JPH05250445A - Three-dimensional model data generating device - Google Patents

Abstract

PURPOSE:To generate three-dimensional model data for generating no discontinuity in a display by deleting apex data, based on a space frequency of a shape of a three-dimensional model. CONSTITUTION:By using a digitizer, three-dimensional coordinate data of the apex for constituting the surface of a three-dimensional model which becomes an object of a display, and texture data on the surface of the model are inputted in detail. Accordingly, apex data is obtained in detail at every apex (S101). A triangular patch prescribed by three adjacent apexes is generated (S102). With regard to each triangular patch, a normal vector is derived by an operation (S103). By using the detected normal vector, a variation of a space frequency is detected, and based on a result of detection the space frequency variation, an integration processing of the adjacent triangular patches is executed (S104). By deleting the apex data in the integrated area, and deleting the apex data, as well of the outline part of the integrated graphic, data of a hierarchical model to be generated is decreased (S105).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a three-dimensional model data generator, and more particularly to a three-dimensional model data generator for generating model data suitable for high-speed display of a three-dimensional model.

[0002]

2. Description of the Related Art Fields in which it is required to display a highly realistic image on a screen at high speed by using computer graphics (CG), for example, provide a sensation as if remote people were in the same space. In the immersive teleconferencing system that is required to do so, there is a demand for display technology for enhancing the sense of unity between the virtual space displayed on the screen and the real space where the viewer of the screen exists. Therefore, realization of wide-field display technology with a wide screen field, high-definition display technology with a large number of pixels on the display screen, and motion-vision display technology that displays the image viewed through the screen from the observer's viewpoint position in correspondence with the viewpoint position. Is desired.

When the motion vision is realized by using computer graphics, a display object is defined by a three-dimensional model,
The viewpoint position of the observer is obtained, and the perspective projection image of the target object viewed from the viewpoint position is displayed on the screen.

In the computer graphics technique for generating and displaying a three-dimensional object by a computer, there is a contradictory relationship between high precision of a target object, that is, high quality of a display image and high-speed image generation. That is, in order to generate a high quality image, the target object must be defined with detailed resolution, the number of vertices forming the target object increases, and the processing time for image generation increases.

On the other hand, in order to display the change of the display object in real time, the current computer lacks the processing capability, and only a simple model with a small number of vertices can be displayed.
In order to solve this problem, two types of research have been conducted so far. One is a method of improving the processing capability of a computer, and the other is a method of reducing the number of vertices as much as possible and utilizing a human visual angle characteristic or the like in order to perform a highly realistic display. Here, the latter will be described in detail.

In the case where a three-dimensional model of a target object is created, if a general model is defined for the target object, no matter what size the display target is on the screen and at what angle. It must be displayed in a finely tuned manner without deterioration of image quality. In order to finely model all target objects, as the number of display objects increases, the time required for image generation and display increases and real-time processing becomes difficult. Therefore, some proposals have been made on a method of reducing the number of vertices without giving the deterioration of the image quality to those who are viewing the display image.

For example, in the case of (1) perspective projection display,
A method that uses the fact that as the distance from the viewpoint to the display target increases, the display target becomes smaller on the screen and the image quality is less degraded even if it is roughly modeled. (2) Since human eyesight has the highest gazing point and the visual acuity decreases as it becomes peripheral vision, it is used that the image in peripheral vision is less likely to deteriorate even if it is roughly modeled and displayed. Technique.
(3) A method that utilizes a reduction in human dynamic visual acuity and uses a fact that a moving object that deviates from the gazing point is less likely to experience deterioration in image quality even if it is roughly modeled and displayed.

By using the above-described method or the like, a plurality of models of which the fineness is changed are prepared in advance for the target object, and the roughest model which does not cause the deterioration of the image quality is selected by using the parameter of the characteristic in each method. This makes it possible to reduce the number of vertices, that is, the vertex data.

Generally, a plurality of models created by changing the fineness of the same object is called "hierarchical model". In order to create this layered model, even if the number of vertices is reduced regardless of the features of the model, that is, the vertex data is deleted, there are a complex portion with a large amount of information and a portion with a small amount of information and a small amount of information. Therefore, the vertex data is uniformly deleted, so that the discontinuity in the display is perceived by the observer when switching the model hierarchy, which gives an unnatural feeling.

[0010]

As described above, the method of preparing the layered model and selecting the optimum model in consideration of the human visual angle characteristics is highly realistic by using computer graphics. It is extremely effective for displaying images in real time. However, as described above, it is not preferable to uniformly reduce the vertices, that is, the vertex data from the entire model to generate the layered model, because it gives the observer a discontinuity in the display. Therefore, it is necessary to prepare a layered model that is suitable for high-speed display and does not give the observer discontinuity in display.

The present invention has been made in order to solve the above problems, and provides a three-dimensional model data generating device capable of generating three-dimensional model data which is suitable for high-speed display and does not cause discontinuity in display. The purpose is to provide.

[0012]

A three-dimensional model data generating device according to the present invention includes means for inputting a plurality of vertex data for defining a three-dimensional model for each vertex forming a three-dimensional model, and , A triangular patch generation means for generating a plurality of triangular patches defined by three neighboring vertices among the vertices defining the three-dimensional model, and a normal vector detection means for detecting a normal vector for each triangular patch, In the integrated area, a spatial frequency change detecting means for detecting a change in spatial frequency using the detected normal vector, a triangular patch integrating means for integrating adjacent triangular patches based on the detection result of the spatial frequency change, Vertex data deleting means for deleting the vertex data of.

[0013]

In the three-dimensional model data generation device according to the present invention, the spatial frequency of the three-dimensional model is detected by using the normal vector of the triangular patch generated for the three-dimensional model. Adjacent triangular patches are integrated based on the detected change in spatial frequency, and the vertex data in the integrated region is deleted. That is, since the vertex data is deleted in consideration of the change in spatial frequency of the three-dimensional model, three-dimensional model data that does not cause discontinuity in display can be generated.

[0014]

FIG. 2 is a block diagram of a three-dimensional display system to which a layered model generating apparatus according to an embodiment of the present invention is applied. Referring to FIG. 2, the three-dimensional display system is
Hierarchical model generator 52 and hierarchical model generator 5
2, a digitizer 51 for inputting data of a three-dimensional model, a storage device 61 for storing the hierarchical model data generated by the hierarchical model generation device 52, and a three-dimensional model of the three-dimensional model by accessing the storage device 61. A display processing device 62 that performs a process for displaying and a display device 63 that displays a three-dimensional model based on the processed data are included. The storage device 61, the display processing device 62, and the display device 63 form a display unit 60.

FIG. 1 is a flow chart for explaining the processing in the hierarchical model generation device 52 shown in FIG.
Referring to FIG. 2, first, in step 101, the digitizer 51 is used to input in detail the three-dimensional coordinate data of the vertices forming the surface of the three-dimensional model to be displayed and the texture data of the model surface. Therefore, the vertex data can be obtained in detail for each vertex.

In step 102, a triangular patch defined by three neighboring vertices is generated. For example, although FIG. 10 shows a wireframe model of a Venus image, many triangular patches are generated for such a 3D model.

In step 103, a normal vector is calculated for each triangular patch. Here, it is assumed that each normal vector is represented by (x, y, z).

At step 104, adjacent triangular patches are integrated based on the detected normal vector. The triangular patch integration process can be performed using either of the two methods described below.

First, a first method for integrating triangular patches will be described. FIG. 3 is a feature space diagram of the normal vector mapped to the feature space. In FIG. 3, each x,
The y and z axes show the values of the elements of the normal vector. Figure 3
In, ◯, Δ, etc. indicate each normal vector mapped to the feature space. After the normal vectors are mapped to the feature space, neighboring normal vectors are grouped as shown in FIG. That is, groups 1 to 5 are generated for neighboring normal vectors, and group names 1 to 5 are labeled as labels for each normal vector.

The number of groups here is changed depending on the required fineness of the layered model. In other words, when a rough model is created, the number of groups is small,
That is, normal vectors that are adjacent in a wide range in the feature space are treated as belonging to one group (the same group name is given). On the contrary, when a fine model is created, the number of groups to be created is large, and therefore feature vectors that are adjacent in a narrow range in the feature space are treated as one group.

The label (corresponding to the group name) attached to each normal vector is attached to the corresponding triangular patch in the three-dimensional model space, and the following processing is further performed.

FIG. 4 is a schematic diagram showing labels attached to triangular patches in the three-dimensional model space. With reference to FIG. 4, each triangle patch is labeled with a label obtained by grouping in the feature space. After that, it is detected whether or not the same label is attached to the adjacent triangular patches, and when the same label is attached, the adjacent triangular patches are integrated. As a result of the integration processing of the example shown in FIG. 4, the integrated figure shown in FIG. 5 is obtained.

FIG. 5 is a schematic diagram showing an integrated figure obtained by integrating triangular patches. At this stage, although triangular patches have been integrated as shown in FIG. 5,
Each vertex data remains. In order to reduce the data of the hierarchical model, the vertex data existing in the integrated area is deleted, and the vertex data of the outline part of the integrated figure is also deleted to generate the data of the hierarchical model to be generated. Reduce.

FIG. 6 is a schematic diagram showing an integrated figure from which vertex data has been deleted. In FIG. 6, it is shown that the vertex data in the portion surrounded by the dotted line is deleted. The decision to delete the vertices of the contour portion of the integrated figure is made as follows.

For example, assume that the three vertices of the contour portion have a relationship as shown in FIG. Referring to FIG. 8, line segments V1V2 for vertices V1, V2 and V3
And the line segment V2V3 form an angle θ. In this case, it is determined whether or not the vertex V2 is deleted by determining whether or not the following inequality relationship is satisfied.

| Θ−180 | <THD (1) Here, THD represents a threshold for deleting a vertex. This threshold THD is determined based on the fineness required for the hierarchical model to be generated.

When the inequality (1) is satisfied, the vertex V2
Is deleted, and thus the vertex data is deleted from the integrated figure (see FIG. 6). The process of deleting the focus data in the integrated figure is the process 1 in the flowchart shown in FIG.
Equivalent to 05.

After the vertex data is deleted, new triangular patches are generated for the remaining vertices. That is, as shown in FIG. 7, when a new triangular patch is generated using the remaining vertices and further data reduction is required, that is, when coarser three-dimensional model data is created, The process returns to step 102.

In step 106, the remaining vertex data is stored in the storage device 61 shown in FIG.

Next, a second method for integrating triangular patches will be described. FIG. 9 is a schematic diagram of triangular patches in the three-dimensional model space. In FIG. 9, a total of 12 triangular patches p1 to p12 surrounding one triangular patch p0.
It is shown. The following equation is calculated with the respective angles formed by one specified triangular patch p0 and the adjacent triangular patches p1 to P12 set as φi.

[0031] Therefore, in the example shown in FIG. 9, the absolute values of the angles φ1 to φ12 formed by the specified one triangular patch p0 and the other triangular patches p1 to p12 are added according to the equation (2), and the value D is Desired. Value D is triangular patch p
Holds for 0. The value D is proportional to the spatial frequency of the shape of the three-dimensional model in the triangular patch p0. Therefore, by obtaining the spatial frequency in each part of the three-dimensional model as described above, the low spatial frequency part and the high spatial frequency part are detected, and the triangular patch integration and the vertex data deletion as described above are performed in the low spatial frequency part. Is performed.

As described above, by using the normal vector of the triangular patch forming the three-dimensional model to be displayed, the vertex data can be efficiently deleted according to the spatial frequency of the shape of the three-dimensional model. , Hierarchical model data that does not give discontinuity in display to the observer can be created. By using this layered model data, the data of the optimal layer is selected with parameters such as the distance between the viewpoint and the object, the direction difference between the gazing point and the object, and the relative movement between the observer and the display target. Thus, it becomes possible to display a high-quality image at high speed without causing the observer to feel the deterioration of the image quality.

[0033]

As described above, according to the present invention, since the vertex data deleting means for deleting the vertex data based on the spatial frequency of the shape of the three-dimensional model is provided, it is suitable for high-speed display and discontinuity in the display. Thus, a three-dimensional model data generation device capable of generating three-dimensional model data that does not cause a sex is obtained.

[Brief description of drawings]

FIG. 1 is a flowchart for explaining a process in a hierarchical model generation device shown in FIG.

FIG. 2 is a block diagram of a three-dimensional display system to which the layered model generation device according to one embodiment of the present invention is applied.

FIG. 3 is a feature space diagram of normal vectors mapped to the feature space.

FIG. 4 is a schematic diagram showing labels attached to triangular patches in a three-dimensional model space.

FIG. 5 is a schematic diagram showing an integrated figure obtained by integrating triangular patches.

FIG. 6 is a schematic diagram showing an integrated figure from which vertex data has been deleted.

FIG. 7 is a schematic diagram showing creation of a new triangular patch.

FIG. 8 is a schematic diagram for explaining a process of deleting vertices of a contour portion of an integrated figure.

FIG. 9 is a schematic diagram showing triangular patches in a three-dimensional model space.

FIG. 10 is a wire frame model diagram showing an example of a three-dimensional model.

[Explanation of symbols]

 51 Digitizer 52 Hierarchical Model Generation Device 61 Storage Device 62 Display Processing Device 63 Display Device

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Seikei Morii Kyoto Prefecture Soraku-gun Seika-cho, Osamu Osamu Osamu, Mihiratani No. 5 ATR Communications Systems Laboratories, Inc. (72) Inventor Shinji Tetsuya Kyoto Prefecture 5 Seiraya, Seika-cho, Soraku-gun, Osamu Osamu, Mihiratani Co., Ltd.

Claims (3)

[Claims] 1. A three-dimensional model data generation device for generating model data used for high-speed display of a three-dimensional model, wherein a plurality of vertices for defining a three-dimensional model are provided for each vertex forming the three-dimensional model. Means for inputting data, triangular patch generating means for generating a plurality of triangular patches defined by three neighboring vertices among the vertices defining a three-dimensional model, and a normal vector for each triangular patch Normal vector detecting means for detecting, spatial frequency change detecting means for detecting spatial frequency change using the detected normal vector, and triangular patch for integrating adjacent triangular patches based on the detection result of spatial frequency change A three-dimensional model data generation device including an integration unit and a vertex data deletion unit that deletes the vertex data in the integrated region. 2. The spatial frequency change detecting means transfers the detected normal vector to the feature space and groups neighboring normal vectors in the feature space, and a normal line of each adjacent triangular patch. A group change detecting means for detecting a change of a group to which the vector belongs, the triangular patch integrating means integrates adjacent triangular patches based on a detection result of the group change detecting means,
The three-dimensional model data generation device according to claim 1. 3. The spatial frequency change detecting means uses the detected normal vector to detect an angle formed by adjacent triangular patches, and a sum of the angles detected by the angle detecting means. The three-dimensional model data generation device according to claim 1, further comprising: an angle-sum detecting unit that detects, and the triangular patch integrating unit integrates adjacent triangular patches based on a detection result of the angle-sum detecting unit.