Virtual simulation method for symptoms of grape fruit mold diseases

Abstract

The present invention relates to computer graphics, virtual reality, and virtual agriculture technology. A virtual simulation method for the main symptoms of grape fruit mold diseases (mold, discoloration, shrinkage) is proposed, which can simulate various molds on the fruit surface, conveniently simulate the discoloration of the fruit peel, and more accurately simulate the volume shrinkage. In the present invention, the virtual simulation method for the symptoms of grape fruit mold diseases includes input, preprocessing, simulation, output, and rendering steps, wherein the simulation step is to calculate the reaction-diffusion model and perform mold growth simulation, discoloration simulation, and volume shrinkage simulation. The present invention is mainly applied to the occasion of plant growth simulation.

Description

Technical Field

[0001] The present invention relates to the fields of computer graphics, virtual reality, virtual agriculture, computer games, film and television special effects, etc., and particularly relates to a virtual simulation method for the symptoms of grape fruit mold diseases. Background Art

[0002] Mold diseases (including powdery mildew, gray mold, downy mildew, etc.) are the main diseases that are likely to occur during grape cultivation. The infected grape fruits will produce symptoms such as color change, surface mold, and volume shrinkage, greatly reducing their commerciality. The virtual simulation of this process can provide insights for food appearance simulation in computer graphics, and is also conducive to content creation in virtual reality, virtual agriculture, computer games, film and television special effects, etc. Currently, the virtual simulation of plant disease symptoms (color change, mold, shrinkage) mostly focuses on fruits and leaves.

[0003] The simulation of plant color change mainly includes image-based methods [1][2] and methods based on reaction-diffusion models [3][4][5]. The former usually preprocesses the lesion image first, and then generates a simulated lesion through an interactive method. Such methods usually have high requirements for the lesion image, such as the need to contain color diffusion characteristics, etc. [2]. The method based on the reaction-diffusion model represents the expansion of the fruit infection area by calculating a soft rot map [3], a fungus map [4], or a nutrient map [5], and updates the fruit color using alpha blending, but the parameters are not easy to adjust.

[0004] There are few simulation methods for plant pathogenic bacteria. The invention patent [6] simulates powdery mildew and rust on leaves, but does not mention the simulation of filamentous fungi. The literature [3] needs to use the software Maya Fur to simulate the filamentous fungi on the fruit surface, but this method is difficult to model the fungi of powdery mildew. The literature [5] uses the L-system to model the fungi on the fruit surface. Although it is a potential method that can simulate various fungi, its grammar is difficult to set. Currently, there is a lack of a method that can simulate various different forms of molds causing grape diseases.

[0005] The simulation of plant shrinkage mainly includes the finite element-based method and the particle spring system-based method. The former, such as the literature [7], [8], can depict details such as the wrinkles of fruit water loss and shrinkage, but it is not suitable for simulating large deformations and the calculation is relatively complex. The latter is commonly used in leaf shrinkage simulation, and a single layer [9] or a double layer

[10] of particle springs is used to simulate the deformation of the leaf. The simulation of fruit shrinkage based on the particle spring system mainly has two methods in the literature [3] and the literature

[11] , but both of these methods regard the fruit as hollow, which does not conform to the actual situation of grapes. And the above methods are all difficult to accurately represent the biological property changes during the grape water loss and shrinkage process.

[0006] References:

[0007] [1] CN 103400410 A, July 12, 2013.

[0008] [2] Wu S, Xiao B, Miao T, et al. An interactive design method for realistic fruit rot modeling and simulation, 2018: 1850038.

[0009] [3] Kider Jr J T, Raja S, Badler N I. Fruit senescence and decay simulation. Computer Graphics Forum, 2011, 30(2): 257 - 266.

[0010] [4] Fan D, Liu S, Wei Y. Fruit ring rot simulation based on reaction - diffusion model. International Conference on Virtual Reality and Visualization, 2013: 199 - 205.

[0011] [5] Cirdei B, Anderson E F. Withering fruits: Vegetable matter decay and fungus growth. ACM SIGGRAPH, 2018: 1 - 2.

[0012] [6] CN 102867325 B, June 29, 2012.

[0013] [7] Liu Y, Chen Y, Wu W, et al. Physically based object withering simulation. Computer Animation and Virtual Worlds, 2012, 23(3 - 4): 395 - 406.

[0014] [8] Liu Y, Yang X, Cao Y, et al. Dehydration of core / shell fruits. Computers and Graphics, 2015, 47: 66 - 87.

[0015] [9]CN 102663816 A, April 1, 2012.

[0016]

[10] Jeong S H, Park S H, Kim C H. Simulation of morphology changes in drying leaves. Computer Graphics Forum, 2013, 32(1): 204 - 215.

[0017]

[11] Liu S, Fan D. Computer modeling and simulation of fruit sunscald. International Journal of Image and Graphics, 2015, 15(3): 1550013 Summary of the Invention

[0018] To overcome the deficiencies of the prior art, the present invention aims to propose a virtual simulation method for the main symptoms of grape fruit mold diseases (mold, discoloration, shrinkage), which can simulate various molds on the fruit surface, conveniently simulate the discoloration of the fruit peel, and more accurately simulate the volume shrinkage. To this end, the technical solution adopted by the present invention is a virtual simulation method for the symptoms of grape fruit mold diseases, and the steps are as follows

[0019] A Input: A grape fruit model, including UV texture and triangular mesh (Mesh);

[0020] B Pretreatment:

[0021] B1: Perform Poisson disk sampling on the texture to calculate the distribution of hyphae on the texture and generate hyphal growth points;

[0022] B2: Establish a template hyphal model according to the microscopic morphology of the pathogenic bacteria of the disease to be simulated. Among them, the hyphae are represented by a segmented cylinder model, and the spores are represented by a spherical model;

[0023] B3: Fill the inside of the fruit mesh with particles. The internal particles are arranged in a hexagonal pattern, and adjacent particles are connected by a spring S in to form a mass - spring system MMS of internal particles in .

[0024] B4: Connect the internal and external particles: The internal particles are the particles filled in B3, and the external particles are the vertices of the Mesh input in step A;

[0025] B4.1: Connect the external particles, using the edges of the Mesh as springs S out, taking the vertices of the Mesh, i.e., the external particles, as mass points, to form the mass point spring system MMS of the external particles out ;

[0026] B4.2: Establish internal proxy particles and connect them to the external particles. For each external particle p o , taking the negative normal direction of p o as the outgoing direction to make a ray, traversing all the triangles formed by any 3 particles selected from the 8 internal particles closest to p o , and determining whether it intersects with the ray: if it intersects, then compare the triangle areas, select the triangle with the smallest area, and the intersection point of the ray and this triangle is used as the proxy particle p agnet , and record the barycentric coordinates of p agnet and the triangle Δagent, and connect the proxy particle p oa with this external particle p agnet using the spring S o ;

[0027] C simulation

[0028] C0 Calculate the reaction-diffusion model on the Mesh: find the active fungal value u o of each external particle p a , the inactive fungal value u i and the nutrient value n, and calculate the fungal value u ai = u a + u i , and the fungal value u of each texture pixel and each hyphal growth point is calculated through its barycentric coordinates and the fungal values u ai of the 3 vertices of the corresponding triangle in the Mesh;

[0029] C1 Mold growth simulation

[0030] For each hypha, use the barycentric coordinates to map the pre-generated hyphal growth points to the surface of the fruit model Mesh to obtain the actual hyphal growth points x h ; use the barycentric coordinates and the vertex normal of the Mesh triangle where the hypha is located to calculate the hyphal normal n h , and calculate the hyphal length l h , radius r h using the following formula:

[0031] l h = k l u, (1)

[0032] r h = k r u, (2)

[0033] where k l is the length coefficient, k ris the radius coefficient, denoted by l h Compare it with the length of the template model used, and instantiate the current l h The part of the template model that can be covered. A spherical model can be instantiated at the top of the hypha to represent the spore.

[0034] For the pathogenic bacteria of powdery mildew, the spore model must be instantiated to simulate the effect of powdery mildew particles;

[0035] C2 color change simulation: The color change of each texture pixel is calculated using a zero-order kinetic model, and the formula is as follows:

[0036] C = C0 - k c C0u, (3)

[0037] where C and C0 represent the current and initial colors of the pixel, and the vector k c stores the color change coefficients for the 3 RGB channels;

[0038] C3 volume shrinkage simulation: Based on the particle spring system MMS in step B3 in calculate the position change of the internal particles due to water loss, thereby driving the external particles p o to move to simulate the volume shrinkage of the fruit. The specific steps are as follows:

[0039] C3.1 Update the water-loss seed particles: If the fungal value of the external particle p o is greater than the set threshold, then the vertices (internal particles) of the triangle Δagent to which it is connected are all set as seed particles; if the water loss of the seed particle is greater than the set threshold, then the internal particles adjacent to it are all set as seed particles; agnet agnet

[0040] C3.2 Calculate the positions of the internal particles: Based on the particle spring system MMS in step B3 in calculate the position change of the internal particles due to water loss;

[0041] C3.2.1 Calculate the forces: The internal particle i is affected by 3 forces, the turgor pressure the damping force the elastic force that is

[0042]

[0043] C3.2.1.1 Calculate the turgor pressure The turgor pressure caused by particle water loss is the driving force for MMS in particle movement and is calculated according to the following formula:

[0044]

[0045] where, kp is the turgor pressure coefficient, P represents the turgor pressure, j represents the particle adjacent to i, the subscript 0 represents the initial value, x represents the particle position, represents the direction of the force, and the turgor pressure P is calculated by the following formula i :

[0046]

[0047] where t is the time, m is the particle mass, ε i is the bulk modulus of elasticity and is calculated by the following formula:

[0048]

[0049] where ε ∞ is the maximum bulk modulus of elasticity, k ε is the rate constant, P a is the plateau pressure, ε ∞ , k ε , P a are all constants;

[0050] In formula (6), the change in the mass m of the dehydrated seed particles i is calculated by the following formula:

[0051]

[0052] where k m is the dehydration coefficient, T is the environmental temperature, H is the environmental humidity, D w is the moisture diffusion coefficient;

[0053] In formula (6), the change in the mass of non-seed particles only calculates moisture diffusion:

[0054]

[0055] C3.2.1.2 Calculate the damping force

[0056]

[0057] where k d is the damping coefficient and v represents the velocity;

[0058] C3.2.1.3 Calculate the elastic force Taking the Lennard-Jones force as the elastic force, its formula is:

[0059]

[0060] where, represents the magnitude of the force and is calculated by the following formula:

[0061]

[0062] Among them, k e is the coefficient of force, and r i is the particle radius, and the calculation formula is:

[0063]

[0064] Among them, r0 is the initial particle radius;

[0065] C3.2.2 Calculate the acceleration according to the particle mass and the force applied, and then update the internal particle velocity and position;

[0066] C3.3 Calculate the external particle positions: According to each proxy particle p agnet The position of the internal particle is at the vertex of the Δagent where it is located and its centroid coordinates are used to update the position of p agnet The elastic force of the spring S oa is used to drive the movement of the external particle it is connected to, and the external particle mass-spring system MMS out is calculated, the external particle positions are updated, and the fruit model Mesh is updated;

[0067] Loop through the entire simulation step C until the set number of loops is reached;

[0068] D Output: In each loop of step C, output the results of the simulation, including the updated hypha model, fruit model texture, and mesh;

[0069] E Rendering: Render the results output in step D to obtain the simulation result image. The result image is composited into an animation of the disease process at 25 frames per second.

[0070] In step B1: Combine the Perlin noise map and Poisson disk sampling to make the hypha distribution more natural;

[0071] In step B2: To increase the hypha diversity, multiple template models are established for each type of mold, and a fixed model is selected for each hypha during preprocessing, and several random variables are set for each hypha to increase the diversity;

[0072] In step B3: First voxelize the Mesh to obtain the voxel grid of the fruit, and then use the seed filling algorithm to fill the particles in a hexagonal arrangement;

[0073] In step B4: B4.1: Add bending springs between the external particles;

[0074] In step C0, to ensure stability, an implicit integration method is used to calculate the diffusion model;

[0075] C3.2.2 Calculate the acceleration based on the particle mass and the force applied, and then update the internal particle velocity and position. Use the leapfrog integration to calculate this process.

[0076] The features and beneficial effects of the present invention are as follows:

[0077] The virtual simulation method for grape fruit mold diseases of the present invention can simulate the morphologies of various pathogenic molds by adjusting the template hypha model; the method based on the zero-order kinetic model proposed can conveniently set parameters to simulate the fruit peel discoloration; the adaptive particle spring system with the cell turgor pressure as the driving force and the Lennard-Jones force as the elastic force can more accurately describe the property changes caused by plant water loss and meet the need for simulating volume shrinkage. The three simulated symptoms combined together can comprehensively simulate the process of grape fruit mold diseases. Description of the Drawings

[0078] Figure 1 is the flowchart of the virtual simulation method for grape fruit mold disease symptoms of the present invention.

[0079] Figure 2 is the schematic diagram of the distribution of hyphae on the texture generated by the present invention.

[0080] Figure 3 is the template hypha model for downy mildew and gray mold established by the present invention.

[0081] Figure 4 is the template hypha model for powdery mildew established by the present invention.

[0082] Figure 5 is the schematic diagram of the internal particle arrangement of grapes of the present invention.

[0083] Figure 6 is the simulation effect of the mold change of grapes infected with powdery mildew generated by the present invention.

[0084] Figure 7 is the simulation effect of downy mildew of grapes generated by the present invention.

[0085] Figure 8 is the simulation effect of gray mold of grapes generated by the present invention.

[0086] Figure 9 is the simulation effect of the discoloration and shrinkage process of grapes generated by the present invention. Detailed Embodiment

[0087] The technical solution of the present invention is described in detail as follows:

[0088] A Input: Grape fruit model, including UV texture and triangular mesh (Mesh).

[0089] B Pretreatment:

[0090] B1: Perform Poisson disk sampling on the texture to calculate the distribution of hyphae on the texture and generate hyphal growth points. Preferably, Perlin noise map can be combined with Poisson disk sampling to make the hyphal distribution more natural.

[0091] B2: Establish a template hyphal model according to the microscopic morphology of the pathogenic bacteria of the disease to be simulated. Among them, the hyphae are represented by segmented cylinder models, and the spores are represented by sphere models. Preferably, to increase the diversity of hyphae, multiple template models can be established for each type of mold, and a fixed model can be selected for each hypha during preprocessing. Preferably, several random variables can be set for each hypha to increase diversity.

[0092] B3: Fill the interior of the fruit mesh with particles, and the internal particles are arranged in hexagons. Preferably, the Mesh is first voxelized to obtain the voxel grid (grid) of the fruit, and then the particles are filled in hexagons through the seed filling algorithm. Connect adjacent particles with spring S in to form the particle mass-spring system MMS of the internal particles in .

[0093] B4: Connect the internal and external particles: The internal particles are the particles filled in B3, and the external particles are the vertices of the Mesh input in step A.

[0094] B4.1: Connect the external particles. Use the edges of the Mesh as springs S out , and use the vertices (external particles) of the Mesh as mass points to form the particle mass-spring system MMS of the external particles out . Preferably, bending springs can be added between the external particles.

[0095] B4.2: Establish internal proxy particles and connect them to the external particles. For each external particle p o , make a ray with the negative normal of p o as the outgoing direction, traverse all triangles formed by any 3 particles among the 8 particles closest to p o , and judge whether it intersects with the ray: If it intersects, compare the triangle areas, take the triangle with the smallest area, and the intersection point of the ray and the triangle is used as the proxy particle p agnet , and record the barycentric coordinates of p agnet and the triangle Δagent. Connect the proxy particle p oa to the external particle p agnet with spring S o .

[0096] C Simulation

[0097] Calculating the reaction-diffusion model on the Mesh: For each external particle p o calculate the active fungal value u a , the inactive fungal value u i and the nutrient value n, and calculate the fungal value u ai = u a + u i . The fungal value u of each texture pixel and each hyphal growth point is calculated through its centroid coordinates and the fungal values u ai of the three vertices of the corresponding triangle in the Mesh.

[0098] Preferably, to ensure stability, an implicit integration method is used to calculate the reaction-diffusion model.

[0099] C1 Mold growth simulation

[0100] For each hypha, use the centroid coordinates to map the hyphal growth points generated by preprocessing to the surface of the fruit model Mesh to obtain the actual hyphal growth point x h . Similarly, use the centroid coordinates and the vertex normal of the Mesh triangle where the hypha is located to calculate the hyphal normal n h . Use the following formula to calculate the hyphal length l h , the radius r h :

[0101] l h = k l u, (1)

[0102] r h = k r u, (2)

[0103] where k l is the length coefficient, k r is the radius coefficient. Compare l h with the length of the template model used, and instantiate the part of the template model that the current l h can cover.

[0104] Optionally, if the hyphal lifetime reaches the set threshold, a sphere model can be instantiated at the top of the hypha to represent the spore.

[0105] For the powdery mildew pathogen, the spore model must be instantiated to simulate the effect of powdery mildew particles.

[0106] C2 Color change simulation: The color change of each texture pixel is calculated using a zero-order kinetic model, and its formula is as follows:

[0107] C = C0 - k c C0u, (3)

[0108] where C and C0 represent the current and initial colors of the pixel, and the vector kc Store the color change coefficients of the RGB three channels.

[0109] C3 volume shrinkage simulation: Based on the particle spring system MMS in step B3 in , calculate the position change of the internal particles caused by water loss, so as to drive the external particle p o to move to simulate the volume shrinkage of the fruit. The specific steps are as follows:

[0110] C3.1 Update the water-loss seed particles: If the fungal value of the external particle p o is greater than the set threshold, then all the vertices (internal particles) of the triangle Δagent to which it is connected are set as seed particles; if the water loss of the seed particle is greater than the set threshold, then all the internal particles adjacent to it are set as seed particles. agnet

[0111] C3.2 Calculate the positions of the internal particles: Based on the particle spring system MMS in step B3 in , calculate the position change of the internal particles caused by water loss.

[0112] C3.2.1 Calculate the force: The internal particle i is affected by three forces, the turgor pressure the damping force the elastic force That is

[0113]

[0114] C3.2.1.1 Calculate the turgor pressure The turgor pressure caused by the water loss of the particle is the driving force for the movement of MMS in particle, and is calculated according to the following formula:

[0115]

[0116] Among them, k p is the turgor pressure coefficient, P represents the turgor pressure, j represents the particle adjacent to i, the subscript 〇 represents the initial value, x represents the particle position, represents the direction of the force. Calculate the turgor pressure P through the following formula i :

[0117]

[0118] ` Among them, t is the time, m is the particle mass, ε i is the bulk modulus of elasticity, preferably, it can be calculated according to the following formula:

[0119]

[0120] Among them, ε ∞is the maximum bulk elastic modulus, k ε is the rate constant, P a is the plateau pressure, ε ∞ 、k ε 、P a are all constants.

[0121] In formula (6), the mass m of the dehydrated seed particles i changes according to the following formula:

[0122]

[0123] where k m is the dehydration coefficient, T is the ambient temperature, H is the ambient humidity, D w is the moisture diffusion coefficient.

[0124] In formula (6), the change in the mass of non-seed particles only calculates moisture diffusion:

[0125]

[0126] C3.2.1.2 Calculate the damping force

[0127]

[0128] where k d is the damping coefficient, and v represents the velocity.

[0129] C3.2.1.3 Calculate the elastic force Taking the Lennard-Jones force as the elastic force, its formula is:

[0130]

[0131] where represents the magnitude of the force and is calculated according to the following formula:

[0132]

[0133] where k e is the force coefficient, r i is the particle radius, and the calculation formula is:

[0134]

[0135] where r0 is the initial particle radius.

[0136] C3.2.2 Calculate the acceleration based on the particle mass and force, and then update the internal particle velocity and position. Preferably, the leapfrog integration is used to calculate this process.

[0137] C3.3 Calculate the position of external particles: Based on each agent particle p agnet Update the position of p according to the position of the Δagent vertex (internal particle) where it is located and its centroid coordinates agnet and use the elastic force of spring S oa to drive the movement of the external particles connected to it, calculate the multi-material spring system MMS of the external particle mass points out , update the position of the external particles, and update the fruit model Mesh.

[0138] Loop through the entire simulation step C until the set number of loops is reached.

[0139] D Output: Output the results of the simulation in each loop of step C, including the updated hypha model, fruit model texture and mesh;

[0140] E Rendering: Render the results output in step D to obtain the simulation result image. Synthesize the disease process animation at 25 frames per second with the result image.

[0141] The following further details the best implementation mode of the present invention in combination with the attached drawings and specific implementation modes.

[0142] Figure 1 is the flow chart of this method, including the following steps:

[0143] A Input: Grape fruit model, including UV texture and triangular mesh (Mesh).

[0144] B Preprocessing:

[0145] B1: Perform Poisson disk sampling on the texture to calculate the distribution of hyphae on the texture. A calculated distribution schematic diagram is as Figure 2 shown. The black dots in the texture represent the hypha growth points. Preferably, Poisson disk sampling can be combined with the Perlin noise map to make the hypha distribution more natural.

[0146] B2: Establish a template hypha model according to the microscopic morphology of the disease-causing pathogen to be simulated, such as Figure 3 (downy mildew, gray mold), Figure 4 (powdery mildew) shown, where the hyphae are represented by segmented cylinder models and the spores are represented by sphere models. Preferably, to increase the diversity of hyphae, Figure 3 10 models of hyphae are established, and a fixed model can be selected for each hypha during preprocessing. Preferably, during preprocessing, a random rotation angle, length coefficient, radius coefficient, etc. can be set for each hypha to increase the diversity, and these values remain unchanged during the simulation process.

[0147] B3: Fill the inside of the fruit mesh with particles. The internal particles are arranged in a hexagonal pattern, asFigure 5 Taking the gray particles as the center, they are arranged in a hexagonal pattern with the white particles in the XoY plane and in a hexagonal pattern with the black particles and two white particles on the X-axis in the XoZ plane. Preferably, the Mesh is voxelized to obtain the voxel grid of the fruit, and then the particles are filled by the seed filling algorithm according to the above hexagonal arrangement. For adjacent particles, they are connected by a spring S in to form the mass point spring system MMS of the internal particles in .

[0148] B4: Connect the internal and external particles: The internal particles are the particles filled in step B3, and the external particles are the vertices of the Mesh input in step A

[0149] B4.1: Connect the external particles. Taking the edges of the Mesh as the spring S out and the vertices of the Mesh (external particles) as the mass points to form the mass point spring system MMS of the external particles out . Preferably, a bending spring can be added between the external particles

[0150] B4.2: Establish internal proxy particles and connect them to the external particles. For each external particle p o , with the negative normal of p o as the outgoing direction, a ray is made, and all triangles formed by taking any 3 particles from the 8 internal particles closest to p o are traversed to determine whether it intersects with the ray: If it intersects, then compare the triangle areas, take the triangle with the smallest area, and the intersection point of the ray and this triangle is used as the proxy particle p agnet , and record the barycentric coordinates of p agnet and this triangle Δagent. Connect the proxy particle p oa to this external particle p agnet with the spring S o .

[0151] C Simulation:

[0152] C0: Calculate the reaction-diffusion model on the Mesh to obtain the fungal value of each external particle p o . Preferably, the formula of the reaction-diffusion model adopted (Reference: Fan D, Liu S, Wei Y. Fruit ring rot simulation based onreaction-diffusion model. International Conference on Virtual Reality andVisualization, 2013: 199 - 205.) is as follows:

[0153]

[0154] Among them, u a , u i and n respectively represent the active fungal value, the inactive fungal value, and the nutrient value. u ai = u a + u i represents the fungal value of the external particle p o . D c and D n are the diffusion coefficients of active fungi and nutrients. θ represents the consumption rate of nutrients by active fungi. s(u a , n) represents the conversion coefficient from active fungi to inactive fungi.

[0155] Preferably, to ensure stability, an implicit integration method is used to calculate formula (1).

[0156] For each texture pixel and each hyphal growth point, the fungal value u is calculated through its centroid coordinates and the fungal values u ai of the three vertices of the corresponding triangle in the Mesh.

[0157] C1 Mold Growth Simulation

[0158] C1.1 Downy Mildew and Gray Mold Pathogenic Mold Growth Simulation: The template hyphal model used is as Figure 3 shown.

[0159] For each hypha, using the centroid coordinates, the hyphal growth points generated by preprocessing are mapped to the surface of the fruit model Mesh to obtain the actual hyphal growth points x h . Similarly, using the centroid coordinates and the vertex normal of the Mesh triangle where the hypha is located, combined with the hyphal rotation angle obtained by preprocessing, the hyphal normal n h is calculated. The hyphal length l h , radius r h are calculated using the following formulas:

[0160] l h = k l u, (2)

[0161] r h = k r u, (3)

[0162] Among them, k l is the length coefficient, and k r is the radius coefficient. Compare l h with the length of the template model used, and instantiate the part of the template model that the current l h can cover.

[0163] Optionally, if the hyphal lifetime reaches the set threshold, a sphere model can be instantiated at the top of the hypha to represent spores.

[0164] C1.2 Powdery Mildew Pathogenic Mold Growth Simulation: The template hyphal model used is as shown in Figure 4 the figure.

[0165] Mold growth point x h , normal direction n h , length l h , radius r h are calculated in the same way as in step C1.1. Such molds must instantiate spore models to simulate the effect of powdery mildew particles.

[0166] C2 Color Change Simulation: The color change of each texture pixel is calculated using a zero-order kinetic model, and its formula is as follows:

[0167] C = C0 - k c C0u, (4)

[0168] where C and C0 represent the current and initial colors of the pixel, and the vector k c stores the color change coefficients for the RGB 3 channels. The last term in formula (4) indicates that the color change is related to the fungal value u and the initial color C0.

[0169] C3 Volume Shrinkage Simulation: Based on the particle spring system MMS in step B3 in , calculate the movement of internal particles caused by water loss, thereby driving the movement of external particles p o to simulate the volume shrinkage of the fruit. The specific steps are as follows:

[0170] C3.1 Update Water-Loss Seed Particles: If the fungal value of the external particle p o is greater than the set threshold, then all the vertices (internal particles) of the triangle Δagent to which it is connected are set as seed particles; if the water loss of the seed particle is greater than the set threshold, then all the internal particles adjacent to it are set as seed particles. agnet

[0171] C3.2 Calculate the Positions of Internal Particles: Based on the particle spring system MMS in step B3 in , calculate the position change of internal particles caused by water loss.

[0172] C3.2.1 Calculate the Forces: The internal particle i is affected by 3 forces, the turgor pressure the damping force the elastic force Among them, the turgor pressure caused by particle water loss is the driving force for MMS in particle movement. The force formula for particle i is as follows:

[0173]

[0174] ​C3.2.1.1 Calculate the turgor pressure

[0175]

[0176] Among them, k p is the turgor pressure coefficient, P represents the turgor pressure, j represents the particle adjacent to i, the subscript 0 represents the initial value, x represents the particle position, represents the direction of the force. Formula (6) indicates that the turgor pressure on particle i is caused by the change in turgor pressure between its own current state and initial state and those of its adjacent particles. Calculate the turgor pressure P through the following formula i :

[0177]

[0178] Among them, t is the time, m is the particle mass, ε i is the bulk modulus of elasticity. Preferably, it can be calculated according to the following formula:

[0179]

[0180] Among them, ε ∞ is the maximum bulk modulus of elasticity, k ε is the rate constant, P a is the plateau pressure, ε ∞ , k ε , P a are all constants. Formula (8) shows that the bulk modulus of elasticity changes with the turgor pressure.

[0181] In formula (7), the mass m i of the dehydrated seed particle changes according to the following formula:

[0182]

[0183] Among them, t is the time, k m is the water loss coefficient, T is the environmental temperature, H is the environmental humidity, D w is the water diffusion coefficient. Formula (9) indicates that the change in the mass of the seed particle consists of water loss (the first term) and water diffusion (the second term).

[0184] In formula (7), the mass m i of the non-seed particle changes only by calculating water diffusion:

[0185]

[0186] C3.2.1.2 Calculate the damping force

[0187]

[0188] where k d is the damping coefficient and v represents the velocity.

[0189] C3.2.1.3 Calculate the elastic force

[0190] Since the Lennard-Jones (LJ) force is often used in cell mechanics to represent the attraction and repulsion of cells, and its principle is similar to the elastic force. Therefore, the LJ force is used as the elastic force to represent the attraction and repulsion between the internal particles of the fruit, and its formula is:

[0191]

[0192] where represents the magnitude of the force and is calculated according to the following formula:

[0193]

[0194] where k e is the force coefficient, r i is the particle radius, and the calculation formula is:

[0195]

[0196] where r0 is the initial particle radius, and this formula indicates that as the particle mass decreases, its radius also decreases.

[0197] Furthermore, formula (12) can also be written as:

[0198]

[0199] where can be regarded as the spring elastic coefficient of MMS in and r i can be regarded as the spring rest length, and both of these parameters change with particle water loss, representing the change of cell properties.

[0200] C3.2.2 Calculate the acceleration based on the calculated particle mass and force, and then update the internal particle velocity and position. Preferably, the leapfrog integration is used to calculate this process.

[0201] C3.3 Calculate the external particle position

[0202] Update the position of each proxy particle p agnet according to the position of the Δagent vertex (internal particle) where it is located and its centroid coordinates, and use the elastic force of the spring S agnet to drive the movement of the external particle connected to it, and calculate the external particle mass-spring system MMS oa out ​, update the positions of external particles and update the fruit model Mesh.

[0203] Loop through the entire simulation step C until the set number of loops is reached.

[0204] D Output: Output the results of the simulation in each loop of step C, including the updated hypha model, fruit model texture, and mesh;

[0205] E Rendering: Render the results output in step D to obtain the simulation result image. Synthesize an animation of the disease process at 25 frames per second with the result image.

[0206] By replacing the template hypha model, the present invention can simulate the morphological forms of pathogenic fungi of various mold diseases, including powdery mildew ( Figure 6 ), downy mildew ( Figure 7 ), gray mold ( Figure 8 ), which cannot be achieved by other methods. Figure 6 It well simulates the spread of the powdery mildew pathogen on the surface of grape fruits, Figure 7 shows the effect that the filamentous fungi causing downy mildew cover the entire surface of grape fruits, Figure 8 shows different hypha distribution situations of the simulated gray mold.

[0207] The present invention also well simulates the color change and volume shrinkage of grapes caused by fungal diseases. Figure 9 Shows the simulation effect of the color change and shrinkage process of grapes, where the discolored area gradually expands, and the wrinkles caused by shrinkage gradually increase and deepen. Figure 8 Shows the simulation effects of different degrees of shrinkage of grape fruits.

[0208] As described above, it is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A virtual simulation method for the symptoms of grape fruit mold diseases, characterized in that the steps As follows: A Input: Grape fruit model, including UV texture and triangular mesh Mesh; B Preprocessing: B1: Perform Poisson disk sampling on the texture to calculate the distribution of hyphae on the texture and generate hyphal growth points; B2: Establish a template hyphal model according to the microscopic morphology of the pathogenic bacteria of the disease to be simulated. Among them, the hyphae are represented by a segmented cylinder model, and the spores are represented by a sphere model; B3: Fill the interior of the fruit grid with particles. The internal particles are arranged in hexagons, and adjacent particles are connected by springs S in to form the mass-spring system MMS of the internal particles in ; B4: Connect internal and external particles: The internal particles are the particles filled in B3, and the external particles are the vertices of the Mesh input in step A; B4.1: Connect external particles, with the edges of the Mesh as springs S out , with the vertices of the Mesh, i.e., the external particles, as mass points, to form the mass point spring system MMS of the external particles out ; B4.2: Establish internal proxy particles, connect them to external particles, and for each external particle p o , with p o as the outgoing direction of the ray in the negative normal direction, traverse all the triangles formed by taking any 3 particles from the 8 internal particles closest to p o to determine whether they intersect with the ray: if they intersect, compare the areas of the triangles, select the triangle with the smallest area, and use the intersection point of the ray and this triangle as the proxy particle p agnet , and record the barycentric coordinates of p agnet and this triangle Δagent, and connect the proxy particle p oa with the spring S agnet to this external particle p o ; C Simulation Calculating the reaction-diffusion model on the Mesh: finding the active fungal value u o for each external particle p a , the inactive fungal value u i and the nutrient value n, and calculating the fungal value u ai = u a + u i where the fungal value u for each texture pixel and each hyphal growth point is calculated through its centroid coordinates and the fungal values u ai of the 3 vertices of the corresponding triangle in the Mesh; C1 Mold growth simulation For each hypha, the hypha growth points generated by preprocessing are mapped to the surface of the fruit model Mesh using barycentric coordinates to obtain the actual hypha growth points x h ; the hypha normal vector n is calculated using barycentric coordinates and the vertex normal vector of the Mesh triangle where the hypha is located h , and the hypha length l is calculated using the following formula h , the radius r h : l h = k l u, (1) r h = k r u, (2) where k l is the length coefficient, and k r is the radius coefficient. Comparing l h with the length of the template model used, the part of the template model that the current l h can cover can instantiate a sphere model at the top of the hypha to represent spores; For the pathogenic bacteria of powdery mildew, the spore model must be instantiated to simulate the effect of powdery mildew particles; C2 Discoloration simulation: The discoloration of each texture pixel is calculated using a zero-order kinetic model, and its formula is as follows: C = C0 - k c C0u, (3) where C and C0 represent the current and initial colors of the pixel, and the vector k c stores the color change coefficients for the 3 RGB channels; C3 Volume Atrophy Simulation: Based on the Mass-Spring System MMS in Step B3 in , calculate the position change of internal particles caused by water loss, so as to drive the movement of external particle p o to simulate the volume atrophy of the fruit. The specific steps are as follows: C3.1 Update dehydrated seed particles: If the external particle p o has a fungal value greater than the set threshold, then set the vertices (internal particles) of the triangle Δagent to which it is connected as seed particles; if the water loss of the seed particle is greater than the set threshold, then set all adjacent internal particles as seed particles; agnet ​ C3.2 Calculate the positions of internal particles: Based on the mass-spring system MMS in step B3 in , calculate the position changes of internal particles caused by water loss; C3.2.1 Calculate the acting forces: The internal particle i is affected by three forces, the swelling pressure The damping force The elastic force That is C3.2.1.1 Calculate the turgor pressure The turgor pressure caused by particle water loss is MMS in The driving force for particle movement is calculated according to the following formula: where k p is the turgor pressure coefficient, P represents the turgor pressure, j represents the particle adjacent to i, the subscript 0 represents the initial value, x represents the particle position, represents the direction of the force, and the turgor pressure P is calculated by the following formula i : where t is time, m is the mass of the particle, and ε i is the bulk modulus of elasticity and is calculated by the following formula: Among them, ε ∞ is the maximum bulk elastic modulus, k ε is the rate constant, P a is the plateau pressure, ε ∞ , k ε , P a are all constants; In formula (6), the change in the mass m of the dehydrated seed particles i is calculated according to the following formula: where k m is the water loss coefficient, T is the ambient temperature, H is the ambient humidity, and D w is the water diffusion coefficient; In formula (6), the mass change of non-seed particles only calculates water diffusion: C3.2.1.2 Calculate the damping force where k d is the damping coefficient and v represents the velocity; C3.2.1.3 Calculate the elastic force Use the Lennard-Jones force as the elastic force, and its formula is as follows: wherein, represents the magnitude of the force and is calculated according to the following formula: where k e is the coefficient of force, r i is the particle radius, and the calculation formula is: Among them, r0 is the initial particle radius; C3.2.2 Calculate the acceleration according to the particle mass and force, and then update the internal particle velocity and position; C3.3 Calculate the position of external particles: Based on each proxy particle p agnet The position of the internal particles and the coordinates of their center of gravity at the Δagent vertex where p is located are used to update the position of p agnet The elastic force of the spring S is used to drive the movement of the external particles connected to it, and the multi-mass-spring system (MMS) of the external particle mass points is calculated oa Update the position of the external particles and update the fruit model Mesh; out ​ Loop through the entire simulation step C until the set number of loops is reached; D Output: Output the results of the simulation in each loop of step C, including the updated hyphal model, fruit model texture and mesh; E Rendering: Render the results output in step D to obtain the simulation result image, and synthesize the disease process animation with the result image at 25 frames per second.

2. The virtual simulation method for the symptoms of grape fruit mold diseases as described in claim 1, characterized in that the steps In B1: Combine the Perlin noise map with Poisson disk sampling to make the hyphal distribution more natural; In step B2: To increase hyphal diversity, establish multiple template models for each type of mold, and select a fixed model for each hypha during preprocessing. Set several random variables for each hypha to increase diversity; In step B3: First voxelize the Mesh to obtain the voxel grid of the fruit, and then fill the particles in a hexagonal arrangement through the seed filling algorithm; In step B4: B4.1: Add bending springs between external particles; In step C0, to ensure stability, use an implicit integration method to calculate the diffusion model; C3.2.2 Calculate the acceleration according to the particle mass and force, and then update the internal particle velocity and position, and calculate this process using leapfrog integration.

Images