Method and device for calculating near plane of camera, storage medium and computer equipment

Abstract

The invention discloses a method and device for calculating a camera near plane, a storage medium and computer equipment, and relates to the field of three-dimensional rendering, and the method comprises the steps: determining an object in a scene, and dividing a bounding box; constructing a bounding ball according to the bounding box; constructing a camera space; projecting the bounding ball into a camera space; position features of the bounding ball and the camera space are judged; and a value of a plane closest to the proximity is found through sorting to determine a camera near plane. The bounding box is used for subdividing an object in an area which can be observed by a camera, the object is divided into a plurality of small objects, the bounding box is constructed into a bounding ball, the bounding ball can cover all parts of the object in the blocked object, and the problem that the object is cut by the camera at a special visual angle and a special position is avoided. Through the combination of the bounding box and the bounding ball, the depth conflict problem of objects can be greatly improved, the discrimination degree of the camera for displaying a plurality of objects in the visual field of the camera is improved, and the object on the nearest plane in the plurality of objects in the visual field of the camera is accurately displayed.

Description

Technical Field

[0001] The present invention relates to the field of three-dimensional rendering, and in particular to a method, device, storage medium and computer equipment for calculating a near plane of a camera. Background Art

[0002] In rendering, the camera is one of the most important mathematical models. Specifically, the camera (or camera space, also called view space) determines the size and position of the 3D model that is finally presented on the display device. More intuitively, the camera can be compared to the eyes in reality. The eyes look at objects in the 3D space at different positions, different orientations, and rotate in different directions (the direction of the top of the head) to determine the content that is finally imaged on the retina. Similarly, the camera also needs to clarify three basic quantities: position, orientation, and upward axis. In addition, when the camera looks at the scene, the range needs to be defined. Without considering perspective, along the camera orientation, the range seen by the camera is determined by a cuboid. The up, down, left, and right of the cuboid correspond to the up, down, left, and right of the final display device. However, when two objects are close in the horizontal direction and need to be imaged by the camera, due to the difference in front and back positions, the camera needs to display the object in front, so it is necessary to determine the camera's near plane to determine the object closer to the camera and image it for display. However, the existing camera imaging technology cannot handle this problem well, so a camera near plane determination method is urgently needed.

[0003] Chinese patent document CN102353931A discloses "a method for relative positioning of space targets". It adopts step 1: a micro-satellite carrying a visible light camera is released from an on-orbit spacecraft; step 2: the distance between them and the azimuth and pitch angles in each other's body coordinate system are measured by microwave and direction-finding equipment respectively; step 3: the on-orbit spacecraft and the micro-satellite constitute a binary observation system, jointly observe the space target, and obtain the azimuth and pitch angle of the target in their respective body coordinate systems; step 4: establish a binary inter-satellite coordinate system, and calculate the rotation matrix from the satellite body coordinate system to the binary inter-satellite coordinate system; step 5: calculate the azimuth and pitch angle of the target in the binary inter-satellite coordinate system; step 6: finally, the three-dimensional position of the space target is obtained by calculating the intersection of the binary line of sight. In step 1, the micro-satellite carrying a visible light camera is released from the on-orbit spacecraft. In step 2, the following steps are used to measure the azimuth and elevation of the on-orbit spacecraft and the micro-satellite in the coordinate system of the other party: step 2-1 detects the target from the background image through image processing, and measures its imaging coordinates on the camera focal plane; step 2-2 calculates the projection coordinates of the target relative line of sight unit vector in the camera measurement coordinate system; step 2-3 converts the target relative line of sight unit vector to the satellite body coordinate system according to the camera optical axis pointing angle measured by the photoelectric axis encoder, and calculates the azimuth and elevation of the on-orbit spacecraft and the micro-satellite in the coordinate system of the other party. The binary intersatellite coordinate system in step 4 takes the center of mass of the binary star as the origin, the direction of the binary star connecting the binary star and pointing to the other party as the X axis, the Z axis is in the plane formed by the binary star and the center of the earth, perpendicular to the X axis and pointing to the center of the earth, and the Y axis, the X axis and the Z axis form a right-handed coordinate system. The azimuth and elevation angles of the target in the binary satellite inter-satellite coordinate system in step 5 are calculated by transferring the azimuth and elevation angles of the target measured in the binary satellite body coordinate system to the binary satellite inter-satellite coordinate system through the rotation matrix calculated in step 4. However, this method cannot clearly image and display two objects that are far away from the camera and close in horizontal distance. Summary of the invention

[0004] The present invention mainly solves the technical problem that the original distance camera cannot accurately display several objects that are close in parallel distance but far away from the camera, and provides a method for calculating the camera near plane by combining a bounding sphere and a subdivided bounding box. This patent divides all objects in the camera's field of view into bounding boxes and wraps them with corresponding bounding spheres, then establishes camera space and world space, expresses the camera and the center of the bounding sphere in the form of coordinates, and finally determines the camera's near plane according to the distance between the coordinates, and finally identifies several objects that are close in the camera's field of view and accurately displays the nearest object.

[0005] The above technical problems of the present invention are mainly solved by the following technical solutions: The present invention comprises S1: Determine the objects in the scene and divide the bounding boxes; S2: construct a bounding sphere based on the bounding box; S3: construct camera space; S4: Project the bounding sphere into the camera space; S5: Determine the position characteristics of the bounding sphere and the camera space; S6: Sort and find the value of the closest plane to determine the camera near plane.

[0006] In the existing camera imaging process, the camera will collect and display objects in the scene. However, when multiple objects are approximately on the same straight line and received by the camera, if the order of the objects cannot be determined, the camera will display two objects at the same time. Since the display screen is a two-dimensional plane, multiple objects on the approximately same straight line will overlap on the screen, or multiple objects will switch back and forth and cannot be displayed normally on the screen. In order to solve this problem, this patent first establishes the camera as a camera space coordinate system, establishes the area where the object is located as a world space coordinate system, divides the objects within the camera imaging range into several small pieces through a bounding box, and then includes all the bounding boxes into a bounding sphere. The camera near plane is determined by calculating the distance from the center of the bounding sphere to the camera plane, and displays objects that are simultaneously on the camera near plane.

[0007] Preferably, in step S3, the camera space is constructed according to the position and orientation of the camera, the Z axis in the coordinate system is the camera shooting direction, the Y axis in the coordinate system is the vertical upward direction of the camera, and the camera Z axis vector n (nx, ny, nz), the camera Y axis vector s (sx, sy, sz), and the camera X axis vector t (tx, ty, tz) are set. By constructing the camera space, the plane where the camera is located and the camera orientation can be accurately determined, which is convenient for the subsequent calculation of the distance between the object and the camera, and also convenient for the subsequent determination of the near plane.

[0008] Preferably, in step S4, all bounding spheres are first included in the same world coordinate system, and then the centers of all bounding spheres are traversed and converted into the camera space to obtain two-dimensional center coordinate values.

[0009] Preferably, when converting the coordinates in S4, the original coordinates in the world space are first set as v, and the calculation method for converting the coordinates P in the camera space is p=v*n+v*s+v*t. The conversion formula can accurately calculate the specific coordinate values ​​of the bounding sphere coordinates in the world space projected into the camera space.

[0010] Preferably, in step S5, different bounding spheres are projected into the camera space to generate a circle, and the camera area is defined by the circle. By judging the relationship between the distance between the centers of the two circles and the sum of the radii, it is possible to judge whether the bounding sphere is in the camera space. By comparing the distance between the centers of the circles and the sum of the radii of the two circles, objects that are not within the enveloping range of the camera space can be filtered out, thereby improving the efficiency of determining the near plane of the camera.

[0011] Preferably, in step S6, the distances from the nearest points of all objects to the camera plane are obtained by subtracting the radius of the enclosing sphere from the distances from the center of all enclosing spheres to the plane where the camera is located, and the camera near plane closest to the camera is obtained by sorting all values. The distances from the center of all constructed enclosing spheres to the camera plane are calculated, and the minimum value can be obtained after sorting and comparison. The plane where the object at the center of the enclosing sphere corresponding to the value is located is the camera near plane. Traversing all values ​​can accurately cover all objects included in the camera space, thereby improving the accuracy of determining the camera near plane.

[0012] Preferably, in step S1, the objects to be determined in the scene are first determined, and the corresponding bounding boxes are segmented according to the actual size of the objects to generate a number of bounding boxes. Some objects are large in size. If they are directly constructed as bounding spheres, the area represented by the original objects will become larger, and the near plane value calculated for the object will be smaller, thereby affecting the final near plane result.

[0013] Preferably, in step S2, the bounding box after segmentation is constructed into a bounding sphere, and the diameter of the bounding sphere is set to the diagonal of the bounding box. If the bounding box is used to segment the object, the small block of the object segmented by the bounding box is directly used to calculate the camera near plane. If the object is at the corner or edge of the bounding box, the camera at a specific viewing angle is likely to cut off the object at this position, affecting the calculation accuracy of the final near plane. The bounding box is constructed as a bounding sphere, which can fully collect and calculate the object parts at all positions, thereby improving the calculation results and accuracy.

[0014] Preferably, a device for calculating the near plane of a camera includes a monitoring module for controlling each plane to receive rendering notifications; a triggering module for controlling rendering events triggered in sequence by each plane object of the monitored rendering notification; an acquisition module for acquiring the triggered rendering effects; and a rendering module for determining the near plane and rendering the image according to the required rendering requirements.

[0015] Preferably, a storage medium is characterized by comprising a near plane determination program, and when the near plane determination program is executed on a computer, the computer is executed according to a predetermined near plane determination method.

[0016] Preferably, a computer device comprises a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, wherein when the computer program is run, the processor executes the method of calculating the camera near plane by combining a bounding sphere and a subdivided bounding box as described in any one of claims 1 to 8.

[0017] The beneficial effects of the present invention are as follows: by using a bounding box to subdivide objects within the area that can be observed by the camera, dividing the objects into several small objects can reduce the interference of larger objects in the scene on the final calculation. Secondly, the bounding box is constructed as a bounding sphere, which can cover all parts of the object in the object after the block, and there will be no problem of objects under special viewing angles and positions being cropped by the camera. The combination of the bounding box and the bounding sphere can greatly improve the depth conflict problem of the object, improve the resolution of the camera displaying several objects that are relatively close in the horizontal direction, and accurately display the relatively close objects in the camera field of view after establishing the camera near plane. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 The present invention is a flow chart of a method for calculating a camera near plane by combining a bounding sphere and a subdivided bounding box.

[0019] Figure 2 It is a schematic diagram of a camera determining a near plane according to the present invention.

[0020] Figure 3 It is a schematic diagram of a bounding box segmenting an object according to the present invention.

[0021] Figure 4 It is a schematic diagram of the principle of generating a bounding sphere from a bounding box of the present invention.

[0022] Figure 5 It is a structural diagram of establishing a camera space and a world space of the present invention.

[0023] In the figure, a. the camera's upward direction, b. the camera's orientation, c. the camera's position, d. the near plane, and e. the far plane. DETAILED DESCRIPTION

[0024] The technical solution of the present invention is further specifically described below through embodiments and in conjunction with the accompanying drawings. Embodiment: A method of calculating the camera near plane d by combining a bounding sphere and a subdivided bounding box in this embodiment is as follows: Figure 1 As shown, it includes S1: determining objects in the scene and dividing bounding boxes; S2: construct a bounding sphere based on the bounding box; S3: construct camera space; S4: Project the bounding sphere into the camera space; S5: Determine the position characteristics of the bounding sphere and the camera space; S6: Sort and find the value of the closest plane to determine the camera near plane d.

[0025] like Figure 2 As shown in the figure, in real-time rendering, the camera is one of the most important mathematical models. Specifically, the camera (or camera space, also called view space) determines the size and position of the 3D model that is ultimately presented on the display device. More intuitively, the camera can be compared to the eyes in reality. The eyes look at objects in the 3D space at different positions, different orientations, and rotate in different directions (the orientation of the top of the head) to determine the content that is ultimately imaged on the retina. Similarly, the camera also needs to define three basic quantities: position, orientation, and upward axis. In addition, when the camera looks at the scene, it needs to define the range. Without considering perspective, along the camera orientation b, the range seen by the camera is determined by a cuboid, and the top, bottom, left, and right of the cuboid correspond to the top, bottom, left, and right of the final display device. Due to precision limitations, it is impossible for a computer to calculate the imaging of objects that are infinitely close to the camera or infinitely far from the camera. The front and back of the cuboid correspond to the values ​​that should be set for the closest and farthest distances of the camera. Since two objects may be located at the same position on the two-dimensional plane of the display device at the same time after calculation, it is necessary to decide which object should be displayed and which object should be discarded. The current hardware practice is to record the distance from different objects (strictly speaking, each pixel after the object is discretized) to the camera, and compare these distances to determine the order of the objects, and ultimately decide which object to display on the display device. The data recorded by the hardware here is a floating point number also called depth. Floating point numbers have precision limits, and different intervals have different precisions. In real-time rendering, the algorithm in the hardware will make corresponding transformations. The transformation formula is: Where F depthis the depth, z is the distance of the object from the camera plane, near is the minimum value of the distance of the object from the camera plane, and far is the maximum value of the distance of the object from the camera plane. The depth of the transformation is in the range of {0,1}. Therefore, this embodiment needs to determine the near plane d of the camera by a certain method. At present, there are two mainstream technologies for calculating the near plane d of the camera in the industry. The first is to directly set the default value: since objects within the near plane d of the camera will be clipped and lost in the camera's perspective, an intuitive idea is to directly set a smaller value for the near plane d of the camera, so that most objects can be guaranteed not to be clipped. This method is feasible to a certain extent, but due to the limitations of the above-mentioned hardware, it will bring new problems. After the near plane d is set to a smaller value, once the two objects are farther away from the camera in the vertical direction, but the distance between the two objects in the horizontal direction is closer. When performing depth comparison, the depth values ​​calculated by the two objects are similar, so when the hardware compares the depth values, it will be unable to distinguish the two objects, resulting in the pixels at the same position on the screen switching back and forth between the two objects, resulting in the problem of Z-Fighting.

[0026] The above method has problems because the near plane d is set too small, so another idea is to set a larger near plane d value. Use the bounding box of each object in the scene to perform calculations. Each object in the scene has a bounding box that defines the scope of the object, so you can traverse the objects in the scene, calculate the minimum value of the depth from the 8 vertices of the bounding box to the camera, and obtain the camera near plane d value by sorting the minimum depth of all objects, that is, the dynamic computer scene near plane d value. When the object is far away from the camera, the calculated near plane d value is relatively large, and the floating point representation interval between objects that are close to each other is larger, so the above Z-Fighting problem can be alleviated. However, since only the 8 vertices of the bounding box are used, if the camera moves to the middle of the object, no matter from which direction you look at it, the part between the camera and the vertex will be clipped, and it will not be able to be imaged in the camera, which will eventually affect the judgment of the camera near plane d.

[0027] The first trick to solving the above problem in real-world rendering is to always separate the distance between different objects and not try to place multiple objects too close together so that some of their triangles overlap. By setting an offset value between two objects that the user cannot notice, depth fighting between the two objects can be completely avoided. For example, when placing a box on the floor, you can move the box slightly horizontally so that this small change in the position of the box will be less likely to be noticed, but it can completely reduce the occurrence of depth fighting. However, this requires manual adjustment for each object and thorough testing to ensure that no object in the scene will cause depth fighting. The second trick is to set the near plane as far away as possible. It is known that the rendering accuracy is very high when it is close to the plane, so if you move the near plane away from the viewer, you will have greater accuracy for the entire frustum. However, setting the near plane too far will cause close objects to be clipped, so it usually takes experimentation and fine-tuning to determine the near plane distance that is best for your scene. The third trick is to sacrifice some rendering performance to get a higher precision depth buffer. Most depth buffers are 24-bit precision, but most graphics cards now support 32-bit depth buffers, which will greatly improve the accuracy. So at the expense of some performance, you can get a higher-precision depth test and reduce the depth. In order to solve the above problem, this embodiment adopts a method that combines the bounding sphere and the subdivided bounding box to calculate the camera near plane d. The specific implementation steps are as follows: Step 1. Because the axis-aligned bounding box of some objects in the scene may be large, enclosing and segmenting the objects in this way will cause great interference to the final calculation of the nearest point, so it is necessary to determine the volume of the object, filter out the larger objects in the required imaging scene, set different division standards for each object according to the actual size, and then divide the bounding boxes of these larger objects according to their respective standards. Figure 3 As shown, the standards can be implemented according to different business scenarios. For example, a standard is set as follows: if the length exceeds 4000, it is divided into four equal parts, then the bounding box will be divided into 4*4*4 smaller bounding boxes. If the length exceeds 2000 but is less than 4000, it is divided into two equal parts, then the bounding box will be divided into 2*2*2 smaller bounding boxes. Subdividing the object bounding box can reduce the interference of larger objects in the scene on the final calculation and improve the accuracy of the near plane d calculation. Step two, the existing bounding box technology only considers the relationship between the 8 vertices and the camera. This method will cause the camera to have inaccurate determination of the near plane d at specific angles. Therefore, after generating the bounding box, it is also necessary to construct the bounding box into a bounding sphere, such as Figure 4As shown in the figure, the center of the original segmented bounding box is used as the center of the sphere, and the longest diagonal length of the original segmented bounding box is used as the diameter of the bounding sphere. The nearest point is calculated within the range of the bounding sphere, and the bounding box is constructed as a bounding sphere to ensure that the object will not be clipped by the camera at a special viewing angle and position, so that any point on the object will not be ignored by the camera. Step 3: As shown in the figure, the center of the original segmented bounding box is used as the center of the sphere, and the longest diagonal length of the original segmented bounding box is used as the diameter of the bounding sphere. The nearest point is calculated within the range of the bounding sphere, and the bounding box is constructed as a bounding sphere to ensure that the object will not be clipped by the camera at a special viewing angle and position, so as to ensure that any point on the object will not be ignored by the camera. Figure 5 As shown, since the position, orientation and upward axis of the camera are known, the camera space can be constructed accordingly. The left side of the figure is the camera space, and the right side is the world space. After the camera space is established, the object in the world space can be converted to the camera space. When the conversion method is executed, the Z axis value of the object is the distance to the plane where the camera is located, and the X axis and Y axis values ​​are the coordinates on the camera plane. Step 4: After completing the establishment of the camera space and the world space where the object is located, traverse all bounding spheres and convert the center of the bounding sphere to the camera space to obtain the two-dimensional coordinates (XY axis values) and the distance from the center of the sphere to the plane where the camera is located (Z axis value). Step 5, after projecting the bounding sphere into the camera space, a circle on the two-dimensional plane is generated, and the area that the camera can observe can also be defined by a circle. Then the problem of determining whether the bounding sphere is in the area viewed by the camera can be simplified to determining the relationship between the two circles on the plane. The relationship between the distance between the center of the circle and the sum of the radius can be used to determine whether the bounding sphere is in the area that the camera can observe. Step 6, step 4 can collect the values ​​of the center of the remaining objects to the plane where the camera is located, and after subtracting the radius, the value of the nearest point of each object to the camera plane can be obtained. The result of step five can eliminate objects that the camera cannot see. Sort the distance values ​​between the center of all bounding spheres and the camera plane. After sorting, you can get the value of the nearest point of all objects to the camera plane, and filter out the minimum value that is not 0. This value is the near plane d of the camera. Since the calculated value may be very small, if it is less than the default value, we can directly use the default value instead in the last step. For example, the default value is 100, but the calculated value is 5, then use 100 instead. Through the combined processing of the above bounding sphere and bounding box technology, it is possible to ensure that all positions of the object are not clipped by the camera, and at the same time greatly improve the depth conflict problem of the object.

[0028] In addition, this embodiment includes a device for calculating the camera near plane by combining a bounding sphere and a subdivided bounding box, including a monitoring module for controlling each plane to receive a rendering notification; a triggering module for controlling the rendering events triggered in sequence by each plane object of the listening rendering notification; an acquisition module for obtaining the triggered rendering effect; a rendering module for determining the near plane and rendering the image according to the required rendering requirements. The rendering device is adapted to the camera near plane determination method in this example, and the application of this method in the rendering process can accurately determine the camera's near plane and improve the rendering accuracy; it also includes a computer-readable storage medium, the medium includes a near plane determination program, and when the near plane determination program is executed on a computer, the computer is executed according to the established near plane determination method; and a computer device. The device includes a memory and a processor, the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, characterized in that when the computer program is running, the processor stably executes the method for calculating the camera near plane by combining a bounding sphere and a subdivided bounding box.

Claims

1. A method for calculating a camera near plane, characterized in that: include S1: Determine the objects in the scene and divide the bounding boxes; S2: construct a bounding sphere based on the bounding box; S3: construct camera space; S4: Project the bounding sphere into the camera space; S5: Determine the position characteristics of the bounding sphere and the camera space; S6: Sort and find the value of the closest plane to determine the camera near plane.

2. A method for calculating a camera near plane according to claim 1, characterized in that: In step S3, the camera space is constructed according to the position and orientation of the camera. The Z axis in the coordinate system is the shooting direction of the camera, and the Y axis in the coordinate system is the vertical upward direction of the camera. The camera Z axis vector n(nx, ny, nz), the camera Y axis vector s(sx, sy, sz), and the camera X axis vector t(tx, ty, tz) are set.

3. A method for calculating a camera near plane according to claim 2, characterized in that: The step S4 first incorporates all bounding spheres into the same world coordinate system, and then traverses the centers of all bounding spheres and converts them into the camera space to obtain two-dimensional center coordinate values.

4. The method for calculating the near plane of a camera according to claim 3, characterized in that: When converting the coordinates in S4, the coordinates in the original world space are first set to v, and the calculation method of converting the coordinates to P in the camera space is p = v*n +v*s + v*t.

5. The method for calculating the near plane of a camera according to claim 1, characterized in that: In step S5, different bounding spheres are projected into the camera space to generate a circle, and the camera area is defined by the circle. By judging the relationship between the distance between the two circle centers and the radius, it is possible to judge whether the bounding sphere is in the camera space.

6. A method for calculating a camera near plane according to claim 4, characterized in that: In step S6, the distances from the nearest points of all objects to the camera plane are obtained by subtracting the radius of the enclosing sphere from the distances from the centers of all enclosing spheres to the plane where the camera is located. All values ​​are sorted to obtain the values ​​of the closest plane, which is the desired near plane of the camera.

7. The method for calculating a camera near plane according to claim 1, characterized in that: In step S1, the object to be determined in the scene is first determined, and a corresponding bounding box segmentation is performed according to the actual size of the object to generate a plurality of bounding boxes.

8. The method for calculating a camera near plane according to claim 1, characterized in that: In step S2, the segmented bounding box is constructed into a bounding sphere, and the diameter of the bounding sphere is set to be the diagonal of the bounding box.

9. A device for calculating a near plane of a camera applied to any one of claims 1 to 8, comprising a monitoring module for controlling each plane to receive a rendering notification; a triggering module for controlling rendering events triggered in sequence by each plane object of the monitored rendering notification; Get module, used to get the triggered rendering effect; The rendering module is used to determine the near plane and render the image according to the required rendering requirements.

10. A storage medium applied to any one of claims 1 to 8, characterized in that: The invention comprises a near plane determination program, which, when executed on a computer, enables the computer to execute according to a predetermined near plane determination method.

11. A computer device according to any one of claims 1 to 8, comprising a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, wherein: When the computer program is executed, the processor is enabled to execute the method for calculating the near plane of a camera according to any one of claims 1 to 8.

Images