KR20060131145A - Randering method of three dimension object using two dimension picture - Google Patents

Abstract

A rendering method of a three-dimensional object using a two-dimensional image is provided to represent the quality of an image like the three dimension image. When a view point of a camera is determined(S111), the position and angle of a 3-D(Dimensional) object rotated from an initial point to the camera(the next view point) are stored. An angle is calculated by the inner product of the first and second vectors(S113). The calculated angle is divided by the offset so that the real texture image number is obtained(S115). The position angle of the object is compared with the angle of the camera view point. At this time, when the position angle of the object is above the angle of the camera view point, the real texture image number is calculated by subtraction of the text image number(S117) and applied to the QBM(Quick Time VR and Billboard Mesh) so that a rendering operation is performed(S119).

Description

Rendering method of three dimension object using two dimension picture}

1 is a flowchart illustrating a method of rendering a 3D object using a 2D image according to an exemplary embodiment of the present invention.

2 is a flowchart illustrating a method of rendering a 3D object using a 2D image according to an exemplary embodiment of the present invention.

3 is a view showing an example of calculating the angle according to the coordinates of a three-dimensional object in accordance with the present invention.

4 is a view showing another example of the angle calculation according to the coordinates of a three-dimensional object in accordance with the present invention.

The present invention relates to a rendering method of a three-dimensional object using a multi-view two-dimensional image, particularly when rendering a three-dimensional object in a portable terminal.

In general, rendering is a technique of computer graphics that expresses a realistic image while considering the shape of an object in consideration of external information such as its shape, position, and light source. In other words, when creating a computer graphics (CG) image in 2D or 3D, the final step is to generate the image. In 2D CG, rendering refers to a process of generating a processed image as a final step of an image processing process for a moving image, and in the case of 3D CG, model data recorded inside a computer can be drawn on a display device. Say that to visualize

Examples of display of 3D graphics include techniques such as Quick Time VR (QTVR), which can display 3D graphics on the web, or billboards to roughly represent non-critical objects. They also produce images that look like 3D graphics.

QTVR captures an image by rotating a three-dimensional object 360 degrees at a point in time, and then displays an image corresponding to the viewpoint and angle according to the user's mouse movement on the web. It is a technique that can be seen.

When expressing such a 3D graphic, in particular, rendering a 3D object such as a leafy tree requires a lot of resources, while the image shown is not only a similar feeling, but is not the main image. The part that does not care about is used.

Rather than representing these objects in full three-dimensional form, a plate is placed to capture the image of the tree and the plate is rotated according to the user's point of view. You can feel the same way. This technique is called the billboard technique.

As such, in the prior art, when expressing a 3D graph, complicated algorithms and many calculation processes are required, and thus, hardware specifications of a personal computer (PDA) or a portable terminal (PDA (personal digital assistant), a mobile phone, a camera phone, etc.) are required to increase the speed of execution. This should be high. In addition, even if the hardware specifications are high, due to the spatial constraints of the portable terminal, there is a limit to the performance that can be supported in the portable terminal, it is not easy to implement three-dimensional graphics.

The present invention provides a method of rendering a 3D object using a 2D image generated between viewpoints.

The first object of the present invention is to render the three-dimensional model as a rectangular mesh by applying the advantages of the QTVR and the billboard when rendering the three-dimensional model, it is possible to represent the whole as a three-dimensional image with a high screen quality even at a low specification It is characterized by what is possible.

The second object of the present invention is to use a large number of meshes in the three-dimensional model, so that high speed values and three-dimensional effects can be expected even at low specifications.

A third object of the present invention is to use an image generated from multiple viewpoints, so that a visual effect corresponding to the viewpoint of a user may be expected.

The fourth object of the present invention is to provide a faster execution speed when using the original by using the image generated in the square mesh (mesh) as a texture.

Rendering method of a three-dimensional object using a two-dimensional image according to the present invention for achieving the above object,

Generating a plurality of images by respectively capturing images by rotating the three-dimensional object 360 degrees at a predetermined offset from an initial viewpoint;

Generating a rectangular mesh covering a maximum size of the captured images, and storing a position of an object on which the generated rectangular mesh is to be placed;

Thereafter, when the viewpoint of the camera looking at the specific square mesh is determined at run time, extracting the mapped image number using an angle generated between the rotated camera viewpoint and the three-dimensional object at the initial viewpoint of the object;

And rotating the rectangular mesh in a predetermined direction by an angle generated between the camera viewpoint and the 3D object, and performing rendering by applying the extracted image as a texture of the rectangular mesh.

Preferably, the angle generated between the camera viewpoint and the three-dimensional object is called a first vector between the reference point and the object and a second vector between the camera viewpoint and the object, and the angle is calculated by obtaining the inner product of the two vectors.

Preferably, the image number is extracted from the plurality of captured images by dividing an angle generated between the camera viewpoint and the 3D object by an offset.

Preferably, when the image number is extracted, one image number is extracted from two images which are symmetrical with each other by comparing the angle of the object with the angle of the camera.

Referring to the accompanying drawings, a three-dimensional object rendering method using a two-dimensional image according to an embodiment of the present invention as described above is as follows.

The present invention combines the billboard technique with QTVR to replace planar objects that rotate images to match an image in a three-dimensional space, and to look at the camera CAM. By creating and replacing them according to the situation, the rendering time is minimized and more realistic object representation is possible.

Specifically, the present invention extracts an image corresponding to 360 degrees by dividing a predetermined angle from a three-dimensional object for a two-dimensional image, and between the three-dimensional object using an image generated between the visible viewpoint and an angle generated between the viewpoint and the object. Allows you to display the image according to the angle.

1 is a diagram illustrating a preprocessing method for rendering a 3D object using a 2D image according to the present invention.

In order to implement a 3D model as a quad mesh (QBM: Quick Time VR & Billboard Mesh), a preprocessing process for applying QTVR and billboard technique is required. The texture image of the square mesh (QBM) for the QTVR application is made by rotating the three-dimensional model 360 degrees.

Accordingly, when there is a 3D object, the angle at which the 3D object is viewed is calculated. In this case, since the 3D object can be rotated 360 degrees, the angle between the image and the image according to the rotation of the 3D object is determined as an offset (S101).

As shown in FIG. 3, the image of the three-dimensional object OBJ viewed from the base point BP is the base line BL, and the image is captured by rotating 360 degrees for each offset angle. (S103). The number of texture images divided by the 360 degrees by the offset angle is generated (S104).

Here, the geometric position of the reference point BP coincides with the initial point of view ORG, the first point of view ORG is a starting point at the time of acquiring an image, and the reference point BP is a texture image number of QBM at runtime. It is the reference point of the inner product operation to calculate.

If the offset is 10, if 360 degrees are divided by 10, 36 images (0 to 35) are generated. The smaller this offset value, the more images are generated, resulting in a more natural visual effect. Here, the reference line BL is a line set to change an image, and is a line segment of a line connecting the first point of view ORG where the first image is visible and the 3D object OBJ.

When an image of 360 degrees / offset is generated, a maximum rectangle (QBM: Quick Time VR & Billboard Mesh) capable of covering all three-dimensional objects is generated (S105). Then, the position where the maximum rectangle is to be placed, that is, the position (coordinate of the object) of the 3D object OBJ is stored (S107).

Here, in order to apply the billboard technique, a rectangle that can cover the largest size among the generated images is set as the billboard. This results in all images appearing the same size by the rectangular billboard covering the largest size. In addition, in order to billboard the image, the background of the image is based on colors that are not in the image. Accordingly, the background except for the image can be transparently processed.

Here, the image size refers to the size of the target object to be viewed in the QBM, and the size may be different according to the position of the camera or other situation. For example, the size of the area occupied by the car is different from the size of the area occupied by the car accurately in the front image of the car.

The image size is a two-dimensional image acquired for use as a texture in QBM, and since the quadrilateral image includes unnecessary objects corresponding to the background along with the object to be expressed, the background part is filtered when QBM is rendered. Will be working. In order to minimize unnecessary image parts filtered as described above, processing is performed to keep the size of the acquired images the same, but to minimize them. And, according to the relative size of the processed image in the 3D space, the size of the QBM is also adjusted.

In order to render a three-dimensional object using a multi-view two-dimensional image as described above, by capturing each image by an offset angle of the three-dimensional object, and storing the object position for each image, the captured two-dimensional image image This completes the preprocessing process to render 3D.

Through this preprocessing process, the image is extracted by using the two-dimensional image and the object's position for each image, that is, the angle between the camera's point of view and the object, and then applied to the texture of the square mesh. do.

2 is a flowchart illustrating a method of rendering a 3D object using a 2D image according to the present invention. 3 and 4 are diagrams illustrating angle calculation based on a reference line BL, an initial point of view ORG, a point of view CAM, and a coordinate OBJ of a three-dimensional object for convenience of description of the present invention.

2, 3, and 4, when the viewpoint of the camera is determined (S111), the position and angle of the rotated 3D object from the first viewpoint ORG to the camera (the next viewpoint) are stored. At this time, as shown in FIG. 3, the vector value of the reference line BL and the object OBJ is set as the first vector Vector1 value, and between the objects at the camera viewpoint as the second vector Vector2, the inner product of the two vectors is obtained. Calculate the angle (S113). Here, the first vector Vector1 = BP-OBJ and the second vector Vector2 = Cam-OBJ are obtained.

Here, the angle between the reference line BL and the object OBJ is θ2, and a value obtained by subtracting 90 degrees from this θ2 is obtained as θ1. That is, since the camera viewpoint CAM maintains an angle of 90 degrees with the rotated three-dimensional object, the camera viewpoint from the first viewpoint using the angle θ2 from the reference line BL to the camera viewpoint CAM and the 90 degrees. The angle θ1 up to can be calculated. This means that if we know θ1 and θ2 through the dot product of the vector, we can get the image index and the rotation angle of QBM.

Accordingly, the actual texture image number can be known by dividing the calculated angle by an offset (S115). The image number is a texture image number of QBM and is an image number corresponding to a value calculated by dividing the inner product of two vectors by an offset.

In other words, the geometric position of the reference point BP coincides with the initial point of view ORG, where the first point of view ORG is the starting point at the time of acquiring the image, and the reference point BP is the texture image number of QBM at runtime. It is the reference point of the inner product operation to calculate. If these two points (BP, ORG) are matched, the image number can be calculated by a simple operation. That is, the internal angle between the two vectors can be known by internalizing the reference vector (BP-OBJ) and the object facing the camera (CAM-OBJ), and dividing the angle by the offset can be any one of 0 to 35. Will come out. Here, the image number has been described for the case where the offset is 10. For example, the angle between the mood vector BP-OBJ and the vector object CAM-OBJ facing the camera is 36.5 degrees, and the offset is 10. If using the ANSIC standard integer division, the image number = (INT) (36.5 / 10) = 3. That is, the third image (the fourth image) is mapped to the rotated plane object.

In order to clearly distinguish the positions where the rotation positions are symmetric with respect to each other based on 360 degrees, the position angle of the object and the angle of the camera viewpoint are compared. At this time, if the position angle of the object is greater than the angle of the camera viewpoint, the text image number is subtracted from 360 to obtain the actual texture image number (S117), and applied to QBM to render (S119).

Also, if the angle of the object is smaller than the angle of the camera, the rotated value of the square mesh (QBM) is multiplied by -1 to calculate the image number by the value located on the opposite side (= -1 * rotation value), and the square mesh (QBM) Will be applied and rendered.

Here, the images to be used as the texture can be obtained by moving the camera by an offset angle around an object. When the reference point (ORG or BP) is referred to as the front image, the left side image when the offset is 90 degrees, and the rear image when it is 180 degrees. If it is 270 degrees, it is a right side image. At runtime, the image index is obtained by determining which offset value the object's position is most similar to the current camera's position. When the offset value is found, the corresponding image is obtained. Will be mapped. Therefore, since the mapped image is different according to the positional relationship between the camera and the object at runtime, the position of the camera becomes important. In order to select the image to be mapped, the image index information according to the outlet must be stored in advance. As a reference point to measure, it also has information about the first time point or reference point.

3 and 4, when the angles obtained from the camera views CAM A are symmetrical with each other, θ1 is a rotation angle, θ2 is an angle for texture of QBM, and θ3 is an OBJ. Angle for the texture of the QBM when> CAM. This means that if 36 (0 ~ 35) images are generated by dividing the 360 degree image number by the offset 10, the moved camera viewpoint is either image number 8 or one of image number 26, so the texture image number is compared by comparing the object angle with the camera viewpoint angle. You will find.

That is, it is set as image 0 when 0 ~ 9 degrees and image 1 when 10 ~ 19 degrees based on a specific vector (ORG-OBJ) (assumes when offset = 10). To do this, the reference vector BP-OBJ and vector 2 (CAM-OBJ) are internalized, and the image number is found according to the obtained angle θ.

In operation S119, the rectangular mesh QBM is rotated in the opposite direction by the angle, and the image calculated above is applied as the texture of the rectangular mesh QBM. That is, the plane QBM continuously rotates to look at the camera, and the texture of the square mesh is applied using angle information of how much the QBM rotates to select an image to be mapped.

In this way, by calculating the angle between the reference line and the three-dimensional object and the line segment generated from the viewpoint, and by applying the image corresponding to the calculated angle, respectively, it becomes possible to render the three-dimensional object using the two-dimensional image. In addition, in order to show the generated image to the user during rendering, the billboarded image is maintained at an angle of 90 degrees with the current view.

The present invention extracts an image corresponding to 360 degrees for the two-dimensional image according to the camera viewpoint (viewing viewpoint), and shows the extracted image using the angle between the camera viewpoint and the three-dimensional object. This can improve the speed by up to 70% by rendering a rectangular mesh (QBM) compared to rendering the entire three-dimensional model. You can also view better images than using the original model for image quality.

The present invention can convey a sense of fact to a user when implementing 3D graphics such as a game or an avatar in a PC, PDA, or portable terminal having a low specification. In addition, in limited specifications such as a mobile phone terminal, it is possible to reduce unnecessary resource waste by using minimum resources.

So far, the present invention has been described with reference to the preferred embodiments, and those skilled in the art to which the present invention pertains to the detailed description of the present invention and other forms of embodiments within the essential technical scope of the present invention. Could be implemented. Here, the essential technical scope of the present invention is shown in the claims, and all differences within the equivalent range will be construed as being included in the present invention.

According to the method of rendering a 3D object using a 2D image, when rendering a 3D object, a 3D mesh is generated using information obtained by using an image generated between a camera viewpoint and an angle generated between the camera viewpoint and the object. Instead, rendering has the effect of rendering a three-dimensional object at a higher speed.

In addition, by using a multi-dimensional 2D image as a rectangular mesh texture, a 3D model can be rendered instead of a 3D model to obtain a better image than the original and can support high speed.

The present invention can see the effect that can be felt by the user when realizing a three-dimensional graphics, such as a game or avatar in a low-spec PC, PDA, portable terminal.

In addition, in a limited specification such as a mobile phone terminal, there is an effect of preventing unnecessary resource waste by using a minimum of resources.

Claims (4)

Generating a plurality of images by respectively capturing images by rotating the three-dimensional object 360 degrees at a predetermined offset from an initial viewpoint; Generating a rectangular mesh covering a maximum size of the captured images, and storing a position of an object on which the generated rectangular mesh is to be placed; Thereafter, when the viewpoint of the camera looking at the specific square mesh is determined at run time, extracting the mapped image number using an angle generated between the rotated camera viewpoint and the three-dimensional object at the initial viewpoint of the object; Rotating a rectangular mesh in a predetermined direction by an angle generated between the camera viewpoint and the three-dimensional object, and performing the rendering by applying the extracted image as a texture of the rectangular mesh, the three-dimensional object using a two-dimensional image Method of rendering. The method of claim 1, The angle generated between the camera viewpoint and the three-dimensional object is called a first vector between the reference point and the object and a second vector between the camera viewpoint and the object, and the angle is calculated by calculating the inner product of the two vectors. How to render three-dimensional objects. The method of claim 1, wherein an image number is extracted from the plurality of captured images by dividing an angle generated between the camera viewpoint and the 3D object by an offset. The method of claim 1, When the image number is extracted, a method of rendering a three-dimensional object using a two-dimensional image, characterized in that by comparing the angle of the object and the angle of the camera to extract one image number from the two symmetric images.

Images