KR20220099899A - Method of piecewise linear scaling of geometry atlas and apparatus using the same - Google Patents

Abstract

A piecewise linear scaling method of a geometry atlas according to an embodiment of the present invention includes the steps of: generating min-max normalized depth values; and generating geometry atlases by scaling the depth values to correspond to gradients of a plurality of linear sections.

Description

Partial linear scaling method of geometric atlas and apparatus using the same

The present invention relates to encoding and decoding of immersive video, and more particularly, to encoding and decoding of geometry atlases corresponding to depth information representing 3D information in immersive video.

With the recent explosive increase in interest in immersive content and the development of broadcasting equipment and image transmission technology, there is an increasing movement to actively utilize immersive content in various fields such as multimedia industries such as movies and TV.

The immersive video provides viewers with images from multiple viewpoints to experience natural motion parallax.

In order to provide the immersive video, a photographing apparatus must capture images of a plurality of viewpoints and provide the captured images of the plurality of viewpoints. As the number of images of the captured viewpoints increases, there is an advantage in that high-quality 3D content can be generated. In addition, there is a disadvantage that a large amount of storage space is required when having a multi-view high-definition image.

Geometry information is depth information representing three-dimensional information in MPEG immersive video, and is generally expressed as a two-dimensional (2D) image of a single channel.

Such a geometry is expressed as a geometry atlas in which redundancy between views is removed from the multi-view geometry.

A geometric image expressed as atlases is compressed through a two-dimensional image codec.

The MPEG immersive video encoder expresses and encodes multi-view depth information as atlases, and the MPEG immersive video decoder decodes the geometric atlas and uses them to generate a virtual viewpoint image through view synthesis or the like.

That is, the geometry can be viewed as a two-dimensional image of depth or disparity information, and is simple image information compared to a texture image. This geometry is used as 3D information when generating a virtual viewpoint image through rendering. In particular, the geometry has high importance at the boundary of the object, and the geometry corresponding to the boundary of the object greatly affects the rendering quality.

As a result, there is an urgent need for a new technique for efficiently encoding/decoding a geometric atlas while minimizing an increase in encoded data in MPEG immersive video.

An object of the present invention is to maximize the quality of immersive video while minimizing loss in compression performance by expressing geometry based on importance.

Another object of the present invention is to maximize encoding/decoding efficiency of immersive video by expressing geometry information with different data representation accuracies (or different scales) according to importance in a plurality of regions.

In addition, an object of the present invention is to minimize syntax information that must be transmitted/received for this purpose while expressing a geometry atlas with different data representation accuracies according to importance in a plurality of areas.

A piecewise linear scaling method of a geometry atlas according to the present invention for achieving the above object includes: generating min-max normalized depth values; and generating geometry atlases by scaling the depth values according to slopes of a plurality of linear sections.

In this case, the partial linear scaling method of the geometry atlas may be performed by the immersive video encoder.

In this case, the linear intervals may be generated by dividing the entire range corresponding to the min-maximum normalized depth values into equal intervals of the same length.

In this case, the partial linear scaling method may further include generating an encoded immersive video signal using the geometry atlas.

In this case, the slopes may be determined based on sample occurrence frequencies of the linear sections.

In this case, each of the slopes may be set to be larger as the frequency of sample generation increases.

In this case, the sample occurrence frequencies may be clipped to a preset range.

In this case, the linear intervals include a first field (dq_interval_num) indicating the number of the linear intervals; and a second field (dq_norm_disp_pivot) indicating a scaled value corresponding to each of the linear sections.

In this case, the syntax fields may further include a third field (dq_scaled_disp_start) indicating a start depth value to which the partial linear scaling is applied.

In addition, an inverse scaling method of a geometry atlas according to an embodiment of the present invention includes the steps of reconstructing the geometry atlas; and generating min-max normalized depth values by inversely scaling the geometry atlas according to slopes of a plurality of linear sections.

In this case, the inverse scaling method of the geometry atlas may be performed by the immersive video decoder.

In this case, the linear intervals may be generated by dividing an entire range corresponding to the min-maximum normalized depth values into equal intervals of the same length.

In this case, the inverse scaling method of the geometry atlas may further include generating a virtual viewpoint image using the min-max normalized depth values.

In this case, the slopes may be determined based on sample occurrence frequencies of the linear sections.

In this case, each of the slopes may be set to allocate more bits as the sample generation frequency increases.

In this case, the linear intervals include a first field (dq_interval_num) indicating the number of the linear intervals; and a second field (dq_norm_disp_pivot) indicating a scaled value corresponding to each of the linear sections.

In this case, the syntax fields may further include a third field (dq_scaled_disp_start) indicating a start depth value to which the partial linear scaling is applied.

In addition, a partial linear scaling apparatus of a geometry atlas according to an embodiment of the present invention includes: a memory in which at least one program is recorded; and a processor executing the program. At this time, the program generates min-max normalized depth values; and instructions for performing the step of generating geometry atlases by scaling the depth values according to slopes of a plurality of linear sections.

In this case, the linear intervals may be generated by dividing the entire range corresponding to the min-maximum normalized depth values into equal intervals of the same length.

In this case, the slopes may be determined based on sample occurrence frequencies of the linear sections.

In this case, each of the slopes may be set to be larger as the frequency of sample generation increases.

In this case, the linear intervals include a first field (dq_interval_num) indicating the number of the linear intervals; and a second field (dq_norm_disp_pivot) indicating a scaled value corresponding to each of the linear sections.

According to the present invention, the image quality of the immersive video can be maximized while minimizing the loss of compression performance by expressing the geometry based on the importance.

In addition, the present invention can maximize the encoding/decoding efficiency of the immersive video by expressing the geometry information with different data representation accuracies (or different scales) according to importance in a plurality of regions.

In addition, the present invention can minimize the syntax information to be transmitted/received for this purpose while expressing the geometry atlas with different data representation accuracies according to importance in a plurality of areas.

1 is a graph illustrating partial linear scaling of a geometry atlas according to an embodiment of the present invention.
2 is a block diagram illustrating an immersive video system according to an embodiment of the present invention.
FIG. 3 is a block diagram illustrating an example of the immersive video encoder shown in FIG. 2 .
4 is a block diagram illustrating an example of the immersive video decoder shown in FIG. 2 .
5 is an operation flowchart illustrating a method of partial linear scaling of a geometry atlas according to an embodiment of the present invention.
6 is an operation flowchart illustrating a method for inverse scaling of a geometry atlas according to an embodiment of the present invention.
7 is a block diagram showing the configuration of a computer system according to an embodiment of the present invention.

Advantages and features of the present invention and methods of achieving them will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but will be implemented in a variety of different forms, only these embodiments allow the disclosure of the present invention to be complete, and common knowledge in the technical field to which the present invention belongs It is provided to fully inform the possessor of the scope of the invention, and the present invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout.

Although "first" or "second" is used to describe various elements, these elements are not limited by the above terms. Such terms may only be used to distinguish one component from another. Accordingly, the first component mentioned below may be the second component within the spirit of the present invention.

The terminology used herein is for the purpose of describing the embodiment and is not intended to limit the present invention. In this specification, the singular also includes the plural, unless specifically stated otherwise in the phrase. As used herein, “comprises” or “comprising” implies that the stated component or step does not exclude the presence or addition of one or more other components or steps.

Unless otherwise defined, all terms used herein may be interpreted with meanings commonly understood by those of ordinary skill in the art to which the present invention pertains. In addition, terms defined in a commonly used dictionary are not to be interpreted ideally or excessively unless clearly specifically defined.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings, and when described with reference to the drawings, the same or corresponding components are given the same reference numerals, and the overlapping description thereof will be omitted. .

Hereinafter, depth may be a concept including disparity. That is, all expressions expressed as depth or depth values below may be replaced with sides or disparities.

1 is a graph illustrating partial linear scaling of a geometry atlas according to an embodiment of the present invention.

Referring to FIG. 1 , it can be seen that the min-max normalized depth values are scaled using a plurality of piecewise linear intervals.

According to an embodiment, the geometry atlas may generally have a 16-bit depth, and may be generated in a downscaled format for 10-bit video coding.

In this case, min-max normalized depth values may be scaled in a range from {0,64} to {511, 1023}. That is, the min-max normalized depth values may be scaled from 0 to 511 or from 0 to 1023 or from 64 to 511 or from 64 to 1023.

In the generation of the geometry atlas, loss of depth information is incurred by this downscaling in the generation of geometry atlas. Furthermore, depth information loss may subsequently occur due to lossy coding of the depth atlas. The method for partial linear scaling of a geometry atlas according to an embodiment of the present invention may redistribute bits in a geometry codeword using partial linear scaling to reduce loss of depth information.

The detailed process by which the depth atlas is generated with min-max normalized values is described in "Test Model 7 for Immersive Video," ISO/ It is detailed in IEC JTC1/SC29/WG4, N0005.

The entire range of min-max normalized depth value is divided into a predetermined number of equal intervals. In this case, each interval is selectively scaled according to the frequency of occurrence of depth samples of the atlas (is adaptively scaled according to the frequency of the depth samples in the atlas). Specifically, a scaled depth range corresponding to the specific section may be adaptively determined according to the frequency of occurrence of depth samples included in the specific section.

1 , an original depth value of each interval is mapped to the corresponding scaled depth range. Geometry atlases are then generated as 10-bit depth representation.

Partial linear scaling can be defined as follows.

d' = (b2 _i - b1 _i ) / (a2 _i - a1 _i ) * (d - a1 _i ) + b1 _i

In this case, d is the depth value to be scaled, i is the interval index of the d value, and a1 _i and a2 _i are on the original depth range, respectively. ) are the minimum and maximum depths of the i-th section. In this case, b1 _i and b2 _i are the minimum and maximum depths of the i-th section in the scaled depth range, respectively.

Inverse scaling may be required to perform rendering on the decoder side, and this may be performed in a reverse way of (forward) scaling.

In order to apply the proposed partial linear scaling, the number of linear intervals and depth values of all scaled intervals may be signaled. As will be described later, such information may be signaled by setting a syntax element dq_quantization_law to 1.

In the example shown in FIG. 1 , a case in which 7 linear intervals are exemplified is exemplified, but the technical spirit of the present invention is not limited thereto. For example, there may be 16 linear intervals.

A partial linear scaling model may be derived at the encoder side. In this case, scaling of each section may be applied according to importance in terms of the rendered view quality of depth values in the section. For example, the partial linear scaling method according to an embodiment of the present invention assigns more codewords to depth intervals that have more frequent occurrence of depth. values). Furthermore, since depth information of regions around an object tends to be important from the viewpoint of rendering view quality, the partial linear scaling method according to an embodiment of the present invention further allocates codewords to depth sections including samples around the object. can do. In this case, the partial linear scaling method according to an embodiment of the present invention may be applied to both a basic view and additional views.

For example, the derivation of a partial linear model can be performed as follows.

Step 1) The original depth region is divided into N equal sections (intervals). (N is a natural number such as 7 or 16)

Step 2) Calculate the percentage of sample occurrence in each depth interval, and set the frequency of occurrence to [1/32, 1/8], and clipping is performed in a preset range.

Step 3) Adjust the interval range based on the statistics resulting from Step 2.

Table 1 below is a table showing an example of fields signaled in relation to partial linear scaling of a geometry atlas according to an embodiment of the present invention.

depth_quantization(　viewID　) { Descriptor dq_quantization_law [ viewID ] u(8) if(　dq_quantization_law[　viewID　]　==　0　) { dq_norm_disp_low [ viewID ] fl(32) dq_norm_disp_high [ viewID ] fl(32) } else if( dq_quantization_law[ viewID ] == 1 ) { dq_norm_disp_low [ viewID ] fl(32) dq_norm_disp_high [ viewID ] fl(32) dq_num_piece_minus1 [viewID] u(8) dq_scaled_disp_start [viewID] for( i = 0; i <= dq_num_piece [ viewID]; i++ ) dq_scaled_disp_range [ viewID ][ i ] fl(32) } if(　vme_embedded_occupancy_enabled_flag　) dq_depth_occ_threshold_default [ viewID ] ue(v) }

Referring to Table 1, when the value of the syntax element of dq_quantization_law is set to 1 in the syntax for signaling depth quantization, it can be seen that information for partial linear scaling is signaled. In this case, dq_quantization_law[ viewID ] indicates the type of a depth quantization method of a view having the same view ID as viewID. In this case, if dq_quantization_law is 0, uniform quantization may be applied to depth values. In this case, if dq_quantization_law is 1, partial linear scaling of the geometry atlas according to an embodiment of the present invention may be applied.

In this case, dq_norm_disp_low[viewID] and dq_norm_disp_high[viewID] represent the minimum normalized depth value (min in FIG. 1) and the maximum normalized depth value (max in FIG. 1) of a view having the same view ID as viewID, respectively. .

In this case, dq_num_piece_minus1[viewID] represents a value obtained by subtracting 1 from the number of linear intervals for partial linear scaling of a view having the same view ID as viewID. Alternatively, instead of dq_num_piece_minus1, a syntax derived by differentiating 2 or more natural numbers from the number of linear sections may be encoded/decoded. In this case, information for specifying the number of linear sections, such as the dq_num_piece_minun1 field, may correspond to the first field of the claim. According to an embodiment, the first field may be dq_num_piece, and in this case, dq_num_piece may indicate the number of linear sections as it is. According to an embodiment, the first field may be signaled based on information by adding or subtracting various natural numbers from the number of linear sections for partial linear scaling in some cases.

In this case, dq_scaled_disp_start[viewID] indicates a starting depth value for partial linear scaling of a view having the same view ID as viewID. According to an embodiment, dq_scaled_disp_start [ viewID ] may not be signaled. When signaling of the dq_scaled_disp_start value is omitted, the starting depth value for partial linear scaling may be inferred to be the same as dq_norm_disp_low. If dq_scaled_disp_start[ viewID ] is signaled separately from dq_norm_disp_low[ viewID ], a section to which partial linear scaling is applied can be more freely set. In this case, information indicating a starting depth value for partial linear scaling, such as the dq_scaled_disp_start field, may correspond to the third field of the claim.

In this case, dq_scaled_disp_range[viewID][i] may indicate the range of the i-th linear section for partial linear scaling of a view having the same view ID as viewID. As an example, dq_scaled_disp_range[viewID][i] may represent a maximum value, a minimum value, or a difference between a maximum value and a minimum value of a scaled depth value in the i-th section. At this time, dq_scaled_disp_range[viewID][i] is defined in a for loop that is repeated as many as the number of linear sections, and signals the upper limit of the scaled depth value for each linear section. have. As another example, in the first section (eg, i is 0) or the last section (eg, i is dq_num_piece_minus1+1), signaling of the above information dq_scaled_disp_range may be omitted. In this case, information indicating the range of the i-th linear section, such as the dq_scaled_disp_range field, may correspond to the second field of the claim.

In this case, in Table 1, dq_num_piece, which is a variable used to set the condition of the for loop, may correspond to dq_num_piece_minus1+1.

Table 2 below is a table showing another example of fields signaled in relation to partial linear scaling of a geometry atlas according to an embodiment of the present invention. Contrary to the example of Table 1, encoding/decoding of information indicating a starting depth value for partial linear scaling is excluded. In this case, the minimum value of the normalized depth value (ie, dq_norm_disp_low) may be set as the minimum depth value at which partial linear scaling is performed.

else if　( dq_quantization_law[ viewID ] == 1 ) { dq_norm_disp_low [viewID] dq_norm_disp_high [viewID] dq_interval_num [ viewID ] for(i=0;i<= dq_interval_num [ viewID ];i++) dq_norm_disp_pivot [viewID][ i ] }

In this case, dq_quantization_law[viewID] indicates the type of a depth quantization method of a view having the same view ID as viewID. In this case, if dq_quantization_law is 0, uniform quantization may be applied to depth values. In this case, if dq_quantization_law is 1, partial linear scaling of the geometry atlas according to the embodiment of the present invention may be applied. The normalized depth value (min in Fig. 1) and the maximum normalized depth value (max in Fig. 1) are shown.

In this case, dq_interval_num[viewID] indicates the number of linear intervals for partial linear scaling of a view having the same view ID as viewID. In this case, the dq_interval_num field may correspond to the first field of a claim.

In this case, dq_norm_disp_pivot[viewID][i] represents the normalized depth or disparity value of the i-th pivot or linear section in the partially linearly scaled domain of the view having the same view ID as viewID. In this case, dq_norm_disp_pivot[viewID ][i] may indicate a scaled depth value (upper limit) of the i-th linear section for partial linear scaling having the same view ID as viewID. At this time, dq_norm_disp_pivot [viewID ][i] is defined in a for loop that is repeated as many as the number of linear sections, and signals the upper limit of the scaled depth value for each linear section. have. In this case, the dq_norm_disp_pivot field may correspond to the second field of the claim. In this case, subtracting 1 from the total number of dq_norm_disp_pivot[viewID][i] may be the same as dq_interval_num[viewID].

2 is a block diagram illustrating an immersive video system according to an embodiment of the present invention.

Referring to FIG. 2 , an immersive video system according to an embodiment of the present invention includes an immersive video encoder 210 and an immersive video decoder 220 .

In this case, the immersive video encoder 210 may be an MPEG immersive video encoder, and the immersive video decoder 220 may be an MPEG immersive video decoder.

The immersive video encoder 210 performs partial linear scaling of the geometry atlas. At this time, the immersive video encoder 210 generates min-max normalized depth values; and generating geometry atlases by scaling the depth values to correspond to gradients of a plurality of linear sections.

The immersive video encoder 210 may perform various functions for encoding the immersive video in addition to partial linear scaling of the geometry atlas.

The immersive video decoder 220 reconstructs the immersive video by receiving the encoded immersive video and decoding it.

The immersive video decoder 220 may perform inverse scaling of the geometry atlas. At this time, the immersive video decoder 220 includes the steps of reconstructing the geometric atlas; and generating min-max normalized depth values by inversely scaling the geometry atlas according to slopes of a plurality of linear sections.

In addition to inverse scaling of the geometry atlas, the immersive video decoder 220 may perform various functions for decoding the immersive video and generating a virtual view image.

FIG. 3 is a block diagram illustrating an example of the immersive video encoder shown in FIG. 2 .

Referring to FIG. 3 , the immersive video encoder shown in FIG. 2 may include a partial linear scaling device 310 and an encoded immersive video generator 320 .

The partial linear scaling device 310 generates min-max normalized depth values and scales the depth values according to slopes of a plurality of linear sections to obtain geometry atlases. can create

The encoded immersive video generator 320 performs various functions for encoding the immersive video.

4 is a block diagram illustrating an example of the immersive video decoder shown in FIG. 2 .

Referring to FIG. 4 , the immersive video decoder illustrated in FIG. 2 may include an inverse scaling device 410 and a view synthesizer 420 .

The inverse scaling apparatus 410 reconstructs the geometry atlas and inversely scales the geometry atlas according to slopes of a plurality of linear sections to generate min-max normalized depth values.

The view synthesizer 420 generates a virtual view image through view synthesis.

5 is an operation flowchart illustrating a method of partial linear scaling of a geometry atlas according to an embodiment of the present invention.

Referring to FIG. 5 , the method for partial linear scaling of a geometry atlas according to an embodiment of the present invention generates min-max normalized depth values ( S510 ).

Also, according to the method of partial linear scaling of a geometry atlas according to an embodiment of the present invention, the depth values are scaled according to slopes of a plurality of linear sections to generate geometry atlases ( S520 ).

In this case, the linear intervals may be generated by dividing the entire range corresponding to the min-maximum normalized depth values into equal intervals of the same length.

In this case, the slopes may be determined based on sample occurrence frequencies of the linear sections.

In this case, each of the slopes may be set to be larger as the frequency of sample generation increases.

In this case, the sample occurrence frequencies may be clipped to a preset range.

In this case, the linear intervals use syntax fields including a first field (dq_interval_num) indicating the number of linear intervals and a second field (dq_norm_disp_pivot) indicating a scaled value corresponding to each of the linear intervals. can be signaled.

In this case, the syntax fields may further include a third field (dq_scaled_disp_start) indicating a start depth value to which the partial linear scaling is applied.

Each step shown in FIG. 5 may be performed in the immersive video encoder shown in FIG. 1 .

Although not explicitly shown in FIG. 5 , the method for partial linear scaling of a geometry atlas according to an embodiment of the present invention may further include generating an encoded immersive video signal using the geometry atlas. .

6 is an operation flowchart illustrating a method for inverse scaling of a geometry atlas according to an embodiment of the present invention.

Referring to FIG. 6 , the inverse scaling method of the geometry atlas according to an embodiment of the present invention reconstructs the geometry atlas ( S610 ).

In addition, the method for inverse scaling of a geometry atlas according to an embodiment of the present invention generates min-max normalized depth values by inversely scaling the geometry atlas according to the slopes of a plurality of linear sections. (S620).

In this case, the linear intervals may be generated by dividing an entire range corresponding to the min-maximum normalized depth values into equal intervals of the same length.

In this case, the slopes may be determined based on sample occurrence frequencies of the linear sections.

In this case, each of the slopes may be set to allocate more bits as the sample generation frequency increases.

In this case, the linear intervals use syntax fields including a first field (dq_interval_num) indicating the number of linear intervals and a second field (dq_norm_disp_pivot) indicating a scaled value corresponding to each of the linear intervals. can be signaled.

In this case, the syntax fields may further include a third field (dq_scaled_disp_start) indicating a start depth value to which the partial linear scaling is applied.

Each step shown in FIG. 6 may be performed in the immersive video decoder shown in FIG. 1 .

Although not explicitly shown in FIG. 6 , the inverse scaling method of the geometry atlas according to an embodiment of the present invention generates a virtual viewpoint image using the min-max normalized depth values. It may further include the step of

7 is a block diagram showing the configuration of a computer system according to an embodiment of the present invention.

The partial linear scaling apparatus of the geometry atlas, the inverse scaling apparatus of the geometry atlas, the encoded immersive video generator, the viewpoint synthesizer, the immersive video encoder, and the immersive video decoder according to the embodiment may include a computer such as a computer-readable recording medium. It may be implemented in system 700 .

Computer system 700 may include one or more processors 710 , memory 730 , user interface input device 740 , user interface output device 750 and storage 760 that communicate with each other via bus 720 . can In addition, computer system 700 may further include a network interface 770 coupled to network 780 . The processor 710 may be a central processing unit or a semiconductor device that executes programs or processing instructions stored in the memory 730 or storage 760 . The memory 730 and the storage 760 may be a storage medium including at least one of a volatile medium, a non-volatile medium, a removable medium, a non-removable medium, a communication medium, and an information delivery medium. For example, the memory 730 may include a ROM 731 or a RAM 732 .

At this time, at least one program may be recorded in the memory 730 .

In this case, the processor 710 may execute the program. At this time, the program generates min-max normalized depth values; and generating geometry atlases by scaling the depth values to correspond to gradients of a plurality of linear sections.

In this case, the linear intervals may be generated by dividing an entire range corresponding to the min-maximum normalized depth values into equal intervals of the same length.

In this case, the slopes may be determined based on sample occurrence frequencies of the linear sections.

In this case, each of the slopes may be set to be larger as the frequency of sample generation increases.

In this case, the linear sections include a first field indicating the number of the linear sections; and a second field indicating a scaled value corresponding to each of the linear sections.

According to another method of partial linear scaling of a geometry atlas according to an embodiment of the present invention, loss of depth information is minimized and efficiency of view synthesis is secured. The partial linear scaling method of a geometry atlas according to an embodiment of the present invention may generate a partial linear scaling model based on an occurrence frequency of a depth value in an additional view.

As described above, the configuration and method of the embodiments described above are not limitedly applicable to the partial linear scaling method, the inverse scaling method, and the apparatuses for the geometry atlas according to the present invention, but the embodiments are various All or part of each of the embodiments may be selectively combined so that modifications may be made.

210: immersive video encoder
220: immersive video decoder

Claims (20)

In a piecewise linear scaling method of a geometric atlas performed in an immersive video encoder,
generating min-max normalized depth values; and
generating geometry atlases by scaling the depth values according to slopes of a plurality of linear sections
A partial linear scaling method comprising: The method according to claim 1,
The linear sections are
A partial linear scaling method, wherein an entire range corresponding to the min-maximum normalized depth values is divided into equal intervals of the same length. The method according to claim 1,
The partial linear scaling method is
generating an encoded immersive video signal using the geometric atlases. The method according to claim 1,
The slopes are
and is determined based on sample occurrence frequencies of the linear intervals. 5. The method according to claim 4,
Each of the slopes is
Partial linear scaling method, which is set larger as the frequency of occurrence of samples is higher. 5. The method according to claim 4,
wherein the sample occurrence frequencies are clipped to a predetermined range. 5. The method according to claim 4,
The linear sections are
a first field indicating the number of the linear sections; and
Signaled using syntax fields including a second field indicating a scaled value corresponding to each of the linear intervals. 8. The method of claim 7,
The syntax fields are
and a third field indicating a starting depth value to which the partial linear scaling is applied. In the inverse scaling method of the geometric atlas performed in the immersive video decoder,
reconstructing the geometric atlas; and
generating min-max normalized depth values by inversely scaling the geometric atlas according to the slopes of a plurality of linear sections;
A method of inverse scaling of a geometry atlas, comprising: 10. The method of claim 9,
The linear sections are
An inverse scaling method of a geometry atlas, wherein an entire range corresponding to the min-maximum normalized depth values is divided into equal intervals of the same length. 10. The method of claim 9,
The inverse scaling method of the geometry atlas is
The method of claim 1, further comprising generating a virtual viewpoint image using the min-max normalized depth values. 10. The method of claim 9,
The slopes are
A method for inverse scaling of a geometry atlas, which is determined based on sample occurrence frequencies of the linear intervals. 13. The method of claim 12,
Each of the slopes is
A method of inverse scaling of a geometry atlas, in which the higher the frequency of sample occurrence, the more bits are set to be allocated. 13. The method of claim 12,
The linear sections are
a first field indicating the number of the linear sections; and
The inverse scaling method of a geometry atlas, which is signaled using syntax fields including a second field indicating a scaled value corresponding to each of the linear sections. 15. The method of claim 14,
The syntax fields are
and a third field indicating a starting depth value to which the partial linear scaling is applied. a memory in which at least one program is recorded; and
a processor executing the program;
The program is
generating min-max normalized depth values; and
and instructions for generating geometry atlases by scaling the depth values according to slopes of a plurality of linear sections. 17. The method of claim 16,
The linear sections are
and an entire range corresponding to the min-maximum normalized depth values is generated by being divided into equal intervals of the same length. 17. The method of claim 16,
The slopes are
and is determined based on sample occurrence frequencies of the linear intervals. 19. The method of claim 18,
Each of the slopes is
Partial linear scaling device, which is set larger as the frequency of sample occurrence is higher. 19. The method of claim 18,
The linear sections are
a first field indicating the number of the linear sections; and
Signaled using syntax fields including a second field indicating a scaled value corresponding to each of the linear sections.

Images