference motion vector) candidates of a temporal reference motion information unit in a spatial position similar to the current prediction block within the temporal reference motion information field, etc. The second temporal reference motion vector can be selected from one or more temporal reference motion information units within an area corresponding to the current prediction block within the temporal reference motion information field. For example, the motion information of the temporal reference motion information unit at the upper left (Fig. 21), center (Fig. 21), or lower right (Fig. 21) positions can be selected.

[0140] The above temporal reference motion information field may already be filled with all temporal reference motion information within the field through the process of FIG. 4 at the frame level, or temporal reference motion information at a position corresponding to the current prediction block at the current time point may be derived through the process of FIG. 4. The spatial reference motion vector, the first temporal reference motion vector, and the second temporal reference motion vector may each be added to the spatiotemporal reference motion vector list in a predetermined order by a predetermined maximum number.

[0141] The residual motion vector restoration and summing step of Fig. 18 can restore the residual motion vector according to the motion reference mode of the current prediction block, and when the residual motion vector is restored, it can be summed with the spatiotemporal reference motion vector to derive the restored motion vector. In the motion coding mode restoration step of Fig. 18, the residual motion vector can be restored when the restored motion coding mode is the motion reference mode. When the current prediction block is a unidirectional prediction that uses one piece of motion information, one residual motion vector can be restored when the motion coding mode is the motion reference mode. When the current prediction block is a bidirectional prediction that uses two pieces of motion information, the motion coding mode can mean that each piece of motion information is a motion merge mode or a motion reference mode. A residual motion vector can be restored for motion information that is a motion reference mode.

[0142] Figure 22 illustrates the process of performing the linear motion information derivation step.

[0143] The linear motion information derivation step of Fig. 18 can derive unidirectional or bidirectional linear motion information when the current picture performs picture-level linear prediction or the current prediction block performs block-level linear prediction. The linear motion information can be composed of a motion vector and a reference picture index. The reference picture index can be derived according to a predetermined rule, and the motion vector can be derived from the temporal reference motion information field of the current picture. This step can be performed by the process of Fig. 22.

[0144] The temporal reference motion information unit selection step of Fig. 22 can select a temporal reference motion information unit to be referenced in order to derive linear motion information of the current prediction block from the temporal reference motion information field. The temporal reference motion information unit can be selected in units of prediction blocks or sub-prediction blocks.

[0145] Figure 23 illustrates an example of a result of performing a temporal reference motion information unit selection step.

[0146] A method for selecting a temporal reference motion information unit on a prediction block basis may be as follows. One or more temporal reference motion information units may be included within an area of ​​one prediction block. For example, if the current prediction block performs picture-level linear prediction, one temporal reference motion information unit may be included within an area of ​​one prediction block, and if the current prediction block performs block-level linear prediction, one or more temporal reference motion information units may be included within an area of ​​one prediction block. A method for selecting a temporal reference motion information unit on a prediction block basis may follow the process of deriving a second temporal reference motion vector in the spatiotemporal reference motion vector derivation step of FIG. 18. If the current prediction block performs block-level linear prediction, a temporal reference motion information unit including a position obtained by adding one motion vector reconstructed in the residual motion vector restoration and summation step of FIG. 18 to the upper left pixel of the current prediction block may be designated as the upper left unit, and then the above-mentioned derivation process may be performed. In the process of selecting a unit according to the above process, if there is no temporal reference motion information unit with motion information, the motion vector of the temporal reference motion information can be regarded as a zero motion vector in a subsequent process. Fig. 23-a can represent an example of the result of selecting a temporal reference motion information unit in a prediction block unit.

[0147] A method for selecting a temporal reference motion information unit in units of sub-prediction blocks may be as follows. A temporal reference motion information unit included in a region of a current prediction block may match the sub-prediction block. A temporal reference motion information unit corresponding to the region of each sub-prediction block may be designated as a temporal reference motion information unit in units of sub-prediction blocks. If motion information does not exist in a temporal reference motion information unit selected in any sub-prediction block, the motion vector may be regarded as a zero motion vector in a subsequent step. Fig. 23-b may show an example of a result of selecting a temporal reference motion information unit in units of sub-prediction blocks.

[0148] The linear motion information count derivation step of FIG. 22 can derive whether the linear motion information of the current prediction block is bidirectional motion information or unidirectional motion information, and if unidirectional, in which direction. As a method, it can always be derived as bidirectional prediction among unidirectional and bidirectional. Alternatively, it can be derived through a judgment process that is identically defined for the encoder / decoder as to whether bidirectional prediction or unidirectional prediction will be performed on the prediction block, and if unidirectional prediction, in which direction will it be performed between forward and backward. The judgment process can be performed, for example, as in the following embodiments.

[0149] In a first embodiment, a step of determining one of bidirectional and unidirectional can be determined through the following process. If the temporal reference motion information of the current prediction block or sub-prediction block is included in group 1 (see FIG. 12), the current prediction block or sub-prediction block can be determined as bidirectional prediction, and if the temporal reference motion information of the current prediction block or sub-prediction block is included in group 2 (see FIG. 12), the current prediction block or sub-prediction block can be determined as unidirectional prediction. The temporal reference motion information may be temporal reference motion information of a unit position derived by a process of deriving a second temporal reference motion vector in the spatiotemporal reference motion vector derivation step of FIG. 18. If the current prediction block or sub-prediction block is determined as unidirectional, a process of determining one of the forward and backward directions can be performed. If the temporal reference motion information of the current prediction block or sub-prediction block is referenced from a past picture based on the current picture, the current prediction block or sub-prediction block can be determined as backward (past direction) unidirectional prediction, and if the temporal reference motion information of the current prediction block or sub-prediction block is referenced from a future picture based on the current picture, the current prediction block or sub-prediction block can be determined as forward (future direction) unidirectional prediction.

[0150] Figure 24 shows an example of a linear motion valid section.

[0151] In a second embodiment, a valid section of linear motion according to a temporal reference motion vector of a current prediction block or sub-prediction block is derived, and if the linear motion valid section spans from a past picture to a future picture, it can be determined as a bidirectional prediction, and if it spans only a past or future section, it can be determined as a backward or forward unidirectional prediction. The linear motion valid section can be derived through the following process. The temporal reference motion information of the current prediction block or sub-prediction block is projected onto each reference picture in the reference picture list, and the projected pixel is set as the upper left pixel, and an area having the same size as the current prediction block or sub-prediction block can be called a corresponding restoration block. Corresponding restoration blocks with high similarity can be grouped together, and a linear motion validity value proportional to the number of grouped corresponding restoration blocks can be assigned to each reference picture. An area including reference pictures having a linear motion validity value greater than a certain value can be designated as a linear motion valid section. Alternatively, the section in which the reference pictures with a linear motion validity value greater than a certain value continue the most can be designated as the linear motion validity section. Any two pictures can be said to continue if the interval between them is less than n (n>=1) pictures.

[0152] The reference picture index derivation step of FIG. 22 can restore the reference picture index, which is one of the components of linear motion information. The number of reference picture indices corresponding to the number of linear motion information derived in the linear motion information count derivation step of FIG. 22 can be restored. For example, when the number of linear motion information is bidirectional motion information, two or more reference picture indices can be restored, and when the number of linear motion information is unidirectional motion information, one reference picture index can be restored. The method for restoring the reference picture index can combine information parsed from the bitstream or information derived by a process defined in the encoder / decoder, and can be, for example, one of the following embodiments.

[0153] In the first embodiment, the number of reference picture indices to be restored can be parsed from the bitstream. The meaning of the information parsed from the bitstream can mean a difference value between a current picture index and a reference picture index, or can mean an index of a reference picture in a reference picture list, or can mean an index of a reference picture in a sequence or a picture group. Using the parsed information, an absolute index value of a reference picture index at a sequence level, or a difference value between a current picture index and a reference picture index can be derived.

[0154] In the second embodiment, the reference picture with the closest temporal distance to the current picture can be selected for each direction. Alternatively, the reference picture with the closest temporal distance to the current picture can be selected among the reference pictures with the highest linear motion validity for each direction.

[0155] In the third embodiment, when the motion vector restored in the residual motion vector restoration and summation step of FIG. 18 is (0, 0), the reference picture used in the motion skip mode can be excluded from the options in each direction and the second embodiment can be performed. The motion vector can refer to the motion vector used in the temporal reference motion information unit selection step of FIG. 22 when the current prediction block is performed by block-by-block linear prediction. The motion skip mode can be one of the motion information coding modes at the coding block or prediction block level. The reference picture used in the motion skip mode can be a past picture or a future picture having the closest temporal distance to the current picture.

[0156] In the fourth embodiment, if there is a prediction block among the reconstructed prediction blocks spatially adjacent to the current prediction block that has a motion vector similar to the temporal reference motion vector of the current prediction block or the sub-prediction block, the reference picture used in the motion skip mode can be excluded from the selections in each direction and the second embodiment can be performed.

[0157]

[0158] The motion vector derivation step of Fig. 22 can derive a motion vector, which is one of the components of linear motion information. In units of prediction blocks or sub-prediction blocks, the motion information of the temporal reference motion information unit determined in the temporal reference motion information unit selection step of Fig. 22 can be scaled using the reference picture index derived in the reference picture index derivation step of Fig. 22 to derive a motion vector. The number or direction of the motion vector can be determined in the linear motion information number derivation step of Fig. 22. The temporal reference motion information includes a reference motion vector ( , ) and motion vector scale (MV_scale) may be included, and the motion vector ((MV_L0) in each direction of the current prediction block may be calculated using the temporal reference motion information and the reference picture index. x , MV_L0 y ), (MV_L1 x , MV_L1 y )) The process of deriving can be the same as the formula. If the reference motion vector is stored in the temporal reference motion information field based on the forward direction, sign_L0 = 1, sign_L1 = -1, and if it is stored based on the reverse direction, sign_L0 = -1, sign_L1 = 1.

[0159]

[0160] Figure 19 illustrates an embodiment of a method for deriving a motion vector when the linear prediction mode is performed on the entire picture area.

[0161] Figure 20 illustrates an example of deriving a prediction signal using motion information.

[0162] The prediction signal generation step of FIG. 18 can generate a prediction signal of the current prediction block using the motion information restored in the motion information restoration step of FIG. 18, and can be performed in the same process and procedure in all cases where the linear prediction mode of the current picture is the full picture area mode, block mode, or no-performance mode. If the current prediction block is a unidirectional prediction, a prediction signal generated using one motion information can be designated as the final prediction signal, and if it is a bidirectional prediction, a final prediction signal can be derived by weighting and adding prediction signals generated using two pieces of motion information respectively.

[0163]

[0164] The video encoder of the present disclosure may include at least one of a picture level information setting unit, a temporal reference motion information field generation unit, an entropy encoding unit, a quantization unit, a transformation unit, an intra prediction unit, an inter prediction unit, or a filtering unit.

[0165] The description for each is described in the video decoder of the present disclosure, and the specific details are omitted.

[0166] In addition, the entropy encoding unit can perform the entropy decoding process of the entropy decoding unit in reverse, the quantization unit can perform the inverse quantization process of the inverse quantization unit in reverse, and the transformation unit can perform the process of the inverse transformation unit in reverse.

[0167]

[0168] While the exemplary methods of this disclosure are presented as a series of operations for clarity of description, this is not intended to limit the order in which the steps are performed, and individual steps may be performed simultaneously or in different orders, if desired. To implement a method according to this disclosure, additional steps may be included in addition to the steps illustrated, some steps may be excluded and the remaining steps included, or some steps may be excluded and additional steps included.

[0169] The various embodiments of the present disclosure are not intended to list all possible combinations but rather to illustrate representative aspects of the present disclosure, and the matters described in the various embodiments may be applied independently or in combinations of two or more.

[0170] Additionally, various embodiments of the present disclosure may be implemented by hardware, firmware, software, or a combination thereof. In the case of hardware implementation, the embodiments may be implemented by one or more ASICs (Application Specific Integrated Circuits), DSPs (Digital Signal Processors), DSPDs (Digital Signal Processing Devices), PLDs (Programmable Logic Devices), FPGAs (Field Programmable Gate Arrays), general processors, controllers, microcontrollers, microprocessors, etc.

[0171] The scope of the present disclosure may include software or machine-executable instructions (e.g., operating systems, applications, firmware, programs, etc.) that cause operations according to the methods of various embodiments to be executed on a device or a computer, and a non-transitory computer-readable medium having such software or instructions stored thereon and executable on the device or computer.

[0172] The present disclosure may be applicable to an industrial field that uses a linear prediction method among inter-screen prediction methods in a picture decoding process.

Claims

1. A step of obtaining picture level information of the current picture; A step of listing the temporal reference motion information of the current picture in units of temporal reference motion information units of the current picture to generate a temporal reference motion information field of the current picture; and A video decoding method, characterized in that it includes an inter-screen prediction performing step of performing inter-screen prediction in block units of the current picture based on a temporal reference motion information field of the current picture.

2. In paragraph 1, A video decoding method, characterized in that the picture level information includes a picture type, a picture quantization parameter, a reference picture list, and a picture level linear prediction mode.

3. In paragraph 1, A video decoding method, characterized in that the listing of the above temporal reference motion information is performed in raster scan order.

4. In paragraph 1, The step of generating the above temporal reference motion information field is: A motion reference picture selection step for selecting a motion reference picture to refer to a motion information field for generating the above temporal reference motion information field; A motion information field spatial resolution changing step for changing the spatial resolution of the motion information field of the above motion reference picture; A motion information projection step of a motion reference picture, wherein the motion information of the motion reference picture is designated as temporal reference motion information of a temporal reference motion information unit of the current picture including a projected pixel obtained by projecting the motion information onto the current picture, thereby generating a temporal reference motion information field of the current picture before modification; A video decoding method, characterized in that it includes a temporal reference motion information modification step of scaling a motion vector of the temporal reference motion information field before modification and filling an empty unit of the first temporal reference motion information field to obtain a temporal reference motion information field of the current picture.

5. In paragraph 4, The above motion reference picture selection step is performed at the picture level, A video decoding method, characterized in that the subsequent steps of the above motion reference picture selection step are performed at a lower level rather than a picture level.

6. In paragraph 5, A video decoding method, characterized in that the above motion reference picture selection step is performed by considering the POC (Picture Order Count) value of the current picture and the motion reference picture.

7. In paragraph 6, A video decoding method, characterized in that the motion information field spatial resolution change step is performed by comparing the spatial resolution of the motion reference picture and the spatial resolution of the current picture.

8. In paragraph 7, A video decoding method, characterized in that, in the motion information projection step of the motion reference picture, in response to the presence of two or more motion information in the temporal reference motion information unit, the motion information is excluded from the generation of the temporal reference motion information field before modification, regardless of the similarity of the motion information.

9. Step of obtaining picture level information of the current picture; A step of listing the temporal reference motion information of the current picture in units of temporal reference motion information units of the current picture to generate a temporal reference motion information field of the current picture; and A video encoding method, characterized in that it includes an inter-screen prediction performing step of performing inter-screen prediction in block units of the current picture based on a temporal reference motion information field of the current picture.

10. In a non-transitory computer-readable recording medium storing a bitstream generated by a video encoding method, The above image encoding method comprises: a step of obtaining picture level information of a current picture; A step of listing the temporal reference motion information of the current picture in units of temporal reference motion information units of the current picture to generate a temporal reference motion information field of the current picture; and A non-transitory computer-readable recording medium, characterized in that it includes an inter-screen prediction performing step of performing inter-screen prediction in block units of the current picture based on a temporal reference motion information field of the current picture.

Images