JP2009146368A - Visible collision detection circuit - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a collision detection circuit for detecting a collision between objects in CG, which is mounted at renderer side such that the collision can be detected by hardware, thereby reducing a load of application processor and increasing a drawing speed thereof. <P>SOLUTION: The means for detecting the collision between the objects includes: a means for defining a hexahedron which surrounds an object video display space, for projecting a stillness object, which is located in the space and viewed from each outer surface of the hexahedron, on each surface of the hexahedron, and for generating and storing an presence/absence flag, an identifier, a value Z, and a normal line of the projected object for each surface of the hexahedron; a means to be used in a rendering phase for converting a rendering point to a coordinate system of each surface of hexahedron when a moving object is interpolated, for reading the flag, the identifier, the value Z, and the normal line of the stillness object for each surface of the hexahedron, and for determining the overlap of the interpolation points, the identifier of the stillness object at that point, and the collision of the interpolation points from the value Z; and a means for outputting information to the application processor, which is used to obtain the movement of a colliding object after the collision on the basis of a direction and a normal line of a moving object, and a normal line of a collided object. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a means for detecting a collision between objects in computer graphics technology, and a circuit implementation for realizing the means.

  Many computer graphics applications have produced animations involving collisions between objects. Conventionally, collision detection has been performed at high speed based on a boundary box or object tree structure method by programming on the application processor (hereinafter referred to as AP) side. In an animation such as a game that requires real-time drawing, the calculation load of these collision detections is extremely large. For this reason, in order to obtain the realism and real-time drawing of CG video, an increase in the drawing amount and complex system processing such as a physical model are essential, and the load on the AP will become more serious in the future. As long as accurate collision detection is supported by the software on the AP side, there has been a problem that has to solve the conflicting problem between speedup and data volume reduction.

  In the present invention, the collision between objects is not detected using a program installed in the AP, but the collision is detected by hardware on the renderer side called a pixel shader or a rasterizer. By eliminating collision detection processing from the AP side, the AP only outputs the same object primitives (polygons) as the general drawing to the renderer, and reduces the load on the AP and speeds up the drawing by performing collision detection by hardware. Is the subject of the present invention.

  The renderer is usually implemented with hardware such as polygon drawing, texture mapping, environment mapping, and a shader for illumination. The present invention introduces a collision detection circuit between objects with a structure assimilated with the existing function group included in these graphics processors.

  Whether the object is visually colliding can be recognized by looking at the object from a plurality of viewing angles if there is no shielding object at the collision location. As this algorithm, a hexahedron (either a cube or a cuboid, hereinafter referred to as Cube) surrounding the entire drawing space is set, and a moving object (hereinafter referred to as a moving object) and another object (hereinafter referred to as a moving object) are viewed from inside these Cubes. Collisions can be recognized by examining the possibility of contact with a still body. For this reason, the entire drawing space is virtually surrounded by Cube, and images of the inside projected from the outside of the Cube are mapped onto each surface of the Cube, and mapping maps for these six planes are stored as Cube patterns. Unless the collision point is in a completely shielded space, it can be regarded as a collision if the moving body and the stationary body overlap each other when viewed from a plurality of surfaces. This is the same process in which humans visually recognize object collisions.

  In the present invention, as a means for projecting an image inside the Cube on each of the six surfaces of the Cube, a Cube internal image is drawn on each of the six surfaces by parallel projection from a coordinate axis orthogonal to each of the six surfaces. If the projection center is the inside and outside of the Cube, or the difference between perspective projection and parallel projection is excluded, this method is similar to the environment Cube generation method of the environment mapping technique created in advance before rendering.

  The Cube pattern must be regenerated each time a moving object occurs as in the environment mapping. This is a drawing for 6 faces, and the load is large. In the present invention, in order to avoid this, all drawing objects are divided into a moving object and a static object, and only the static object is created in advance as a pattern on each of the Cube 6 surfaces. In the image display stage, the static object is rendered first, and then the moving object is rendered. At the stage of interpolating the polygon of the moving object, whether or not the interpolation point intersects with the static object is determined by comparing with the patterns on the Cube 6 surfaces (hereinafter referred to as image information). For this determination, the interpolation point of the moving object is converted into the coordinate system of each of the six surfaces, and a test is performed to determine whether or not the surface image and the interpolation point overlap.

  If there is no shielding object, only the xy coordinate overlap test between the Cube image information and the interpolation point is performed, and the information stored as the image information does not need color information such as environment mapping. The information stored is only a flag indicating the presence or absence of a 1-bit image per pixel and a still body identifier. When a collision is detected, the values returned to the AP are the identifiers of the moving object and the stationary object, and the three-dimensional coordinate value of the collision point. Thereby, the collision object name and position can be determined on the AP side. By separating the moving body and the stationary body, writing of the stationary body, which occupies the majority, to the Cube is performed only once.

  When there are a plurality of moving objects, it is also necessary to detect a collision between the moving objects. In the present invention, when there are a plurality of moving objects, a six-surface Cube dedicated to the moving object is separately prepared and drawn on the Cube in parallel with the collision detection process at the time of interpolation. By testing each of the stationary body and the moving body Cube, redrawing of the stationary body becomes unnecessary.

  When defining a moving object Cube, whether a plurality of moving objects are collectively written in a set of Cube or a dedicated Cube is created for each moving object is related to the number of moving objects or the hardware cost. If there are a large number of moving bodies, and there are those that move temporarily and those that move constantly, efficiency can be achieved by creating a Cube for each object. In this case, Cube for a moving body may be a space size surrounding the moving body.

  It is a condition for speeding up the present invention that static body Cube image information is obtained by preprocessing. In the display (rendering) stage, the drawing order of either a static object or a moving object may be used. However, the dynamic object is drawn after the static object is drawn, and the speed can be increased by performing the collision detection process only at the time of the dynamic object drawing. In real-time animation, one frame is drawn within 1/30 seconds or 1/60 seconds, and it is only necessary to be able to detect within a frame. In the present invention, during video rendering, a collision is detected as one of interpolation processes, and this is returned to the AP (after the collision occurs), the AP determines the next video with a delay of one frame. It will be.

  The coordinate system at the time of rendering is a viewpoint coordinate system having an arbitrary visual field direction. In the overlap test with the image information of the Cube 6 surface of the interpolation point, the interpolation point coordinate value is converted into the coordinate system of each Cube surface. When the viewpoint direction is an arbitrary angle, three-dimensional rotation is required. In order to avoid the load of rotation calculation for each interpolation point, in the present invention, Cube is defined by world coordinates, and the six planes of Cube are parallel to the xy, yz, and xz planes, respectively, thereby reducing the rotation calculation.

  Also, in the present invention, a world coordinate value is added to a polygon vertex of a moving object together with a world coordinate value, and this is rendered at the same time, and the viewpoint coordinate value is used for drawing, while the world coordinate value is used for collision detection with a Cube coordinate. Therefore, the rotation calculation process for each surface is omitted and the replacement process is used.

  When there is no shielding object between the viewpoints when viewed from any of the six surfaces, a collision can be detected as long as they intersect at or above the Cube3 surfaces that are orthogonal to each other. If there is a shield, additional information is required. In the present invention, as means for detecting a collision between complex shapes, the Z value (distance from the Cube surface) of the still body, the identifier of the still body, and the flag indicating the presence or absence of the still body image are stored as Cube image information. In the collision determination, when the interpolation point is overlapped with the still body having the same identifier on all the Cube surfaces, the collision is made with the still body. In addition, a moving body overlaps with a stationary body having the same identifier between the faces facing each other, and both of them are positioned behind the stationary body in the Z value comparison, and can be regarded as a collision. Depending on the Cube surface, a plurality of objects may block, and the identifier of the Cube surface may be different at the interpolation point. In this case, the Z value of the moving body is compared with the Z value of the overlapping still body. When the moving body is located on the Cube surface (front side) side with respect to the stationary body, it is assumed that there is no collision with the stationary body. On the other hand, the moving object also overlaps in the distance of the object having the same identifier (this is referred to as identifier A) in the other Cube4 planes, each of which is orthogonal to the Cube plane located inside (distant) of the stationary object. If there is a collision. Furthermore, when the same identifier A on a large number of surfaces overlaps in the distance but an object with a different identifier (hereinafter referred to as a shield) overlaps on either of the Cube surfaces, each Z value of the moving object and the shield If the Z value of the moving object is farther from the Cube surface than the shield, it is assumed that it collides with the object of the identifier A, and does not collide when it is in front.

  Reflection of moving object drawing is performed in units of scanning frames for video display. When a moving body collides with a stationary body (a moving body that has finished drawing is also defined as a stationary body), a part of the moving body intersects the stationary body when viewed from the Cube surface. When a moving object moves at a very high speed and passes through a still body within 1/60 seconds, it is usually impossible to detect a collision. However, if each object is in an intersecting state, interpolation of the moving object and the Z value of the moving object is performed. If there is a difference for each pixel in time, the difference code changes from behind to the front or vice versa at a certain point. This change can be caught and regarded as a collision. For this purpose, the Z value and identifier of the stationary body are read out at the interpolation stage of the moving object, and the interpolation interpolation Z value and the read Z value are respectively compared.

  In [008] as return information to the AP by collision detection, the renderer returns the object identifier and collision position of each collision. On the other hand, when the rebound is caused by the collision or the stationary body is moved by the collision, the attributes of the object such as the moving direction of the moving body, the surface normal of the collision point of each of the moving body and the stationary body, and mass and elasticity are required. The speed and attributes of an object are owned by an identifier, and can be judged at high speed on the controlling AP side. However, the direction of movement and surface normal at the time of collision can be determined at the renderer side or at an intermediate processing stage in the graphics engine. There is a case where it is generated with a series change, and if it is to be obtained on the AP side, geometrical development of the shape model is required from the collision point. This load is extremely large. In the present invention, in addition to the identifier and the collision position, the moving body and stationary surface normals of the collision point and the moving direction vector of the moving body are returned to the AP.

  As the means, the surface normal of the moving object is included in the interpolation processing as polygon vertex data or normal map in a normal system, and the surface normal at the collision point can be obtained at the collision detection stage. Based on this, the next coordinate value of the moving object is determined on the engine side, and if the motion vector is output to the renderer as updated information for each frame cycle, this is sent from the renderer side to the AP. Can return. On the other hand, the surface normal of the collision point of the stationary body is additionally stored as Cube surface information in which the stationary body is stored in advance. As described above, the present invention has means for storing a flag, an identifier, a z-coordinate value, and a surface normal as Cube image information.

  Since the surface normal of the stationary body is necessary only at the time of collision, a two-dimensional memory structure is not necessary for high-speed reading such as Cube image information, and a low-speed one-dimensional buffer may be used. Further, with respect to LOD (Level of Detail), there is a problem as to whether or not a collision point needs to be detected strictly in units of pixels. In the present invention, considering the optimization of the resolution versus cost corresponding to the application, the memory cost of the Cube surface information can be reduced by setting the pattern size to about 1/4 to 1/16 of the display resolution as the Cube surface still body information. Significant reduction is possible.

  As described above, the present invention defines a hexahedron surrounding the image display space to be detected as means for detecting a collision between objects, and six stationary objects inside the space as viewed from outside the respective surfaces. Means for generating / storing the presence / absence flag, identifier, Z value, and normal of the projected object as image information for each of the six planes before rendering, and in the rendering stage for displaying video, At the time of interpolation, the interpolation point is converted to the coordinate system of each of the six planes, the flag, identifier, Z value, and normal line of the still body in the six planes are read, , Means for determining the collision of the interpolation point from the identifier or Z value of the stationary object at that point, the direction and normal of the moving object, and the respective normals of the colliding object Motion after collision The visualization collision detection circuit comprising a respective means for outputting order information to the AP.

  According to the present invention, collision between objects in computer graphics is detected not by the software on the application processor side but by the renderer hardware or embedded software, thereby reducing the computational load on the application processor and increasing the drawing speed. This is effective for a real-time drawing system.

  The circuit of the present invention is mounted on a graphics LSI or implemented as an IP (Intelligent Property).

  Examples of the present invention will be described below. FIG. 1 shows a Cube according to the present invention. In FIG. 1A, the six surfaces of Cube are defined as a front surface, a rear surface, a right surface, a left surface, an upper surface, and a lower surface, respectively. The image projected on the Cube surface is an image of an internal object viewed from the outside in the direction indicated by the arrow from the outside of each surface, and stores the image information. This image is a parallel projection perpendicular to the surface, and the pixel of the object closest to each surface is drawn by hidden surface removal. Moreover, when each surface of Cube is developed on a plane, the relationship shown in FIG. 1B is obtained. In the present invention, image information is not color information but a projected object presence / absence flag, an identifier, a Z value, and a normal line.

FIG. 2 shows the positional relationship (6 plane view) on the Cube surface 20 between a stationary object and a moving object according to the present invention. In FIG. 2A, the gray sphere 21 indicates a static body already stored in the Cube, and the black small sphere 22 indicates a moving body. The sphere 22 is not a figure drawn on the Cube surface, but indicates an interpolation point position at the time of rendering. The white small sphere on the top surface of the Cube indicates that the sphere 22 is located behind the object (21 in the figure). In FIG. 2A, there is no object other than the stationary object 21 and the moving object 22. In FIG. 2A, the stationary object 21 and a part of the moving body 22 overlap on all surfaces. The overlap means an overlap on the xy coordinate axes. The collision detection procedure of the present invention is as follows.
For each interpolation of the moving object 22, the interpolation point is converted into the coordinate system of the Cube 6 plane, and a flag and an identifier indicating the presence / absence of a stationary body are read from each plane. It is assumed that the moving body 22 collides with the stationary body 21 if the identifiers of all the faces are the same and the moving bodies 22 overlap on all the faces. When the interpolation point is given in the world coordinate system, the coordinate conversion of the Cube 6 plane is simplified.

One surface, for example, four surfaces orthogonal to the lower surface of the Cube, that is, all surfaces except the upper surface, can be regarded as a collision when the interpolation point of the object 22 overlaps with a stationary body having the same identifier.
On the other hand, the Z value of the interpolation point of the moving object is compared with the Z value of the stationary object for each facing surface (any one of the upper and lower surfaces, the left and right surfaces, and the front and rear surfaces). When the interpolation point of the object 22 is located behind the stationary body 21 having the same identifier on both the faces facing each other, the interpolation point can be regarded as a point inside the object 21 and is determined to be in a collision state. The
FIG. 2B shows another relationship diagram between the object 22 and the object 21. As one of the collision determination methods, a method that does not use the Z value does not correspond to a collision. In the method using the Z value, the object 22 is not positioned far from the object 21 in the comparison result of the Z values facing each other in FIG. For example, the object 22 is located behind the object 21 on the rear surface, but the object 22 is located in front of the object 21 on the front surface facing the rear surface, and does not satisfy the condition. In other aspects, the object 21 and the object 22 do not overlap each other. As a result, it can be determined that the moving body 22 has not collided.

FIG. 3 shows the relationship when there is a shielding object according to the present invention. In FIG. 3A, the still body stored in the Cube surface 30 is a gray sphere 31 and a triangular planar body 33, and the moving body (black circle) is 32. Similarly to the above, the figure of the moving body 32 is not stored in the Cube surface but indicates a positional relationship with the static body. In FIG. 3A, if each object is examined on the basis of the Cube front surface, the triangular body 33 is at the forefront, the moving object 32 is behind it, and the sphere 31 is further behind. The moving body 32 overlaps with the stationary body 33. When the other four surfaces orthogonal to the front surface are examined, there is no overlap between the triangular body 33 and the object having the same identifier on the upper and lower surfaces and the left and right surfaces. Therefore, it is determined that there is no collision with the stationary body 33. In addition, the moving body 32 overlaps the triangular body 33 on the rear surface located on the opposite side. However, here, as a result of the Z value comparison, it is located in front of the triangular body 33 and does not satisfy the collision condition.
Further, on the rear surface, the moving body 32 has a point that overlaps with the stationary body 31, and the identifier at this point is 31. The four surfaces orthogonal to the rear surface are left and right, and upper and lower, but there is no surface that overlaps with the same identifier as 31. Therefore, the moving body 32 is determined to have no colliding object.

  On the other hand, in FIG. 3B, as in FIG. 3A, the moving body 32 is the front surface of the Cube and overlaps with the stationary body 33, and the moving body 32 is located behind the 33. The identifier to be read is a triangle 33. Next, when the four surfaces orthogonal to this surface are tested, no other overlap with the object having the identifier of the triangle 33 is detected. On the other hand, the Cube rear surface overlaps with the stationary body 31, and four surfaces orthogonal to the Cube rear surface are tested with respect to the stationary body 31. It can be determined that the interpolation points of the moving object 32 all overlap with the objects 31 on these four surfaces, and the moving body 32 can be determined to collide with the stationary body 31. This can also be detected by a test on the left and right or upper and lower Cube facing surfaces (each surface has the same identifier and the moving object is located behind).

  FIG. 4 shows a Cube surface 40 when surrounded by a shielding object according to the present invention. The shielding object 41 has a cubic shape without a bottom surface and a front surface. The shielding object 41 and the gray sphere 42 are static bodies, and the moving body 43 is interpolated. The moving body 43 overlaps the stationary body 42 on the front surface. Examining the four surfaces orthogonal to the front surface, the moving body 43 is an upper surface, and the left and right surfaces are overlapped with the still body 41 having an identifier different from that of the Cube front surface. Become. However, the moving body 43 is entirely behind the stationary body 41 that overlaps on the upper surface and the left and right surfaces. By comparison with the Z value, there are two or more objects with the same identifier that overlap with the moving object, the position is behind the object of that identifier (42 in the figure), and the object with other identifiers In the overlap with (41 in the figure), if they are all located behind, it is considered a collision. On the other hand, when the test is performed based on the rear surface of the Cube due to the relationship between the stationary body 41 and the moving body 43, the moving body 43 is on the rear surface of the stationary body 41, with respect to the front surface of the Cube facing it or the four surfaces orthogonal to the rear surface of the Cube. When the Z-value test is performed, the moving body 43 is in front of the stationary body 41 either on the front surface or the lower surface, and does not satisfy the collision condition. If there is a Z value of a moving object in front of a stationary object having at least one identifier, it is not regarded as a collision with an object having that identifier.

  In the present invention, all objects that overlap the moving object have the same identifier, and if all the moving objects are located behind them, that is, if the moving object is located inside a stationary object, it is regarded as a collision. Therefore, when the inside is a cavity and the moving body is floating inside, it is regarded as a collision even if the collision does not occur inside. Such a geometric relationship cannot be handled as long as the visualization method is used. On the other hand, in FIG. 4, if the stationary body 41 also has a bottom surface, the moving body on the lower surface of the Cube is behind the stationary body 41 and the black circle of the moving body 43 is a white circle. Therefore, the surface information indicating the relationship between the stationary body 42 and the moving body 43 is only the Cube front surface. Here, in the process of interpolating the moving object 43, there are some pixels where the Z value of the moving object 43 is located at the closest position (position where it can be visualized), and some pixels move behind the object 42. If it becomes (or vice versa), it means that a part of the moving body 43 has slipped into the stationary body 42. As a result, the objects 42 and 43 can be regarded as collisions.

FIG. 5 shows a collision detection system diagram of the present invention. The projection image information of the stationary object viewed from each surface of the Cube is stored in advance in the memory 52 at 0-5 as a two-dimensional pattern. These pieces of information are the identifier ID0-5, Z value Zd0-5, surface normal Nd0-5, and stationary object presence / absence flag fd0-5 of the object closest to the surface. However, the surface normal may not be a two-dimensional pattern.
An object is usually composed of a large number of polygons. Polygons are converted into data in a vertex coordinate value sequence. In order to draw an image of an object from this polygon row, first, a polygon vertex row is given to the interpolation circuit 50 for painting the inside of the polygon. The interpolation circuit 50 sequentially outputs the interpolation point coordinate values Xs, Ys, Zs and the surface normal inside the polygon, and IDs is output once for each object or polygon. This coordinate value is obtained in the viewpoint coordinate system, and this position is coordinate-converted by the coordinate conversion circuit 51 to each of the Cube 6 plane coordinate systems. The circuit 51 outputs the XY addresses Xd0-5 and Yd0-5 of the six surfaces of the memory 52. With these addresses, the identifier IDd0-5, Z value Zd0-5, surface normal Nd0-5, and stationary object presence / absence flag fd0-5 stored in the memory are read out from the memory 52. If the vertex coordinate values of the polygon are defined not only in the viewpoint coordinate system but also in the world coordinate system and both are interpolated simultaneously by the circuit 50, the coordinate conversion circuit 51 makes the Cube 6 plane coordinate system conversion extremely simple. These pieces of information are then applied to the collision detection circuit 53. The circuit 53 is also given a Z value Z0-5 corresponding to each surface of the Cube of the interpolation point obtained by the coordinate conversion circuit 51. The collision detection circuit performs collision detection processing corresponding to the conditions from [023] to [028] based on these pieces of information.

  FIG. 6 shows the internal configuration of the collision detection circuit 53 of FIG. The presence / absence flag fd0-5 and the identifier among the six sets of identifiers IDd0-5, Z value Zd0-5, surface normal Nd0-5, and stationary object presence / absence flag fd0-5 read from the memory 52 of FIG. IDd 0-5 is added to the presence / absence flag / identifier selection circuit 60. Here, the identifiers IDd0-5 of the surface 0-5 of the memory 52 in which the presence / absence flag is set, that is, the still body is present are selected. In the circuit 60, a register corresponding to each of the 0-5 planes is provided, and an identifier and a valid bit are stored therein. If there is no presence / absence flag, an invalid bit is set in the register. Among the coordinate values converted from the interpolation point to the six-plane coordinates in FIG. 5, the Z value Z0-5 and the Z value Zd0-5 read from the memory 52 are given to the Z value comparison circuit 61. Here, the difference codes for whether Z0-5 is in front of or behind Zd0-5 are obtained, and the result is output to the collision ID detection circuit 62. In the circuit 62, only the identifiers having valid bits are selected based on the result of the selection circuit 60, and the presence / absence of the collision is determined between the same identifiers and the Z value comparison result from the circuit 61. The determination is made according to the respective conditions of [024] to [028]. This determination can be easily made by means such as adding the surface number having the same identifier and the Z value comparison code corresponding to the surface to the determination table. That is, as the maximum possible combination, the plane information is 6 bits (one of which is valid and has the same identifier is 1 bit), and the Z value comparison result is 6 bits. It becomes. The table may be a capacity with 4096 addresses and an output indicating a 1-bit collision. With this output, the identifier under test is output from the circuit 62. At the same time, the multiplexer circuit 63 receives one of the surface numbers having the identifier under test from the circuit 62, and determines the surface normal Nd of the predetermined surface from the surface normals of the collision points of the six types of stationary objects on the respective surfaces. select. The coordinate value XYZ of the interpolation point, the surface normal Ns, and the object motion direction vector Ms are output through the multiplexer circuit. In the circuit 62, when there are a plurality of identifiers, it can be dealt with by sequentially scanning six surfaces for each identifier, detecting the same identifier, setting a flag for that surface, and adding it to the table in the same manner as described above. .

  On the other hand, there is a case where collision cannot be detected only from the identifier and the Z value determination. In the condition of [015], information (pixel difference result) of one pixel before interpolating a moving object is stored using a buffer corresponding to each of the six surfaces, and the Z value is the same identifier. Detects when the top changes from front to back. The history circuit 64 in FIG. 6 stores a stationary object presence / absence flag, an identifier, and Z difference result information from a circuit 62 that receives the stationary object presence / absence flag, Z value difference code, and identifier for each of the six surfaces for each interpolation. The information is interpolated and input to the circuit 62, and the information sent to the circuit 64 is compared. When the pixel point of the moving object moves from the front of the same stationary object to the inside, a signal is sent to the circuit 62 as a collision. The identifier, normal, etc. described in [029] at the time are output.

On the other hand, a stationary object is stored in advance as image information before drawing for display. However, when there are a plurality of moving objects, a six-surface Cube similar to the stationary object is separately created at the display drawing stage, Is projected onto each surface and stored. Therefore, in the present invention, the six-surface Cube image information of the moving object is refreshed and redrawn for each frame, unlike the information of the stationary object. As a circuit, a temporary memory is added to the memory 52 of FIG. 5, and the output signal to the collision detection circuit also increases. However, the collision detection method is the same as the collision with a stationary object. If a collision is detected on both sides, predetermined information is output.
With each of the above means, the present invention is a method suitable for hardware detection of collision within a range in which collision between objects can be visually detected from six surfaces and collision can be detected.

  The circuit of the present invention is provided as an IP (Intelligent property) or mounted on a graphics processor LSI so that it can be used for CG video production and an amusement system.

“The six-sided cube A according to the present invention and its plan development view B are shown.” “The geometrical relationship between collision A and non-collision B between a stationary object and a moving object according to the present invention is shown.” “The geometrical relationship between non-collision A and collision B between a plurality of stationary objects according to the present invention is shown.” “Shows the geometrical relationship of the collision within the shielding object according to the present invention.” “The circuit configuration of the collision detection system of the present invention is shown.” “A collision detection circuit of the present invention is shown.”

Explanation of symbols

FIG.
10 Cube 11 Cube plane development figure 2
20 Cube plane development view 21 Static spherical object 22 Motion figure 3
30 Cube plane development 31 Stationary spherical object 32 Moving body 33 Stationary triangular object diagram 4
40 Cube plane development view 41 Shielding object 42 Static spherical object 43 Moving object diagram 5
50 Interpolation Circuit 51 Coordinate Conversion Circuit 52 Cube 6-Plane Image Information Memory 53 Collision Detection Circuit FIG.
60 Valid flag / identifier selection circuit 61 Z value comparison circuit 62 Collision identifier detection circuit 63 Multiplexer 64 History circuit

Claims (5)

  The present invention relates to a collision detection circuit for a display object in a computer graphics system, wherein a virtual hexahedron surrounding a graphic display space is defined, and a stationary object included in the display space is viewed from the outside of the hexahedron as a viewpoint. The projected image information includes a stationary object presence / absence flag, an identifier, a surface normal, and a distance from the surface, each of which is stored in advance in memory, and a viewpoint coordinate for video display. In the stage of drawing all the objects in the system, the means for drawing in the order of moving objects from the stationary object, and the drawing of the moving objects, the coordinate values of the objects from the viewpoint coordinate system to the respective hexahedrons are drawn. By converting to a plane coordinate system, each of the previously stored image information is read from the memory, and the overlapping state of the moving object and the stationary object is detected on a plurality of faces of the hexahedron. To, collision detection circuit for determining the presence or absence of a collision with an object and a stationary object moving based on the overlap condition between the objects.   2. The circuit according to claim 1, wherein the hexahedron is defined in the world coordinate system, the polygon vertex sequence of the moving object is defined by both the viewpoint coordinate value and the world coordinate value, and the viewpoint coordinate value is determined by interpolation from the polygon vertex sequence. A collision detection circuit having means for interpolating the world coordinate values together and simplifying the conversion process of the hexahedron to each plane coordinate system using the world coordinate values of the interpolation point.   In the circuit of claim 1, when there are a plurality of moving objects, a plurality of hexahedral memories surrounding the moving object are provided in addition to a memory storing a stationary object, and each of the six surfaces is viewed from the outside. Means for storing image information of a moving object to be projected in each of the plurality of sets in the interpolation step, and the image information of the moving object from an object presence flag, an identifier, a surface normal, and a surface The distance information, the pre-stored image information of the stationary object of claim 1 and the image information of the moving object are read simultaneously, and the stationary object and the moving object are read based on the respective information. A circuit that detects collisions.   4. The circuit according to claim 1, wherein a polygon identifier is held in a polygon drawing stage of a moving object, and a surface normal is obtained together with a coordinate value for each interpolation point in a polygon display stage to detect a collision. In this case, the application processor outputs the collision detection point coordinate value, the identifier and surface normal read from the stationary object image information, and the identifier and surface normal held by the moving object, respectively, A collision detection circuit that informs the application processor of each identifier, collision point coordinate value, and surface normal of the collision surface.   A computer graphic image apparatus using each of the means according to claim 1.

Images