KR102042343B1 - Apparatus and method for genaration of depth image based on point-cloud data obtained by 3d scanning and method for identifing 3d object using the same - Google Patents

Abstract

The present invention includes a shape data generation unit for generating three-dimensional shape data consisting of point group data obtained by scanning an object; A viewpoint processing unit for estimating a center point from the 3D shape data and projecting a unit sphere having the same distance as the center point to the 3D shape data; And determining a first viewpoint on the surface of the unit sphere, determining a first resolution based on the first viewpoint, and generating a first depth image having the first viewpoint and the first resolution from the three-dimensional shape data. An image processing unit; The image processing unit may include a plurality of information acquisition rates A (v, w) and information densities D (v, w) through a plurality of depth images each having a plurality of viewpoints on the surface of the unit sphere. A calculation unit for calculating; And a determining unit to determine the first time point and the first resolution based on the information acquisition rate and the information density. Disclosed are an apparatus and method for generating a depth image based on point cloud data obtained through 3D scanning including a 3D object and a 3D object identification method using the same.

Description

Apparatus and method for generating depth image based on point cloud data obtained through 3D scanning and method for identifying 3D object using the same 3D OBJECT USING THE SAME}

The present invention relates to an apparatus and method for generating a depth image based on point cloud data obtained through three-dimensional scanning, and a three-dimensional object identification method using the same.

A plant is a collection of facilities that process and process raw materials to produce a specific product. For example, a marine plant is a facility that extracts, drills and produces resources such as oil and gas buried in the sea, and a freshwater plant is a factory facility that processes seawater and obtains fresh water therefrom. Such plants generally have a complex intertwining of mechanical equipment for processing or controlling the fluid and plumbing equipment for providing the fluid.

It is often the case that the plant's 3D CAD data is needed for the maintenance of the plant during operation of the plant. However, plants built a long time ago often have only 2D drawings, and it is time-consuming to produce 3D CAD data from the plant's 2D drawings to the entire plant. In addition, even if a new plant is being built, the owner of the plant is given 3D CAD data that reflects the actual appearance of the plant to ensure that the finished plant is built as originally designed, even if the plant is provided with the plant's 3D CAD data. May be required.

The present invention generates a depth image of the facility from the point group data obtained by three-dimensional scanning of the facility, such as offshore structures, and detects the shape data of the facility matching the depth image to generate the shape modeling data of the facility And to provide a three-dimensional object identification method using the same.

On the other hand, the technical problems to be achieved in the present invention are not limited to the technical problems mentioned above, and other technical problems that are not mentioned are clearly to those skilled in the art from the following description. It can be understood.

Depth image generating apparatus according to an embodiment of the present invention comprises a shape data generation unit for generating three-dimensional shape data consisting of point group data obtained by scanning the object; A viewpoint processing unit for estimating a center point from the 3D shape data and projecting a unit sphere having the same distance as the center point to the 3D shape data; And determining a first viewpoint on the surface of the unit sphere, determining a first resolution based on the first viewpoint, and generating a first depth image having the first viewpoint and the first resolution from the three-dimensional shape data. An image processing unit; The image processing unit may include a plurality of information acquisition rates A (v, w) and information densities D (v, w) through a plurality of depth images each having a plurality of viewpoints on the surface of the unit sphere. A calculation unit for calculating; And a determining unit to determine the first time point and the first resolution based on the information acquisition rate and the information density. It includes.

The calculating unit may calculate the information acquisition rate through the following [Equation 1].

[Equation 1]

는 단위구의 표면 상의 한 지점인 위치이고,

는 깊이 이미지의 해상도이고, N_pixel은 시점 개수와 해상도 개수를 기반의 깊이 이미지의 전경(foreground) 픽셀의 개수이며, N_p는 점군 데이터의 점 개수이다.)(here,

Is a location on the surface of the unit sphere,

Is the resolution of the depth image, N _pixel is the number of foreground pixels of the depth image based on the number of viewpoints and the resolution number, and N _p is the number of points of the point cloud data.)

The calculating unit may calculate the information density through the following [Equation 2].

[Formula 2]

(N _8-connected is the number of neighboring eight foreground pixels in the depth image in the up, down, left, right and diagonal directions.)

The determination unit may determine the first time point through the following [Equation 3].

[Equation 3]

(Here, Viewpoint_1st is the first time point, and Normalized_A (v, w) normalizes the information acquisition rate derived for each resolution to [0,1].)

The determination unit may determine the first resolution through Equation 4 below.

[Equation 4]

The viewpoint processing unit proportionally reduces the 3D shape data such that the distance between the center point of the 3D shape data and the outermost point of the point group data is 1, and proportionally reduces the unit sphere having the center point and distance 1 It may be 1 projecting onto the shape data.

The surface of the unit sphere is composed of a triangular network or a square network, the viewpoint processing unit may set each of the intersections of the straight lines constituting the triangular or square network to the plurality of viewpoints.

Depth image generation method according to an embodiment of the present invention comprises the steps of generating three-dimensional shape data consisting of point group data obtained by scanning the object; Estimating a center point from the three-dimensional shape data and projecting a unit sphere having the same distance as the center point to the three-dimensional shape data; Setting each specific point of the surface of the unit sphere to a plurality of viewpoints; Determining a first viewpoint at the surface of the unit sphere; Determining a first resolution based on the first viewpoint; And generating a first depth image having the first viewpoint and the first resolution from the three-dimensional shape data. It may include.

The generating of the depth image may include obtaining information rate A (v, w) and information density D (v, w) through a plurality of depth images each having a plurality of viewpoints on the surface of the unit sphere. Calculating; Determining the first time point and the first resolution based on the information acquisition rate and the information density; It may include.

In the calculating of the information acquisition rate and the information density, the information acquisition rate may be calculated through Equation 1 below.

[Equation 1]

는 단위구의 표면 상의 한 지점인 위치이고,

는 깊이 이미지의 해상도이고, N_pixel은 시점 개수와 해상도 개수를 기반의 깊이 이미지의 전경(foreground) 픽셀의 개수이며, N_p는 점군 데이터의 점 개수이다.)(here,

Is a location on the surface of the unit sphere,

Is the resolution of the depth image, N _pixel is the number of foreground pixels of the depth image based on the number of viewpoints and the resolution number, and N _p is the number of points of the point cloud data.)

In the calculating of the information acquisition rate and the information density, the information density may be calculated through Equation 2 below.

[Formula 2]

(N _8-connected is the number of neighboring eight foreground pixels in the depth image in the up, down, left, right and diagonal directions.)

In the determining of the first time point, the first time point may be determined through Equation 3 below.

[Equation 3]

(Here, Viewpoint_1st is the first time point, and Normalized_A (v, w) normalizes the information acquisition rate derived for each resolution to [0,1].)

The determining of the first resolution may determine the first resolution through Equation 4 below.

[Equation 4]

The projecting of the unit sphere on the 3D shape data may include: proportionally reducing the 3D shape data such that a distance between a center point of the 3D shape data and the outermost point of the point group data is 1; And projecting the unit sphere whose distance from the center point of the proportionally reduced three-dimensional shape data is 1 to the proportionally reduced three-dimensional shape data. It may include.

In the projecting of the unit sphere on the 3D shape data, the surface of the unit sphere may be composed of a triangular network or a square network.

In the setting of the plurality of viewpoints, each of the intersection points of the straight lines constituting the triangular network or the square network in the unit sphere may be set as the plurality of viewpoints.

3D object identification method according to an embodiment of the present invention comprises the steps of generating three-dimensional shape data consisting of point group data obtained by scanning the object; Estimating a center point from the three-dimensional shape data and projecting a unit sphere having the same distance as the center point to the three-dimensional shape data; Setting each specific point of the surface of the unit sphere to a plurality of viewpoints; Determining a first time point at the surface of the unit sphere; Determining a first resolution of with respect to the first viewpoint; Generating a first depth image having the first viewpoint and the first resolution from the three-dimensional shape data; And matching the first depth image with pre-stored shape modeling data to identify the object. It includes.

The generating of the depth image may include obtaining information rate A (v, w) and information density D (v, w) through a plurality of depth images each having a plurality of viewpoints on the surface of the unit sphere. Calculating; Determining the first time point and the first resolution based on the information acquisition rate and the information density; It may include.

The calculating of the information acquisition rate and the information density may be performed by calculating the information acquisition rate and the information density by the following Equation 1 and Equation 2, and determining the first time point by the following Equation 3. The determining of the first view point and determining the first resolution may determine the first resolution through Equation 4 below.

[Equation 1]

는 단위구의 표면 상의 한 지점인 위치이고,

는 깊이 이미지의 해상도이고, N_pixel은 시점 개수와 해상도 개수를 기반의 깊이 이미지의 전경(foreground) 픽셀의 개수이며, N_p는 점군 데이터의 점 개수이다.)(here,

Is a location on the surface of the unit sphere,

Is the resolution of the depth image, N _pixel is the number of foreground pixels of the depth image based on the number of viewpoints and the resolution number, and N _p is the number of points of the point cloud data.)

 [Formula 2]

(N _8-connected is the number of neighboring eight foreground pixels in the depth image in the up, down, left, right and diagonal directions.)

[Equation 3]

(Here, Viewpoint_1st is the first time point, and Normalized_A (v, w) normalizes the information acquisition rate derived for each resolution to [0,1].)

 [Equation 4]

Placing shape modeling data of the identified object into three-dimensional modeling data; It may further include.

According to an embodiment of the present invention, by generating the shape modeling data of the facility from the point group data obtained by scanning the facility, it is possible to save time and manpower for securing the 3D CAD data of the facility.

In particular, an embodiment of the present invention provides an algorithm for selecting an optimal viewpoint and an optimal image resolution for a point group obtained by 3D scanning a large facility, thereby enabling automated shape modeling data generation.

On the other hand, the effect obtained in the present invention is not limited to the above-mentioned effects, other effects that are not mentioned will be clearly understood by those skilled in the art from the following description. Could be.

1 is a block diagram schematically showing a three-dimensional modeling apparatus according to an embodiment of the present invention,
2 is a diagram illustrating three-dimensional shape data composed of point group data acquired according to an embodiment of the present invention.
3 is an exemplary view of projecting a unit sphere after proportionally reducing the three-dimensional shape data of FIG. 2;
4 is an exemplary view comparing the results of generating depth images of a specific viewpoint from the 3D shape data of FIG. 2,
5 is a diagram illustrating an information acquisition rate in a depth image generated according to a viewpoint and a resolution.
6 is a diagram illustrating a normalized information acquisition rate in a depth image generated according to a viewpoint and a resolution.
7 is a diagram illustrating a sum of normalized information acquisition rates at each time point,
8 is an exemplary flowchart of a 3D object identification method according to an embodiment of the present invention.

The embodiments of the present invention are provided to more fully explain the present invention to those skilled in the art, and the following examples can be modified in various other forms, and the scope of the present invention is It is not limited to an Example. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the inventive concept to those skilled in the art.

In addition, in the following drawings, each configuration is exaggerated for convenience and clarity of description, the same reference numerals in the drawings refer to the same elements. As used herein, the term “and / or” includes any and all combinations of one or more of the listed items.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

As used herein, the singular forms "a", "an" and "the" may include the plural forms as well, unless the context clearly indicates otherwise. Also, as used herein, "comprise" and / or "comprising" specifies the presence of the mentioned shapes, numbers, steps, actions, members, elements and / or groups of these. It is not intended to exclude the presence or the addition of one or more other shapes, numbers, acts, members, elements and / or groups.

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

1 is a block diagram schematically showing a three-dimensional modeling apparatus according to an embodiment of the present invention.

First, referring to FIG. 1, the 3D modeling apparatus 10 according to an exemplary embodiment of the present invention may include a shape data generator 111, a viewpoint processor 112, an image processor 113, and an object identifier 114. It may include.

Here, the shape data generator 111, the viewpoint processor 112, and the image processor 113 may be a depth image generator described in an embodiment of the present invention.

The shape data generation unit 111 may generate three-dimensional shape data from the point group data obtained by scanning the facility. Hereinafter, the object to be scanned may be mixed and described as an object included in the facility, without being limited to the facility.

Meanwhile, the three-dimensional shape data in the present invention refers to a three-dimensional shape formed of the point group data, and may be mixed with the point group data, but may be distinguished from the three-dimensional modeling data to be described later.

The viewpoint processor 112 may proportionally reduce the three-dimensional shape data of the object, and project a unit sphere onto the proportionally reduced three-dimensional shape data to set a plurality of viewpoints on the surface of the unit sphere.

The image processor 113 determines an optimal first viewpoint on the surface of the unit sphere, and then determines an optimal first resolution based on the first viewpoint, converts three-dimensional shape data at the first viewpoint and the first resolution, One depth image can be created.

The object identifier 114 may identify the scanned facility through the shape modeling data of the facility that matches the first depth image. Moreover, shape modeling data can be arrange | positioned suitably to facility layout data.

Meanwhile, the shape data generator 111, the viewpoint processor 112, the image processor 113, and the object identifier 114 execute a program for generating the 3D modeling data to generate the 3D modeling data from the point cloud data. As an example of the processor 110, the processor 110 may include a CPU and a GPU.

In addition, a program for generating three-dimensional modeling data according to an embodiment of the present invention may be stored in the storage unit 120 connected to the processor 110, the processor 110 is a program from the storage unit 120 Can be invoked and executed. In addition, the storage unit 120 may store and provide various types of data necessary for generating 3D modeling data to the processor 110 according to an exemplary embodiment.

The 3D modeling apparatus 10 according to an exemplary embodiment of the present invention receives the point group data of the object for which the shape modeling data is to be generated and generates the shape modeling data of the object therefrom. The point cloud data can be obtained by three-dimensional scanning of the object, for example, by processing the optical signal returned by projecting a laser to the facility to obtain three-dimensional coordinate information for a plurality of points on the object surface.

Point group data is a collection of multiple points located on the surface of an object, each point having three-dimensional coordinates in a given coordinate system. The three-dimensional coordinates may be orthogonal coordinates composed of x-axis coordinates, y-axis coordinates, and z-axis coordinates, but is not limited thereto.

In addition, the point group data of the object may be obtained by scanning not only a laser but also white light on the object, or may be obtained by using other measuring means such as sound waves in addition to light.

For example, when three-dimensional scanning is performed by irradiating a laser at a plurality of different locations on a facility such as a plant, individual point group data is generated for each three-dimensional scanning. Such individual point group data may be applied to a predefined reference coordinate system to generate one matched point group data for the corresponding facility.

In particular, in one embodiment of the present invention, the shape data generator 111 may divide the point group data of a specific object from the point group data of the facility.

Here, the point group data of the entire facility may be matched point group data or single point group data, and the point group data of a specific object may be segmented point cloud data obtained by dividing points having the same attributes from the matched (single) point group data. Can be.

That is, FIG. 2 is a diagram illustrating three-dimensional shape data composed of point group data obtained according to an embodiment of the present invention.

As shown in FIG. 2, the shape data generation unit 111 may generate 3D shape data of the entire facility and 3D shape data of a specific object from point group data obtained by scanning an object.

3 is an exemplary view of projecting a unit sphere after proportionally reducing the three-dimensional shape data of FIG. 2.

Referring to FIG. 3, the viewpoint processor 112 estimates the center point C of the 3D shape data 1a generated by the shape data generation unit 111, and the unit sphere 2 formed around the center point C. ) Can be projected onto the three-dimensional shape data 1a, and a specific point facing the center point C on the surface of the unit sphere 2 can be set to a plurality of viewpoints V1, V2, ?? Vn _V.

The center point C of the three-dimensional shape data 1a may be set to a coordinate having the lowest standard deviation of the distance formed from the plurality of point coordinates constituting the point group data.

On the other hand, the viewpoint processing unit 112 selects the outermost point having the largest distance to the center point C among the plurality of point coordinates constituting the point group data, so that the distance between the outermost point coordinate and the center point C becomes 1. The three-dimensional shape data 1a can be proportionally reduced.

In addition, the viewpoint processor 112 may generate a unit sphere 2 having a distance of 1 from the center point C, and may project the unit sphere 2 onto the three-dimensional shape data 1a which is scaled down.

Here, the unit sphere 2 may be a three-dimensional sphere shape of the surface consisting of a triangular network or a square network, the viewpoint processing unit 112 a plurality of intersection points of the straight lines forming a triangular or square network in the unit sphere (2). Can be set to the time points V1, V2, and Vn _V.

The image processor 113 may generate depth images viewed from the plurality of viewpoints V1, V2, and Vn _V processed by the viewpoint processor 112.

In particular, the image processor 113 may determine the first view point and the first resolution and generate a first depth image having the first view point and the first resolution.

Here, the first depth image is an optimal depth image with little information loss when converting the 3D shape data into a depth image, the first viewpoint is an optimal viewpoint for generating the first depth image, and the first resolution is 1 means the optimal resolution for creating a depth image.

On the other hand, since the viewpoint processing unit 112 performs position and pose normalization on the point group data, since it is known that all the point group data exist around the center point, the image processing unit 113 looks at the center point when the first view point is determined. You can define the scanning position.

In addition, when the point cloud data is converted into normalized device coordinates (NDC) through objects, viewpoints, and projection matrices, the point group data is converted into screen coordinates according to the resolution. Decisions must also be made.

FIG. 4 is an exemplary view comparing the results of generating depth images of specific views from the 3D shape data of FIG. 2.

1 to 4, the image processor 113 generates a depth image at each of the plurality of viewpoints V1, V2, and Vn _V , and defines an information acquisition rate, A (v), as defined in the present invention. , w)) and information density (D (v, w)), and then the first viewpoint and the first resolution may be determined based on this.

That is, the image processor 113 may determine a first view point in which as many point group data are projected as pixels of the depth image and a first resolution capable of calculating SIFT features well.

The image processor 113 includes a calculator 113a and a determiner 113b.

The calculation unit 113a may calculate the information acquisition rate A (v, w) and the information density D (v, w).

First, the image processing unit 113 may generate a N * N _{v w} of the depth image through a resolution of N _w N _v determined as the view points generated by the unit surface of a sphere.

Thereafter, the calculating unit 113a calculates N _pixels by calculating the number of pixels in each depth image, and calculates the information acquisition rate A (v, w) as shown in Equation 1 below. Can be.

는 단위구의 표면 상의 한 지점인 위치이고,

는 깊이 이미지의 해상도이고, N_pixel은 시점 개수와 해상도 개수를 기반의 깊이 이미지의 전경(foreground) 픽셀의 개수이며, N_p는 점군 데이터의 점 개수이다.here,

Is a location on the surface of the unit sphere,

Is the resolution of the depth image, N _pixel is the number of foreground pixels of the depth image based on the number of viewpoints and the resolution number, and N _p is the number of points of the point group data.

That is, the information acquisition rate A (v, w) indicates how much information of the point group data is reflected in the depth image as a ratio with respect to the input point group data.

In addition, the calculator 113a may calculate the information density D (v, w) through pixel information in each depth image as shown in Equation 2 below.

Here, N _{8 -connected} is the number of pixels adjacent to eight foreground pixels in up, down, left, right and diagonal directions in the depth image.

That is, the information density D (v, w) indicates how dense the points reflected in the depth image exist as pixels as a ratio of the total number of pixels projected on the depth image.

5 is a diagram illustrating an information acquisition rate in a depth image generated according to a viewpoint and a resolution.

Referring to FIG. 5, an example of an information acquisition rate A (v, w) calculated according to a viewpoint and a resolution may be confirmed by the calculator 113a.

As the resolution is higher, the information acquisition rate value tends to be higher overall. This is because the higher the resolution of an image, the more likely that adjacent points are projected onto the same pixel, thereby increasing saturation. Each pixel in a depth image can encode only one depth value, so when multiple points in a point group correspond to one pixel, only the depth value of the nearest point (the smallest depth value) can be written to the pixel. .

In other words, when creating a depth image, when the resolution is high, all depth values of adjacent points can be recorded, whereas at low resolution, only the depth value of a specific point among the dense points is recorded, and the depth value of the remaining points is not included in the image. Therefore, the number of projected point groups may be reduced overall.

Thereafter, the determination unit 113b may determine the first time point as described above using the normalized information acquisition rate (Normalized_A) value.

First, the determiner 113b may normalize the information acquisition rate derived for each resolution to the range [0,1] as shown in Equation 3 below.

In addition, the determination unit 113b may determine the optimal first view point Viewpoint_1st as shown in Equation 4 below.

FIG. 6 is a diagram illustrating a normalized information acquisition rate in a depth image generated according to a viewpoint and a resolution, and FIG. 7 is a diagram illustrating a sum of normalized information acquisition rates at each viewpoint.

6 and 7 together, before determining the proper resolution, the determination unit 113b may probabilistically determine a first time point for deriving a high normalized information acquisition rate value over several resolutions.

In particular, as illustrated in FIG. 7, the normalized information acquisition rate (Normalized_A) sum value is highest at a specific time point, and a time point indicating this peak value may be determined as the first time point in the determination unit 113b. have.

Also, the determiner 113b may determine, as the first resolution of the depth image, a resolution at which the information density becomes the maximum value based on the determined first viewpoint. That is, the determination unit 113b may determine the optimal first resolution View_1_1 as shown in Equation 5 below.

That is, the image processor 113 may provide an algorithm for selecting an optimal viewpoint and an optimal image resolution without requiring a user's input for information on the viewpoint and the resolution, thereby automating the generation of the first depth image. .

The object identifier 114 may identify the scanned object by comparing and analyzing the first depth image provided by the image processor 113 and the shape modeling data previously stored in the storage 120. In addition, the object identification unit 114 may provide shape modeling data of the identified object to the facility arrangement data.

8 is an exemplary flowchart of a 3D object identification method according to an embodiment of the present invention.

Referring to FIG. 8, in the method of identifying a 3D object according to an embodiment of the present disclosure, the method may include generating a depth image (S10) and identifying the object based on the generated depth image, and 3D shape shape data of the object. The step S20 includes the facility layout data.

Meanwhile, the generating of the depth image (S10) may include generating three-dimensional shape data of the object (S11), setting a plurality of viewpoints (S12), determining a first viewpoint (S13), and first Determining the resolution (S14), generating a first depth image (S15) may be included.

Identifying the object based on the generated depth image, and placing the shape modeling data of the object in the three-dimensional facility layout data (S20), the step of identifying the object (S21) and the shape modeling data to the three-dimensional facility layout data It may include the step (S22) to place.

In operation S11 of generating the 3D shape data of the object, the shape data generation unit 111 may generate the 3D shape data of the entire facility and the 3D shape data of the specific object from the point group data obtained by scanning the object. .

In the step S12 of setting a plurality of viewpoints, the viewpoint processing unit 112 proportionally reduces the three-dimensional shape data of the object, projects the unit sphere on the proportionally reduced three-dimensional shape data, and displays the plurality of viewpoints on the surface of the unit sphere. Can be set.

In the determining of the first viewpoint (S13), as described above, the information acquisition rate A (v, w) and the information density D through [Equation 1] and [Equation 2] in the image processing unit 113. (v, w)), and a first time point may be determined through Equation 3 and Equation 4.

In the determining of the first resolution (S14), as described above, the first resolution may be determined through Equation 5 based on the first viewpoint determined by the image processor 113.

In operation S15 of generating the first depth image, a first depth image having the determined optimal first viewpoint and the first resolution may be generated.

Here, the first depth image is an optimal depth image with little information loss when converting the 3D shape data into a depth image, the first viewpoint is an optimal viewpoint for generating the first depth image, and the first resolution is 1 means the optimal resolution for creating a depth image.

In the step of identifying the object (S21), the object identifying unit 114 may identify the scanned object through the shape modeling data of the facility that matches the first depth image.

In the step S22 of disposing the shape modeling data in the three-dimensional facility layout data, the object modeling unit 114 transmits the shape modeling data corresponding to the object as the layout data of the entire facility, on the layout data of the entire facility. You can place shape modeling data of an object.

What has been described above is only one embodiment for implementing a three-dimensional object identification method and apparatus and method for generating a depth image based on the point group data obtained through the three-dimensional scanning according to the present invention, Not limited to one embodiment, as claimed in the following claims to those skilled in the art to which the present invention pertains without departing from the gist of the present invention to the extent that various modifications can be made Would have a technical spirit.

10: three-dimensional modeling device 110: processor
111: shape data generation unit 112: viewpoint processing unit
113: image processing unit 114: object identification unit
120: storage unit

Claims (20)

A shape data generator for generating three-dimensional shape data composed of point group data obtained by scanning an object;
A viewpoint processing unit for estimating a center point from the 3D shape data and projecting a unit sphere having the same distance as the center point to the 3D shape data; And
A first viewpoint is determined on a surface of the unit sphere, a first resolution is determined based on the first viewpoint, and a first depth image having the first viewpoint and the first resolution is generated from the three-dimensional shape data. An image processing unit; Including,
The image processing unit
A calculation unit for calculating each information acquisition rate (A (v, w)) and information density (D (v, w)) through a plurality of depth images each having a plurality of viewpoints on the surface of the unit sphere; And
A determination unit to determine the first time point and the first resolution based on the information acquisition rate and the information density; Including.
And the calculating unit calculates an information acquisition rate through the following [Equation 1].
[Equation 1]

(here,

Is a location on the surface of the unit sphere,

Is the resolution of the depth image, N _pixel is the number of foreground pixels of the depth image based on the number of viewpoints and the resolution number, and N _p is the number of points of the point cloud data.)
delete The method of claim 1,
The calculating unit is a depth image generating device for calculating the information density through the following [Equation 2].
[Formula 2]

(N _8-connected is the number of neighboring eight foreground pixels in the depth image in the up, down, left, right and diagonal directions.)
The method of claim 3, wherein
And the determining unit determines the first viewpoint through the following [Equation 3].
[Equation 3]

(Here, Viewpoint_1st is the first time point, and Normalized_A (v, w) normalizes the information acquisition rate derived for each resolution to [0,1].)
The method of claim 4, wherein
And the determining unit determines the first resolution through the following [Equation 4].
[Equation 4]

The method of claim 5,
The viewpoint processing unit
Proportionally reducing the 3D shape data such that the distance between the center point of the 3D shape data and the outermost point of the point group data is 1,
A one-person depth image generating device for projecting a unit sphere having a center point and a distance of one to proportionally scaled three-dimensional shape data.
The method of claim 6,
The surface of the unit sphere is composed of a triangular network or a square network,
The viewpoint processing unit
And a depth image generating device for setting each of the intersection points of the straight lines forming the triangular or quadrangular network as the plurality of viewpoints.
Generating three-dimensional shape data composed of point group data obtained by scanning an object;
Estimating a center point from the three-dimensional shape data and projecting a unit sphere having the same distance as the center point to the three-dimensional shape data;
Setting each specific point of the surface of the unit sphere to a plurality of viewpoints;
Determining a first viewpoint at the surface of the unit sphere;
Determining a first resolution based on the first viewpoint; And
Generating a first depth image having the first viewpoint and the first resolution from the three-dimensional shape data; Including,
The generating of the first depth image may include the information acquisition rate A (v, w) and the information density D (v, w) through a plurality of depth images each having a plurality of viewpoints on the surface of the unit sphere. Calculating; And
Determining the first time point and the first resolution based on the information acquisition rate and the information density; Including,
The calculating of the information acquisition rate and the information density may include calculating the information acquisition rate through the following [Equation 1].
[Equation 1]

(here,

Is a location on the surface of the unit sphere,

Is the resolution of the depth image, N _pixel is the number of foreground pixels of the depth image based on the number of viewpoints and the resolution number, and N _p is the number of points of the point cloud data.)
delete delete The method of claim 8,
The calculating of the information acquisition rate and the information density may include calculating an information density using Equation 2 below.
[Formula 2]

(N _8-connected is the number of neighboring eight foreground pixels in the depth image in the up, down, left, right and diagonal directions.)
The method of claim 11,
The determining of the first viewpoint may include determining the first viewpoint through the following [Equation 3].
[Equation 3]

(Here, Viewpoint_1st is the first time point, and Normalized_A (v, w) normalizes the information acquisition rate derived for each resolution to [0,1].)
The method of claim 12,
The determining of the first resolution may include determining the first resolution through Equation 4 below.
[Equation 4]

The method of claim 13,
Projecting the unit sphere to the three-dimensional shape data
Proportionally reducing the 3D shape data such that a distance between a center point of the 3D shape data and the outermost point of the point group data is 1; And
Projecting the unit sphere whose distance from the center point of the proportionally reduced three-dimensional shape data is 1 to the proportionally reduced three-dimensional shape data; Depth image generation method comprising a.
The method of claim 14,
In the step of projecting the unit sphere to the three-dimensional shape data,
Surface of the unit sphere is a depth image generation method consisting of a triangular network or a square network.
The method of claim 15,
In the step of setting to the plurality of viewpoints,
The depth image generation method of setting each intersection point of the straight lines constituting the triangular or square network in the unit sphere as the plurality of viewpoints.
Generating three-dimensional shape data composed of point group data obtained by scanning an object;
Estimating a center point from the three-dimensional shape data and projecting a unit sphere having the same distance as the center point to the three-dimensional shape data;
Setting each specific point of the surface of the unit sphere to a plurality of viewpoints;
Determining a first time point at the surface of the unit sphere;
Determining a first resolution of with respect to the first viewpoint;
Generating a first depth image having the first viewpoint and the first resolution from the three-dimensional shape data; And
Matching the first depth image with pre-stored shape modeling data to identify the object; Including,
Generating the first depth image
Calculating each information acquisition rate (A (v, w)) and information density (D (v, w)) through a plurality of depth images each having a plurality of viewpoints on the surface of the unit sphere; And
Determining the first time point and the first resolution based on the information acquisition rate and the information density; Including,
Computing the information acquisition rate (A (v, w)) and the information density (D (v, w)) of each of the three-dimensional object identification method using the following equation (1).
[Equation 1]

(here,

Is a location on the surface of the unit sphere,

Is the resolution of the depth image, N _pixel is the number of foreground pixels of the depth image based on the number of viewpoints and the resolution number, and N _p is the number of points of the point cloud data.)
delete The method of claim 17,
The calculating of the information acquisition rate and the information density may be performed by calculating the information density through the following [Formula 2],
The determining of the first time point may include determining the first time point through the following Equation 3,
The determining of the first resolution may include determining the first resolution through Equation 4 below.
[Formula 2]

(N _8-connected is the number of neighboring eight foreground pixels in the depth image in the up, down, left, right and diagonal directions.)
[Equation 3]

(Here, Viewpoint_1st is the first time point, and Normalized_A (v, w) normalizes the information acquisition rate derived for each resolution to [0,1].)
[Equation 4]

The method of claim 19,
Placing shape modeling data of the identified object into three-dimensional modeling data; Three-dimensional object identification method further comprising.

Images