Three-dimensional net capture visualization method based on simulation data
By determining the optimal bounding box and fitting the mesh nodes through pre-calculated finite element physical simulation data, and combining it with a lightweight real-time fitting process, the problems of low computational efficiency and low rendering efficiency of rope net capture targets in Web 3D scenes are solved, and a highly realistic adaptive mesh wrapping effect is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHONGKE XINGTU MEASUREMENT & CONTROL TECH CO LTD
- Filing Date
- 2025-12-24
- Publication Date
- 2026-05-19
AI Technical Summary
When simulating and rendering rope net capture targets in real time in a Web 3D scene, there are problems of low computational efficiency and frame rate drop, making it difficult to achieve both realistic net movement and rendering efficiency in complex business scenarios on the browser side.
By pre-calculating finite element physical simulation data, the optimal bounding box is determined and the mesh nodes are fitted. Combined with a lightweight real-time fitting process, the mesh model is driven to adaptively fit the target shape, avoiding real-time recalculation.
It achieves a highly realistic adaptive wrapping effect where the net's motion conforms to physical laws, while improving rendering efficiency on the browser side and avoiding heavy physics calculations.
Smart Images
Figure CN122066889A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of computer graphics technology, specifically to a three-dimensional net capture visualization method based on simulation data. Background Technology
[0002] Currently, simulating and rendering the effect of net capture targets in real time in web 3D scenes faces significant challenges. As a complex and flexible physical model, performing full real-time finite element physical calculations on the browser side incurs enormous performance overhead, leading to low computational efficiency, a severe drop in frame rate, and making it difficult to apply to complex business scenarios. To balance the physical realism of net motion with rendering efficiency on the web, a method is urgently needed that can drive the net model to achieve realistic net capture visualization by adapting to targets, without requiring real-time recalculation, using pre-calculated physical law data. Summary of the Invention
[0003] Based on the above analysis, this invention provides a three-dimensional net-capturing visualization method based on simulation data, applicable to nets used to capture targets, comprising the following steps:
[0004] S1. Confirm the optimal bounding box of the target to be captured: Calculate the bounding box of the target to be captured, and generate N sampling points by uniformly sampling the surface of the bounding box. and from the sampling point Collision detection is performed by emitting rays, and the distance between the intersection point of each ray and the surface of the bounding box and the intersection point of each ray and the surface of the target to be captured is calculated. The bounding box type with the smallest average distance of all rays is selected as the optimal bounding box.
[0005] S2. After determining the optimal bounding box, fit the mesh nodes of the mesh to the surface of the optimal bounding box one-to-one and output a mapping dictionary M to complete the fitting.
[0006] S3. After the fitting is completed, the motion trajectory of the net nodes is updated in the time dimension to realize the visual transition from free physical motion to the state of fitting the target and finally complete the visualization of net capture.
[0007] Preferably, the bounding box in S1 includes an OBB (oriented bounding box) and an AABB (axial bounding box), wherein for the AABB, sampling points are uniformly generated along two sides on its six rectangular surfaces at predetermined step sizes. For the OBB, the sampling points are generated using the same strategy as the AABB in its local coordinate system. Then, the sampling points are rotated and translated. Transform to the world coordinate system.
[0008] Preferably, the collision detection is performed for each of the sampling points. A ray Ray is emitted along the outward normal direction of the bounding box surface at that point, and the intersection point of the ray Ray and the bounding box surface is calculated. The position of the ray is used to calculate the intersection point between the ray and the target to be captured. The position; then calculate the... With the Euclidean distance between :
[0009]
[0010] This distance reflects the distance at the sampling point The degree of offset between the surface of the enclosing box and the surface of the target to be captured.
[0011] Preferably, the optimal bounding box refers to: traversing the N sampling points Calculate each sampling point separately The distance set corresponding to the AABB and the OBB { AABB} and { OBB calculates the average fit distance for each type of bezel. :
[0012]
[0013] Compare and Size:
[0014]
[0015] Will and The smaller bounding box type is selected as the optimal bounding box.
[0016] Preferably, step S2 specifically includes obtaining the sampling point density of the optimal bounding box:
[0017] Fitting surface selection: Based on the incoming direction of the mesh, the optimal bounding box surface that is opposite to the incoming direction is excluded, and the remaining five surfaces are selected as the fitting surfaces;
[0018] Sampling point generation: On each surface to be fitted, the density of the sampling grid is determined based on the physical mesh size of the mesh.
[0019]
[0020] in, The physical mesh size is represented by Width, the width of a certain fitting surface is represented by Height, and ⌈⋅⌉ indicates rounding up. Then, a [missing information] is generated on a certain fitting surface. × The two-dimensional point matrix is the set of sampling points for the surface to be fitted. ,in ∈ Repeat this process for all five surfaces to be fitted to obtain the total set of sampling points. Through the total set of sampling points The sampling point density is obtained.
[0021] Preferably, the one-to-one fitting in S2 is as follows: mapping each sampling point to a unique network node in the network, the mapping rule being based on the minimum spatial distance principle: for each sampling point Traverse all nodes of the network and calculate their spatial Euclidean distances. Then, select the node with the smallest distance. The candidate nodes are fitted, and the final output is the mapping dictionary M, which includes keys and values, wherein the keys are unique identifiers of the network nodes. The value is the target position that the network node needs to be fitted to. .
[0022] Preferably, if multiple sampling points occur Associated with the same In the case of these conflicting sampling points In, only those related to the above are retained. The nearest sampling point Establish a one-to-one binding relationship with it, and the remaining sampling points If the sampling point is not found, the allocation process will re-enter, searching for the nearest, unbound network node as the new target, until all sampling points are found. All have been uniquely bound.
[0023] 2. Preferably, updating the motion trajectory of the network node in S3 includes determining parameters, the parameters including:
[0024] The simulated positions of the network nodes ,in The moment when the net comes into contact with the target to be captured;
[0025] The network node is located at the target position corresponding to the mapping dictionary M. ;
[0026] The fitting duration of the network nodes: A fixed interpolation time window ΔT is set, so that the fitting process... Completed in a timely manner.
[0027] Preferably, S3 further includes: performing trajectory interpolation calculation: for any network node, at any time t after contact... ≤t≤ Its new rendering location Calculated using the following linear interpolation formula:
[0028]
[0029] in, It is a normalized interpolation factor whose value increases linearly from 0 to 1 when t> After that, the position of the node is fixed. .
[0030] Preferably, for each node in the mapping dictionary M, its position is calculated according to the linear interpolation formula based on the current time t; for nodes not in the mapping dictionary M, their positions are directly read from the pre-calculated physical simulation data. The updated position data of all the mesh nodes are submitted to the WebGL rendering pipeline to drive the mesh model to deform, thereby visually presenting the dynamic effect of the mesh adaptively wrapping the target.
[0031] The beneficial effects of this invention are as follows:
[0032] (1) The motion of the net body is driven by offline finite element simulation data, which ensures that the net body conforms to the physical laws during the flight and deployment stages and has a high sense of realism;
[0033] (2) After contact, the inherent characteristics of the mesh model are used to perform lightweight fitting with the target geometry, which realizes the adaptive wrapping effect for targets of different sizes and shapes, while avoiding heavy physical calculations on the browser side, resulting in extremely high rendering efficiency. Attached Figure Description
[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0035] Figure 1 This is a fitting relationship diagram in this invention. Detailed Implementation
[0036] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0037] A 3D net-capture visualization method based on simulation data is applied to nets used to capture targets. This method is based on a mature 3D earth visualization library (such as Cesium.js), and its core lies in processing the specific physical model of the net. The net is represented in 3D space as an n×n two-dimensional net structure physical model composed of nodes and edges. Its initial state and motion trajectory during launch and deployment are pre-calculated using high-precision offline finite element physical simulation. This scheme initiates a lightweight real-time fitting process after the net contacts the target, enabling the net to adaptively conform to the target shape. The specific steps are as follows:
[0038] S1. Confirm the optimal bounding box of the target: Calculate the bounding box of the target by uniformly sampling the surface of the bounding box to generate N sampling points. and from sampling points Collision detection is performed on emitted rays. The distance between the intersection point of each ray and the bounding box surface, and the intersection point of each ray and the target surface, is calculated. The bounding box type with the smallest average distance of all rays is selected as the optimal bounding box. Specifically, this includes:
[0039] Generation of bounding box surface sampling points:
[0040] The S1 bounding box includes an OBB (oriented bounding box) and an AABB (axial bounding box). For the AABB, sampling points are uniformly generated along two sides of its six rectangular surfaces at a set step size (either a fixed step size or an adaptive step size based on the bounding box size). For OBB, in its local coordinate system, the same strategy as AABB is used to generate sampling points. Then, the sampling points are moved by rotation and translation. Transform to the world coordinate system.
[0041] Ray collision detection and distance calculation:
[0042] For each sampling point Draw a ray Ray along the outer normal direction of the bounding box surface at that point (or in the direction pointing towards the target's centroid), and calculate the intersection point of the ray Ray and the bounding box surface. The position is used to calculate the intersection point of the ray and the target to be captured. The position was determined using the Möller–Trumbore (ray-triangle intersection) algorithm; then the calculation was performed. and Euclidean distance between :
[0043]
[0044] This distance reflects the distance at the sampling point. The degree of offset between the surface of the bounding box and the surface of the target to be captured.
[0045] Average distance calculation and optimal selection:
[0046] The optimal bounding box refers to the bounding box that has traversed N sampling points. Calculate each sampling point separately The distance set corresponding to AABB and OBB { AABB} and { OBB calculates the average fit distance for each type of bezel. :
[0047]
[0048] Compare and Size:
[0049]
[0050] Will and The smaller bounding box type was selected as the optimal bounding box. The bounding box type with a smaller average distance indicates that it has a better overall geometric fit with the target model, and therefore was selected as the optimal bounding box.
[0051] S2. After determining the optimal bounding box, fit each node of the mesh to the surface of the optimal bounding box one-to-one and output a mapping dictionary M to complete the fitting. The regular n×n mesh structure is the key to this step, specifically including:
[0052] Determine the bounding box fitting surface and sampling point density:
[0053] Fitting surface selection: Based on the incoming direction of the mesh (determined by simulation data or the initial direction of projection), the optimal bounding box surface opposite to the incoming direction is excluded, and the remaining five surfaces are selected as the fitting surfaces;
[0054] Sampling point generation: On each surface to be fitted, the density of the sampling grid is determined based on the physical mesh size of the mesh.
[0055]
[0056] in, The physical mesh size is represented by Width, the width of a fitted surface is represented by Height, and ⌈⋅⌉ indicates rounding up. Then, a [missing information - likely a specific mesh size] is generated on the fitted surface. × The two-dimensional point matrix is the set of sampling points for the surface to be fitted. ,in ∈ Repeat this process for all five surfaces to be fitted to obtain the total set of sampling points. Through the total set of sampling points The sampling point density is obtained.
[0057] Perform the "sampling point - network node" mapping:
[0058] Each sampling point Mapping to a unique node in the network, the mapping rule is based on the principle of minimum spatial distance: for each sampling point Traverse all nodes of the network and calculate their spatial Euclidean distances. Select the node with the smallest distance. The candidate nodes are fitted, and the final output is a mapping dictionary M, which includes keys and values, where the key is a unique identifier of a network node. The value is the target position that the network node needs to be fitted to. .
[0059] If multiple sampling points occur Associated with the same In the case of these conflicting sampling points In the middle, only those related to The nearest sampling point Establish a one-to-one binding relationship with it, and the remaining sampling points If the target is not found, the allocation process will restart, searching for the nearest unbound network node as the new target, until all sampling points are reached. All have been uniquely bound.
[0060] The source code for the fitting process in S2 includes, but is not limited to:
[0061] javascript
[0062] / / Example: A core mapping function snippet
[0063] function mapPointsToNetNodes(samplePoints, netNodeArray) {
[0064] const nodeMap = new Map();
[0065] for (const sample of samplePoints) {
[0066] / / Core logic: Find the nearest node in the network for each sampling point
[0067] let closestNode = findClosestNode(sample.position,netNodeArray);
[0068] / / Processing mapping relationships may involve calculating node weights or position averages.
[0069] if (!nodeMap.has(closestNode.id)) {
[0070] nodeMap.set(closestNode.id, []);
[0071] }
[0072] nodeMap.get(closestNode.id).push(sample.position);
[0073] }
[0074] / / Calculate the final target location for each node (e.g., take the center of the associated sampling points).
[0075] return calculateFinalTargetPositions(nodeMap);
[0076] }
[0077] The resulting executable program requires protection of the static JavaScript library file (e.g., GVIZ-NetCapture.js) generated after compiling and packaging the above source code. This library can be independently integrated into any web 3D application project based on Cesium or a similar engine, providing complete netcapture visualization capabilities.
[0078] S3. After fitting, the motion trajectories of the net nodes are updated in the time dimension to achieve a visual transition from free physical motion to a state of conforming to the target, ultimately completing the net capture visualization. This is achieved through interpolation calculations, specifically including:
[0079] Determine the interpolation parameters:
[0080] Initial state: The moment when the net makes contact with the target. Record the original simulation position of each node (Nodeid) that needs to be fitted. .
[0081] Target state: Retrieve the target location corresponding to this node from the mapping dictionary M. .
[0082] Interpolation duration: Set a fixed interpolation time window ΔT, so that the fitting process is within... Completed in a timely manner.
[0083] Perform trajectory interpolation calculation (taking linear interpolation as an example):
[0084] For any node in the network, at any time t after contact, ≤t≤ Its new rendering location Calculated using the following linear interpolation formula:
[0085]
[0086] in, It is a normalized interpolation factor whose value increases linearly from 0 to 1 when t> After that, the position of the node is fixed. .
[0087] Data update and rendering:
[0088] In each frame rendering loop, for each node in the mapping dictionary M, its position is calculated according to the linear interpolation formula described above based on the current time t.
[0089] For nodes not in the mapping dictionary M (i.e., those that do not need to be fitted), their positions are directly read from the pre-computed physical simulation data. .
[0090] The updated node position data is submitted to the WebGL rendering pipeline to drive the mesh model to deform, thus visually presenting a dynamic effect of the mesh adaptively wrapping the target.
[0091] The fitting result is shown in Figure 1. The coordinates represent the cross-section of the mesh, with black circles representing mesh nodes and the rectangle on the right representing the cross-section of the bounding box of the target. The red star-shaped points indicate the positions of the mesh nodes to be fitted. The green lines show the correspondence between mesh points and fitting points.
[0092] The prior description of this disclosure is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not intended to be limited to the examples and designs described herein, but should be accorded the widest scope consistent with the principles and novel features disclosed herein.
[0093] The embodiments of the present invention have been described in detail above. The description of the embodiments above is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.
Claims
1. A three-dimensional net-capturing visualization method based on simulation data, applied to a net for capturing targets, characterized in that, Includes the following steps: S1. Confirm the optimal bounding box of the target to be captured: Calculate the bounding box of the target to be captured, and generate N sampling points by uniformly sampling the surface of the bounding box. and from the sampling point Collision detection is performed by emitting rays, and the distance between the intersection point of each ray and the surface of the bounding box and the intersection point of each ray and the surface of the target to be captured is calculated. The bounding box type with the smallest average distance of all rays is selected as the optimal bounding box. S2. After determining the optimal bounding box, fit the mesh nodes of the mesh to the surface of the optimal bounding box one-to-one and output a mapping dictionary M to complete the fitting. S3. After the fitting is completed, the motion trajectory of the net nodes is updated in the time dimension to realize the visual transition from free physical motion to the state of fitting the target and finally complete the visualization of net capture.
2. The three-dimensional net-capture visualization method according to claim 1, characterized in that, S1 The bounding box includes an OBB (oriented bounding box) and an AABB (axial bounding box), wherein for the AABB, sampling points are uniformly generated along two sides on its six rectangular surfaces at a set step size. For the OBB, the sampling points are generated using the same strategy as the AABB in its local coordinate system. Then, the sampling points are rotated and translated. Transform to the world coordinate system.
3. The three-dimensional net-capture visualization method according to claim 1, characterized in that, The collision detection is as follows: for each of the sampling points A ray Ray is emitted along the outward normal direction of the bounding box surface at that point, and the intersection point of the ray Ray and the bounding box surface is calculated. The position of the ray is used to calculate the intersection point between the ray and the target to be captured. The position; then calculate the... With the Euclidean distance between : This distance reflects the distance at the sampling point The degree of offset between the surface of the enclosing box and the surface of the target to be captured.
4. The three-dimensional net-capturing visualization method according to claim 1, characterized in that, The optimal bounding box refers to: traversing the N sampling points Calculate each sampling point separately The distance set corresponding to the AABB and the OBB { AABB} and { OBB calculates the average fit distance for each type of bezel. : Compare and Size: Will and The smaller bounding box type is selected as the optimal bounding box.
5. The three-dimensional net-capturing visualization method according to claim 1, characterized in that, S2 specifically includes obtaining the sampling point density of the optimal bounding box: Fitting surface selection: Based on the incoming direction of the mesh, the optimal bounding box surface that is opposite to the incoming direction is excluded, and the remaining five surfaces are selected as the fitting surfaces; Sampling point generation: On each surface to be fitted, the density of the sampling grid is determined based on the physical mesh size of the mesh. in, The physical mesh size is represented by Width, the width of a certain fitting surface is represented by Height, and ⌈⋅⌉ indicates rounding up. Then, a [missing information] is generated on a certain fitting surface. × The two-dimensional point matrix is the set of sampling points for the surface to be fitted. ,in ∈ Repeat this process for all five surfaces to be fitted to obtain the total set of sampling points. Through the total set of sampling points The sampling point density is obtained.
6. The three-dimensional net-capturing visualization method according to claim 1, characterized in that, The one-to-one fitting described in S2 is as follows: mapping each sampling point to a unique network node in the network, with the mapping rule based on the minimum spatial distance principle: for each sampling point... Traverse all nodes of the network and calculate their spatial Euclidean distances. Then, select the node with the smallest distance. The candidate nodes are fitted, and the final output is the mapping dictionary M, which includes keys and values, wherein the keys are unique identifiers of the network nodes. The value is the target position that the network node needs to be fitted to. .
7. The three-dimensional net-capture visualization method according to claim 6, characterized in that, If multiple sampling points appear Associated with the same In the case of these conflicting sampling points In, only those related to the above are retained. The nearest sampling point Establish a one-to-one binding relationship with it, and the remaining sampling points If the sampling point is not found, the allocation process will re-enter, searching for the nearest, unbound network node as the new target, until all sampling points are found. All have been uniquely bound.
8. The three-dimensional net-capturing visualization method according to claim 1, characterized in that, Updating the motion trajectory of the network node in S3 includes determining parameters, which include: The simulated positions of the network nodes ,in The moment when the net comes into contact with the target to be captured; The network node is located at the target position corresponding to the mapping dictionary M. ; The fitting duration of the network nodes: A fixed interpolation time window ΔT is set, so that the fitting process... Completed in a timely manner.
9. The three-dimensional net-capture visualization method according to claim 8, characterized in that, S3 further includes: performing trajectory interpolation calculation: for any network node, at any time t after contact, ≤t≤ Its new rendering location Calculated using the following linear interpolation formula: in, It is a normalized interpolation factor whose value increases linearly from 0 to 1 when t> After that, the position of the node is fixed. .
10. The three-dimensional net-capturing visualization method according to claim 8 or 9, characterized in that, S3 further includes submitting the updated position data of all the mesh nodes to the WebGL rendering pipeline to drive the mesh model to deform, thereby visually presenting a dynamic effect of the mesh adaptively wrapping the target.