Method and device for 3d reconstruction of vocs fluid based on neural fluid field

Abstract

The application discloses a VOCs fluid three-dimensional reconstruction method and device based on a neural fluid field, and belongs to the technical field of computers. The VOCs fluid three-dimensional reconstruction method based on the neural fluid field comprises the following steps: inputting a sample fluid image corresponding to petrochemical gas in a target space in a sample period and a sample label corresponding to the sample fluid image into a neural fluid field model to obtain a first fluid image corresponding to the petrochemical gas output by the neural fluid field model; constructing a target loss function based on at least one of the first fluid image, the sample fluid image and the sample label; and training the neural fluid field model based on the target loss function. The VOCs fluid three-dimensional reconstruction method based on the neural fluid field can recover a velocity field and a density field of VOCs fluid from multi-view video data, is suitable for processing VOCs fluid lacking stable visual features, can better support real-time monitoring and three-dimensional reconstruction and the like applications, and has high detection precision.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of computers, and particularly relates to a VOCs fluid three-dimensional reconstruction method and device based on a neural fluid field. BACKGROUND

[0002] Petrochemical volatile organic compounds (VOCs) are a kind of gas with wide sources and high pollution, which has an important influence on regional atmospheric ozone pollution and PM2.5 pollution, and needs to be monitored in real time so as to timely process the gas. In the related art, there is a method for reconstructing VOCs fluid by using optical flow technology. Since the VOCs fluid usually has no fixed shape and stable visual features, the method is difficult to accurately estimate the density and velocity of the fluid, so that the VOCs gas is difficult to be three-dimensionally reconstructed, the detection accuracy is poor, and it is difficult to be applied to real-time monitoring of VOCs. SUMMARY

[0003] The present application aims to at least solve one of the technical problems in the related art. To this end, the present application provides a VOCs fluid three-dimensional reconstruction method and device based on a neural fluid field. The neural fluid field model can recover the velocity field and density field of the VOCs fluid from multi-view video data, is suitable for processing VOCs fluid lacking stable visual features, can better support real-time monitoring and three-dimensional reconstruction applications, and has high detection accuracy.

[0004] In a first aspect, the present application provides a VOCs fluid three-dimensional reconstruction method based on a neural fluid field, which comprises:

[0005] inputting a sample fluid image corresponding to a petrochemical gas in a target space in a sample period and a sample label corresponding to the sample fluid image into a neural fluid field model, to obtain a first fluid image of the petrochemical gas output by the neural fluid field model; the sample label comprises at least one of an actual density field and an actual velocity field of the petrochemical gas in the target space;

[0006] constructing a target loss function based on at least one of the first fluid image, the sample fluid image and the sample label;

[0007] training the neural fluid field model based on the target loss function; the neural fluid field model is used to reconstruct at least one of a target density field and a target velocity field of the petrochemical gas in the target space, and the at least one of the target density field and the target velocity field is used to three-dimensionally reconstruct a fluid corresponding to the petrochemical gas.

[0008] According to the VOCs fluid three-dimensional reconstruction method based on the neural fluid field provided in the embodiments of the present application, the density field and the velocity field of the petrochemical gas in the sample fluid image are recovered by constructing a neural fluid field model to obtain a first fluid image, and a target loss function is constructed based on at least one of the first fluid image, the sample fluid image and a sample label corresponding to the sample fluid image, and then the neural fluid field model is trained based on the target loss function to improve the accuracy and precision of the neural fluid field model, so that the neural fluid field model can recover the velocity field and the density field of the VOCs fluid from multi-view video data, is suitable for processing VOCs fluid lacking stable visual features, and can better support real-time monitoring and three-dimensional reconstruction and other applications, and has high detection accuracy.

[0009] The VOCs fluid three-dimensional reconstruction method based on the neural fluid field of one embodiment of the present application, the target loss function is constructed based on at least one of the first fluid image, the sample fluid image and the sample label, including:

[0010] At least one of a density loss function, a projection loss function, a laminar flow regularization loss function and a rendering loss function is constructed based on at least one of the first fluid image, the sample fluid image and the sample label;

[0011] At least one of the density loss function, the projection loss function, the laminar flow regularization loss function and the rendering loss function is fused to obtain the target loss function.

[0012] The VOCs fluid three-dimensional reconstruction method based on the neural fluid field of one embodiment of the present application, the target loss function is constructed based on at least one of the first fluid image, the sample fluid image and the sample label, including:

[0013] The density loss function is constructed based on the product of the diffusion rate of the density of the petrochemical gas along a target dimension in multiple spatial dimensions and the velocity of the petrochemical gas along the target dimension, and the density variation rate of the petrochemical gas;

[0014] The projection loss function is constructed based on the difference between the actual velocity field and the divergence corresponding to the actual velocity field;

[0015] The laminar flow regularization loss function is constructed based on the actual velocity field and the actual density field;

[0016] The rendering loss function is constructed based on the first fluid image and the sample fluid image.

[0017] The VOCs fluid three-dimensional reconstruction method based on the neural fluid field of one embodiment of the application is based on the actual velocity field and the actual density field, and the laminar flow regularization loss function is constructed, including:

[0018] The hinge loss function is constructed based on the actual velocity field and the actual density field.

[0019] The laminar flow regularization loss function is constructed based on the hinge loss function.

[0020] The VOCs fluid three-dimensional reconstruction method based on the neural fluid field of one embodiment of the application is based on the first fluid image and the sample fluid image, and the rendering loss function is constructed, including:

[0021] The rendering brightness of the first fluid image under the target camera position and the target ray direction is obtained, and the observation brightness of the sample fluid image under the target camera position and the target ray direction is obtained;

[0022] The rendering loss function is constructed based on the difference between the rendering brightness and the observation brightness.

[0023] The VOCs fluid three-dimensional reconstruction method based on the neural fluid field of one embodiment of the application fuses at least one of the density loss function, the projection loss function, the laminar flow regularization loss function and the rendering loss function to obtain the target loss function, including:

[0024] The loss weight corresponding to the density loss function, the loss weight corresponding to the projection loss function, the loss weight corresponding to the laminar flow regularization loss function and the loss weight corresponding to the rendering loss function are obtained.

[0025] The density loss function, the projection loss function, the laminar flow regularization loss function and the rendering loss function are weighted and summed based on the loss weight corresponding to the density loss function, the loss weight corresponding to the projection loss function, the loss weight corresponding to the laminar flow regularization loss function and the loss weight corresponding to the rendering loss function to obtain the target loss function.

[0026] The VOCs fluid three-dimensional reconstruction method based on the neural fluid field of one embodiment of the application is based on the following steps to construct the neural fluid field model:

[0027] The base neural velocity field is constructed based on the neural graph primitive, and the residual turbulent velocity field driven by the vortex particle is constructed based on the target motion equation.

[0028] The neural fluid field model is constructed based on the base neural velocity field and the residual turbulent velocity field driven by the vortex particle.

[0029] The VOCs fluid three-dimensional reconstruction method based on the neural fluid field of one embodiment of the application comprises the following steps:

[0030] An initial fluid image of the petrochemical gas in the target space corresponding to a target time period is obtained.

[0031] The initial fluid image is input into the trained neural fluid field model, and a target fluid image of the petrochemical gas output by the trained neural fluid field model is obtained.

[0032] In a second aspect, the application provides a VOCs fluid three-dimensional reconstruction device based on a neural fluid field, comprising:

[0033] A first processing module is configured to input a sample fluid image of a petrochemical gas in a target space corresponding to a sample time period and a sample label corresponding to the sample fluid image into a neural fluid field model, to obtain a first fluid image of the petrochemical gas output by the neural fluid field model; the sample label comprises at least one of an actual density field and an actual velocity field of the petrochemical gas in the target space.

[0034] A second processing module is configured to construct a target loss function based on at least one of the first fluid image, the sample fluid image and the sample label.

[0035] A third processing module is configured to train the neural fluid field model based on the target loss function; the neural fluid field model is used to reconstruct at least one of a target density field and a target velocity field of the petrochemical gas in the target space, and the at least one of the target density field and the target velocity field is used to perform three-dimensional reconstruction on a fluid corresponding to the petrochemical gas.

[0036] The VOCs fluid three-dimensional reconstruction device based on the neural fluid field provided by the embodiments of the application restores the density field and the velocity field of the petrochemical gas in the sample fluid image by constructing the neural fluid field model to obtain the first fluid image, constructs the target loss function based on at least one of the first fluid image, the sample fluid image and the sample label corresponding to the sample fluid image, trains the neural fluid field model based on the target loss function, improves the accuracy and the precision of the neural fluid field model, and enables the neural fluid field model to restore the velocity field and the density field of the VOCs fluid from the multi-view video data, is suitable for processing the VOCs fluid lacking stable visual features, and can better support real-time monitoring and three-dimensional reconstruction and other applications, and has high detection precision.

[0037] In a third aspect, the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for three-dimensional reconstruction of VOCs fluid based on neural fluid field according to the first aspect.

[0038] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium, having a computer program stored thereon, wherein the computer program is executed by a processor to implement the method for three-dimensional reconstruction of VOCs fluid based on neural fluid field according to the first aspect.

[0039] In a fifth aspect, the present application provides a computer program product, comprising a computer program, wherein the computer program is executed by a processor to implement the method for three-dimensional reconstruction of VOCs fluid based on neural fluid field according to the first aspect.

[0040] The one or more technical solutions described above in the embodiments of the present application have at least one of the following technical effects:

[0041] By constructing a neural fluid field model to restore the density field and velocity field of petrochemical gas in the sample fluid image to obtain a first fluid image, and constructing a target loss function based on at least one of the first fluid image, the sample fluid image, and the sample label corresponding to the sample fluid image, and training the neural fluid field model based on the target loss function, the accuracy and precision of the neural fluid field model are improved, so that the neural fluid field model can restore the velocity field and density field of VOCs fluid from multi-view video data, is suitable for processing VOCs fluid lacking stable visual features, and can better support real-time monitoring and three-dimensional reconstruction and other applications with high detection accuracy.

[0042] Further, by constructing physical losses, the velocity field inferred by the neural fluid field model is divergence-free and can drive the transmission of the density field; in addition, the density field is regularized based on differentiable volume rendering and physical fluid transmission constraints to improve the accuracy of the neural fluid field model in recovering continuous 3D density field from 2D video data, and the constructed multiple physical losses are weighted and combined, so that the parameters of the neural fluid field model can be jointly optimized based on all loss terms, so that the fluid density and velocity field recovered from the video not only conform to the physical law, but also are consistent with the observed visual data, thereby improving the prediction accuracy and precision of the neural fluid field model.

[0043] Further, by adopting the hybrid neural velocity representation, the VOCs fluid velocity field is decomposed into a base neural velocity field and a vortex particle driven residual turbulent velocity field, the large-scale flow characteristics of the VOCs fluid can be obtained using the base neural velocity field, and the small-scale vortex characteristics of the VOCs fluid can be obtained based on the vortex particle driven residual turbulent velocity field, so that the neural fluid field model can better obtain the turbulent characteristics of the fluid velocity.

[0044] Further, by constructing and training the neural fluid field model, and using the neural fluid field model to recover the fluid density and velocity field of the VOCs gas from the video data, the three-dimensional reconstruction of the VOCs fluid smoke is realized, which can be used for detecting the leakage of VOCs and tracing, etc., and the detection precision and accuracy are higher.

[0045] Additional aspects and advantages of the application will be set forth in part in the description that follows, and in part will become apparent to those skilled in the art upon examination of the following description and drawings or can be learned by practice of the application. BRIEF DESCRIPTION OF DRAWINGS

[0046] The above and / or additional aspects and advantages of the application will become apparent and be readily understood by considering the following detailed description, from which the above-mentioned aspects and advantages will become apparent and be readily understood, in which:

[0047] Figure 1 is a flow diagram of a VOCs fluid three-dimensional reconstruction method based on a neural fluid field provided by an embodiment of the application;

[0048] Figure 2 is a principle diagram of a VOCs fluid three-dimensional reconstruction method based on a neural fluid field provided by an embodiment of the application;

[0049] Figure 3 is a result diagram of a VOCs fluid three-dimensional reconstruction method based on a neural fluid field provided by an embodiment of the application;

[0050] Figure 4 is a structure diagram of a VOCs fluid three-dimensional reconstruction device based on a neural fluid field provided by an embodiment of the application;

[0051] Figure 5 is a structure diagram of an electronic device provided by an embodiment of the application. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the application will be described clearly below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are some of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by a person of ordinary skill in the art belong to the scope of protection of the application.

[0053] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0054] The following description, in conjunction with the accompanying drawings, details the VOCs fluid three-dimensional reconstruction method based on neural fluid fields, the VOCs fluid three-dimensional reconstruction device based on neural fluid fields, the electronic device, and the readable storage medium provided in this application, through specific embodiments and application scenarios.

[0055] Among them, the VOCs fluid three-dimensional reconstruction method based on neural fluid fields can be applied to terminals, specifically executed by hardware or software in the terminal.

[0056] The terminal includes, but is not limited to, portable communication devices such as mobile phones or tablets with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads). It should also be understood that, in some embodiments, the terminal may not be a portable communication device, but rather a desktop computer with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads).

[0057] The following embodiments describe a terminal including a display and a touch-sensitive surface. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, mouse, and joystick.

[0058] The VOCs fluid three-dimensional reconstruction method based on neural fluid field provided in this application embodiment can be executed by an electronic device or a functional module or functional entity in an electronic device that can implement the VOCs fluid three-dimensional reconstruction method based on neural fluid field. The electronic devices mentioned in this application embodiment include, but are not limited to, mobile phones, tablets, computers, cameras and wearable devices. The following uses an electronic device as the execution subject to illustrate the VOCs fluid three-dimensional reconstruction method based on neural fluid field provided in this application embodiment.

[0059] like Figure 1 As shown, the three-dimensional reconstruction method of VOCs fluid based on neural fluid field includes steps 110, 120 and 130.

[0060] In step 110, the sample fluid image of the petrochemical gas in the target space and the sample label corresponding to the sample fluid image are input into the neural fluid field model, and a first fluid image of the petrochemical gas output by the neural fluid field model is obtained.

[0061] In this step, the petrochemical gas is petrochemical volatile organic compounds (VOCs).

[0062] The sample fluid image can include a fluid video corresponding to VOCs collected at at least one collection angle, and the fluid video includes multiple video frames.

[0063] The sample fluid image can include actually collected images or videos, and can also include simulated images or videos corresponding to the petrochemical gas.

[0064] For example, multiple fixed-position cameras can be arranged in the target space to obtain multi-angle video data of the same scene.

[0065] The sample period is a collection period corresponding to the sample fluid image, for example, video data of the petrochemical gas from time A to time B can be collected.

[0066] The sample fluid image is used to represent the flow state of the petrochemical gas in the sample period.

[0067] The sample label includes at least one of the actual density field and the actual velocity field of the petrochemical gas in the target space.

[0068] The actual density field is used to represent the density distribution of the VOCs fluid in the target space, and the actual density field can include the rate of change of the density of the VOCs fluid with time and the diffusion rate of the density along multiple spatial dimensions.

[0069] The actual velocity field is used to represent the flow direction and speed of the VOCs fluid, and the actual velocity field can include the velocity vector of the fluid at each point in space.

[0070] The neural fluid field model is used to recover the density field and velocity field of the VOCs fluid from the sample fluid image, and then render the density field and velocity field of the VOCs fluid to the first fluid image.

[0071] The neural fluid field model can be trained end-to-end to ensure that the neural fluid field model can recover the density field and velocity field of the VOCs fluid from multi-angle video data.

[0072] In actual execution, multi-angle video data of the same scene can be captured by multiple fixed-position cameras, and real data is used to evaluate the simulation.

[0073] The real dataset can be obtained based on the ScalarFlow dataset, and the real data of the first five scenes can be determined as the real dataset. The real dataset in each scene includes video data collected from five collection perspectives. The five cameras are fixed in position during the entire collection process and are uniformly distributed on a 120° arc centered on the rising smoke.

[0074] The frame rate of each video data is 150 frames, the resolution is 1062*600, and the video data has been processed to remove background information. For each scene, four video data can be used for training, and one video data can be used for testing.

[0075] Synthetic simulation can also be used to evaluate 3D velocity and density. Synthetic data can be generated by generating code from the ScalarFlow synthetic dataset, including scenes with different inflow sources and different viscosities. For example, five scenes with different inflow sources and high viscosity and another five scenes with low viscosity can be generated for model training.

[0076] Step 120, constructing a target loss function based on at least one of the first fluid image, the sample fluid image, and the sample label;

[0077] In this step, the target loss function is used to represent the difference between the output of the neural fluid field model and the actual fluid state.

[0078] The target loss function can be used to constrain the density field and velocity field output by the neural fluid field model to be physically reasonable.

[0079] In actual implementation, different sub-loss functions can be constructed based on different physical quantities. For example, a sub-loss function can be constructed based on density to constrain the density field output by the model, and a sub-loss function can be constructed based on velocity to constrain the velocity field output by the model. The target loss function can be obtained by fusing multiple sub-loss functions.

[0080] Step 130, training the neural fluid field model based on the target loss function.

[0081] In this step, the neural fluid field model is used to reconstruct at least one of the target density field and the target velocity field of the petrochemical gas in the target space.

[0082] The neural fluid field model can recover the density field and velocity field of the petrochemical gas from the image data corresponding to the petrochemical gas.

[0083] After reconstructing the target density field and the target velocity field of the petrochemical gas in the target space, the VOCs fluid can be reconstructed in three dimensions, thereby realizing 3D visualization of the VOCs fluid.

[0084] The neural fluid field model can also be applied to fluid simulation, editing, and future prediction.

[0085] The neural fluid field model can predict the future state of the fluid based on historical data and can be applied to environmental pollution warning and industrial process control.

[0086] During the training of the neural fluid field model, the target loss function can update the corresponding model parameters in the neural fluid field model based on the loss value between the predicted value and the true value, to reduce the loss between the predicted value and the true value, so that the output result of the neural fluid field model is more accurate and accurate.

[0087] According to the VOCs fluid three-dimensional reconstruction method based on the neural fluid field provided in the embodiments of the present application, the density field and the velocity field of the petrochemical gas in the sample fluid image are recovered by constructing a neural fluid field model to obtain a first fluid image. At least one of the first fluid image, the sample fluid image, and the sample label corresponding to the sample fluid image is used to construct a target loss function, and the neural fluid field model is trained based on the target loss function to improve the accuracy and accuracy of the neural fluid field model. The neural fluid field model can recover the velocity field and the density field of the VOCs fluid from the multi-view video data, which is suitable for processing VOCs fluid lacking stable visual features, and can better support real-time monitoring and three-dimensional reconstruction applications, and has high detection accuracy.

[0088] In some embodiments, step 120 can include:

[0089] At least one of a density loss function, a projection loss function, a laminar flow regularization loss function, and a rendering loss function is constructed based on at least one of the first fluid image, the sample fluid image, and the sample label;

[0090] At least one of the density loss function, the projection loss function, the laminar flow regularization loss function, and the rendering loss function is fused to obtain a target loss function.

[0091] In this embodiment, the density loss function can be constructed using the density transport equation to ensure that the velocity field learned by the neural fluid field model can reasonably transport the density field.

[0092] The projection loss function can be constructed using a pressure projection solver to make the velocity field divergence-free, so as to meet the physical properties of incompressible fluid and ensure physical reasonableness.

[0093] The laminar flow regularization loss function can be constructed to improve the velocity reconstruction in laminar flow (a state in which the fluid flows slowly and smoothly), and thus improve the velocity field reconstructed by the neural fluid field model.

[0094] The density field can be regularized based on differentiable volume rendering and physical fluid transport constraints.

[0095] Based on at least one of the first fluid image, the sample fluid image and the sample label, at least one loss function can be constructed, and then at least one loss function can be fused to obtain a target loss function.

[0096] In some embodiments, based on at least one of the first fluid image, the sample fluid image and the sample label, constructing at least one of a density loss function, a projection loss function, a laminar regularization loss function and a rendering loss function can include:

[0097] Based on the product of the diffusivity of the density of the petrochemical gas along a target dimension in the plurality of spatial dimensions and the velocity of the petrochemical gas along the target dimension, and the rate of change of the density of the petrochemical gas, the density loss function is constructed;

[0098] Based on the difference between the actual velocity field and the divergence corresponding to the actual velocity field, the projection loss function is constructed;

[0099] Based on the actual velocity field and the actual density field, the laminar regularization loss function is constructed;

[0100] Based on the first fluid image and the sample fluid image, the rendering loss function is constructed.

[0101] In this embodiment, the target dimension is any spatial dimension in the plurality of spatial dimensions, for example, the plurality of spatial dimensions can include x, y and z.

[0102] The velocity field can include three components, respectively representing the velocity in the x, y and z directions.

[0103] For the target dimension in the plurality of spatial dimensions, the density loss function can be constructed based on the diffusivity of the density of the petrochemical gas along the target dimension and the velocity of the petrochemical gas along the target dimension, and the rate of change of the density of the petrochemical gas.

[0104] In actual execution, the density loss Using the density transport equation, it is ensured that the learned velocity field can reasonably transport the density field. The density transport equation in incompressible flow can be used, The physical information supervision signal is introduced, and the density loss function is as follows:

[0105]

[0106] wherein, is the density loss function, is an expectation operator for averaging the loss over the entire dataset, x, y and z are three spatial dimensions, σ is the density, u, v and w are respectively the velocities of the petrochemical gas in the x, y and z directions, is the rate of change of the density over time (i.e. the rate of change of the density), and respectively are the diffusivity of the density along the three spatial dimensions.

[0107] The density loss indicates that the three-dimensional velocity field (u, v, w) at any given time t should transport the density σ so that it evolves into the density field at the next time.

[0108] The divergence corresponding to the actual velocity field can be calculated by the pressure projection solver, and then the projection loss function can be constructed according to the difference between the actual velocity field and the divergence corresponding to the actual velocity field.

[0109] In actual execution, the projection loss The pressure projection solver can be used to make the velocity field divergence-free, which meets the physical properties of incompressible fluid and ensures physical rationality.

[0110] By using the pressure projection solver, the learned velocity field is constrained to be divergence-free, and the projection loss function is as follows:

[0111]

[0112] wherein, is the projection loss function, the expectation of the spatial coordinates (x, y, z) and time t, that is, the loss is averaged over the entire flow field and time period, u p is the velocity field calculated by the pressure projection solver, which is the projection of the original velocity field u on the divergence-free subspace, and ||*|| 2 is the squared Euclidean norm, which is used to calculate the difference between the actual velocity field and the divergence corresponding to the actual velocity field.

[0113] In some embodiments, based on the actual velocity field and the actual density field, a laminar regularization loss function can be constructed, which can include:

[0114] Based on the actual velocity field and the actual density field, a hinge loss function is constructed;

[0115] Based on the hinge loss function, a laminar regularization loss function is constructed.

[0116] In this embodiment, the hinge loss function will affect the laminar regularization loss function when the actual velocity field and the actual density field meet the conditions.

[0117] The laminar regularization loss function can cause the neural fluid field model to output a non-zero velocity field in the high-density area, thereby improving the velocity reconstruction in the laminar area.

[0118] In actual execution, the laminar regularization loss function can be constructed based on the following formula:

[0119]

[0120] wherein, is the laminar regularization loss function, is the expectation over spatial coordinates (x, y, z) and time t, i.e., the loss is averaged over the entire flow field and time period, max(0, γσ-||u||) is the hinge loss function, which contributes to the laminar regularization loss when γσ-||u||>0, i.e., the loss function only increases when the magnitude of the velocity field u is smaller than a certain threshold determined by γ and the density σ.

[0121] γ is a hyperparameter for scaling the threshold according to the unit of velocity magnitude, which can be adjusted according to actual conditions to balance the accuracy of the velocity field and the strength of regularization.

[0122] σ is the density, which represents the mass distribution of the fluid, and the partial derivative of the density may be zero in laminar flow, but the absolute value of the density can be non-zero in areas where there is a difference in mass distribution.

[0123] ||u|| is the modulus (or magnitude) of the velocity field u, i.e., the absolute value of the velocity, which represents the speed of fluid motion.

[0124] In some embodiments, based on the first fluid image and the sample fluid image, the rendering loss function can be constructed, which can include:

[0125] obtaining the rendering brightness of the first fluid image at the target camera position and the target ray direction, and the observed brightness of the sample fluid image at the target camera position and the target ray direction;

[0126] based on the difference between the rendering brightness and the observed brightness, constructing the rendering loss function.

[0127] In this embodiment, the target camera position and the target ray direction can be determined based on the user.

[0128] The rendering brightness is used to represent the luminosity rendered at the target camera position and the target ray direction, and the observed brightness is the luminosity observed at the target camera position and the target ray direction.

[0129] The difference between the rendering brightness and the observed brightness can be calculated based on a structural similarity algorithm or a squared Euclidean norm, or other algorithms can also be used, which can be selected based on user needs, and the present application is not limited.

[0130] In actual execution, the density, appearance and velocity can be learned together, and based on the visual imaging signal and the physical fluid transport constraint, the rendering loss function shown in the following formula can be constructed:

[0131]

[0132] where, is a rendering loss function that measures the difference between the rendered image (first fluid image) and the observed (real) image (sample fluid image), is an expectation operator that averages the loss over the dataset, o is the camera position, d is the ray direction, t is the time, L render (o,d) is the radiance (brightness) rendered at camera position o and direction d, L observe (o,d) is the radiance (brightness) observed at camera position o and direction d, ||*|| 2 is the squared Euclidean norm that computes the difference between two radiance.

[0133] In some embodiments, fusing at least one of the density loss function, the projection loss function, the laminar regularization loss function, and the rendering loss function to obtain a target loss function can include:

[0134] obtaining a loss weight corresponding to the density loss function, a loss weight corresponding to the projection loss function, a loss weight corresponding to the laminar regularization loss function, and a loss weight corresponding to the rendering loss function;

[0135] performing weighted sum processing on the density loss function, the projection loss function, the laminar regularization loss function, and the rendering loss function based on the loss weight corresponding to the density loss function, the loss weight corresponding to the projection loss function, the loss weight corresponding to the laminar regularization loss function, and the loss weight corresponding to the rendering loss function to obtain the target loss function.

[0136] In this embodiment, the target loss function can be designed as In combination with the rendering loss function, the density loss function, the projection loss function, and the laminar regularization loss function, joint optimization is performed.

[0137] A plurality of weight combination schemes can be set, and then the performance of the neural fluid field model under each weight combination scheme is evaluated based on the training set and the validation set, and the weight combination scheme is selected based on the validation result.

[0138] Alternatively, the weights corresponding to each loss function can also be optimized based on other manners, which are not limited by the present application.

[0139] In actual execution, the target loss function can be constructed based on the following formula:

[0140]

[0141] where, is the target loss function, is a rendering loss function, βrender is a loss weight corresponding to the density loss function, is a density loss function, β density is a loss weight corresponding to the density loss function, is a projection loss function, β proj is a loss weight corresponding to the projection loss function, is a laminar regularization loss function, β laminar is a weight corresponding to the laminar regularization loss function.

[0142] The density loss function, the projection loss function, the laminar regularization loss function and the rendering loss function can jointly optimize the model parameters to ensure that the fluid density and velocity field recovered from the video not only conform to the physical law, but also consistent with the observed visual data.

[0143] According to the VOCs fluid three-dimensional reconstruction method based on the neural fluid field provided in the embodiments of the present application, by constructing physical losses, the velocity field inferred by the neural fluid field model is divergence-free and can drive the transmission of the density field. In addition, the density field is regularized based on the differentiable volume rendering and the physical fluid transmission constraint, the accuracy of the neural fluid field model in recovering continuous 3D density field from 2D video data is improved, the constructed multiple physical losses are weighted and combined, the parameters of the neural fluid field model can be jointly optimized based on all loss terms, so that the fluid density and velocity field recovered from the video not only conform to the physical law, but also consistent with the observed visual data, the prediction accuracy and accuracy of the neural fluid field model are improved.

[0144] As shown in FIG. 1, in some embodiments, the neural fluid field model can be constructed based on the following steps: Figure 2

[0145] Construct a base neural velocity field based on the Instant Neural Graphics Primitives, and construct a residual turbulent velocity field driven by vortex particles based on a target motion equation.

[0146] Construct the neural fluid field model based on the base neural velocity field and the residual turbulent velocity field driven by vortex particles.

[0147] In this embodiment, the Instant Neural Graphics Primitives is a graphics basis that uses neural networks as basic building blocks to express and render the surface of a scene through mathematical functions.

[0148] The base neural velocity field can be used to obtain large-scale flow motion of the VOCs fluid.

[0149] The target motion equation can include the Navier-Stokes equation (N-S equation), which is a motion equation representing the momentum conservation of viscous incompressible fluid.​

[0150] The vortex particle driven residual turbulent velocity field can be used to obtain small-scale vortex details of the VOCs fluid.

[0151] That is, the underlying flow field can be decomposed into a base neural velocity field and a vortex particle driven residual turbulent velocity field.

[0152] In actual implementation, a hybrid neural velocity representation can be designed, including a base neural velocity field and a vortex particle driven residual turbulent velocity field, to capture the turbulent characteristics of the VOCs fluid velocity. Specifically, the following steps are included:

[0153] S1: decompose the underlying flow field u = u base + u vort into a base neural velocity field u base and a residual vorticity driven velocity field u vort . The base neural velocity field obtains large-scale flow motion, and the vorticity driven velocity represents small-scale vortex details.

[0154] S2: create a base neural velocity field (u base ), extended to the time domain using instance neural graph primitives (iNGP), to obtain large-scale flow.

[0155] S3: create a vortex particle driven residual turbulent velocity field (u vort ), which uses a particle-based method to represent small-scale vortex details. Wherein u vort complements the base flow by utilizing the physical structure of turbulence. The physical model behind the vorticity driven flow can be represented by the curl form of the Navier-Stokes equation:

[0156]

[0157] where, is the vorticity, u is the velocity field, v is the dynamic viscosity, and f is the external body force. In the case of incompressible fluid, only the vortex stretching term is not zero, and the equation becomes the vorticity transport equation, which represents the stretching and advection of vorticity by the velocity flow.

[0158] In the vortex particle method, a particle-based vorticity representation can be used, which is low-dimensional and easy to evolve in time, and can be embedded in the cyclic physical flow structure.

[0159] The vortex particle induced vorticity driven u vort (x, t) can be represented as:

[0160]

[0161] where u vort(x, t) is a residual turbulent velocity field driven by vortex particles, N p is a unit vector from the particle position to x, is the adjusted vorticity, intensity I p , position and vorticity The triplet of (x, t) represents a vortex particle p at a certain timestamp t.

[0162] K is a Gaussian distribution used to smooth the data, where r is the width parameter of the kernel.

[0163] S4: It can be assumed that most of the energy is captured by the base flow, and only the base flow is used to transmit the vortex particles. First, the base flow can be learned, and the base flow is fixed to learn the residual flow. Second, the learning parameters can be obtained from the particle transmission based on the seeding strategy, enough vortex particles are seeded at high curling spatiotemporal positions, and the trajectories of all t and p are pre-calculated through the learned base flow and Then the learnable parameters can be simplified to I p . Redundant vortex particles are automatically suppressed by learning with zero intensity. Through the simplification of step S4, the poor local minimum caused by directly learning the vortex particles can be effectively avoided.

[0164] According to the VOCs fluid three-dimensional reconstruction method based on the neural fluid field provided in the embodiments of the present application, by adopting the hybrid neural velocity representation, the VOCs fluid velocity field is decomposed into a basic neural velocity field and a residual turbulent velocity field driven by vortex particles. The large-scale flow characteristics of the VOCs fluid can be obtained using the basic neural velocity field, and the small-scale vortex characteristics of the VOCs fluid can be obtained based on the residual turbulent velocity field driven by vortex particles, so that the neural fluid field model can better obtain the turbulent characteristics of the fluid velocity.

[0165] As shown in Figure 3 , Figure 3 (d) illustrates a sample fluid image, and the sample fluid image corresponds to a sample label, Figure 3 (a) and Figure 3 (b) illustrate fluid images obtained by processing the sample fluid image based on the related art. The inventors have found through testing that the neural fluid field model provided in the embodiments of the present application can better recover the fluid density and velocity field of the VOCs gas from the sample fluid image, and obtain the fluid image as shown in Figure 3 (c). The neural fluid field model provided in the embodiments of the present application has high accuracy in recovering the density and velocity of the VOCs fluid from the video.

[0166] In some embodiments, after step 130, the method can further include:

[0167] acquire an initial fluid image corresponding to the petrochemical gas in the target space within a target period;

[0168] input the initial fluid image into the trained neural fluid field model to acquire a target fluid image corresponding to the petrochemical gas output by the trained neural fluid field model.

[0169] In this embodiment, the target period is the acquisition period in the actual application process.

[0170] The initial fluid image is the actually acquired multi-view video data.

[0171] The target fluid image is a predicted fluid image output by the neural fluid field model, and the target fluid image is used to represent the density field and velocity field corresponding to the recovered petrochemical gas from the initial fluid image.

[0172] The actually acquired multi-view video data is input into the trained neural fluid field model to acquire the recovered fluid density and velocity field of the VOCs gas output by the neural fluid field model.

[0173] According to the VOCs fluid three-dimensional reconstruction method based on the neural fluid field provided in the embodiments of the present application, the neural fluid field model is constructed and trained, and the fluid density and velocity field of the VOCs gas are recovered from the video data using the neural fluid field model, so that the three-dimensional reconstruction of the VOCs fluid smoke is realized. The VOCs leakage can be detected and traced, and the detection precision and accuracy are high.

[0174] The VOCs fluid three-dimensional reconstruction device based on the neural fluid field provided in the present application is described below. The VOCs fluid three-dimensional reconstruction device based on the neural fluid field described below can be mutually corresponding to the VOCs fluid three-dimensional reconstruction method based on the neural fluid field described above.

[0175] The VOCs fluid three-dimensional reconstruction method based on the neural fluid field provided in the embodiments of the present application can be executed by the VOCs fluid three-dimensional reconstruction device based on the neural fluid field. In the embodiments of the present application, the VOCs fluid three-dimensional reconstruction method based on the neural fluid field is executed by the VOCs fluid three-dimensional reconstruction device based on the neural fluid field, which is taken as an example to illustrate the VOCs fluid three-dimensional reconstruction device based on the neural fluid field provided in the embodiments of the present application.

[0176] The embodiments of the present application also provide a VOCs fluid three-dimensional reconstruction device based on a neural fluid field.

[0177] As shown in Figure 4 The VOCs fluid three-dimensional reconstruction device based on the neural fluid field includes a first processing module 410, a second processing module 420, and a third processing module 430.

[0178] The first processing module 410 is configured to input the obtained sample fluid image of the petrochemical gas in the target space within a sample period and a sample label corresponding to the sample fluid image into a neural fluid field model to obtain a first fluid image of the petrochemical gas output by the neural fluid field model; the sample label includes at least one of an actual density field and an actual velocity field of the petrochemical gas in the target space.

[0179] The second processing module 420 is configured to construct a target loss function based on at least one of the first fluid image, the sample fluid image, and the sample label.

[0180] The third processing module 430 is configured to train the neural fluid field model based on the target loss function; the neural fluid field model is used to reconstruct at least one of a target density field and a target velocity field of the petrochemical gas in the target space.

[0181] According to the VOCs fluid three-dimensional reconstruction device based on the neural fluid field provided in the embodiments of the present application, the density field and the velocity field of the petrochemical gas in the sample fluid image are restored by constructing the neural fluid field model to obtain the first fluid image, and the target loss function is constructed based on at least one of the first fluid image, the sample fluid image, and the sample label corresponding to the sample fluid image, and then the neural fluid field model is trained based on the target loss function to improve the accuracy and the precision of the neural fluid field model, so that the neural fluid field model can restore the velocity field and the density field of the VOCs fluid from the video data of multiple perspectives, is suitable for processing the VOCs fluid lacking stable visual features, and can better support real-time monitoring and three-dimensional reconstruction and other applications, and has high detection precision.

[0182] In some embodiments, the second processing module 420 can be further configured to:

[0183] construct at least one of a density loss function, a projection loss function, a laminar flow regularization loss function, and a rendering loss function based on at least one of the first fluid image, the sample fluid image, and the sample label;

[0184] fuse at least one of the density loss function, the projection loss function, the laminar flow regularization loss function, and the rendering loss function to obtain the target loss function.

[0185] In some embodiments, the second processing module 420 can be further configured to:

[0186] construct the density loss function based on a product of a diffusion rate of the density of the petrochemical gas along a target dimension in multiple spatial dimensions and a velocity of the petrochemical gas along the target dimension, and a rate of change of the density of the petrochemical gas;

[0187] construct the projection loss function based on a difference degree between the actual velocity field and a divergence corresponding to the actual velocity field;

[0188] construct a laminar regularization loss function based on the actual velocity field and the actual density field;

[0189] construct a rendering loss function based on the first fluid image and the sample fluid image.

[0190] In some embodiments, the second processing module 420 can be further configured to:

[0191] construct a hinge loss function based on the actual velocity field and the actual density field;

[0192] construct the laminar regularization loss function based on the hinge loss function.

[0193] In some embodiments, the second processing module 420 can be further configured to:

[0194] obtain a rendering brightness of the first fluid image at a target camera position and a target ray direction, and an observation brightness of the sample fluid image at the target camera position and the target ray direction;

[0195] construct the rendering loss function based on a difference between the rendering brightness and the observation brightness.

[0196] In some embodiments, the second processing module 420 can be further configured to:

[0197] obtain a loss weight corresponding to the density loss function, a loss weight corresponding to the projection loss function, a loss weight corresponding to the laminar regularization loss function, and a loss weight corresponding to the rendering loss function;

[0198] perform weighted sum processing on the density loss function, the projection loss function, the laminar regularization loss function, and the rendering loss function based on the loss weight corresponding to the density loss function, the loss weight corresponding to the projection loss function, the loss weight corresponding to the laminar regularization loss function, and the loss weight corresponding to the rendering loss function, to obtain a target loss function.

[0199] In some embodiments, the VOCs fluid three-dimensional reconstruction device based on the neural fluid field can further include a fourth processing module configured to construct a neural fluid field model based on the following steps:

[0200] construct a base neural velocity field based on a neural graph primitive, and construct a vortex particle driven residual turbulent velocity field based on a target motion equation;

[0201] construct the neural fluid field model based on the base neural velocity field and the vortex particle driven residual turbulent velocity field.

[0202] In some embodiments, the VOCs fluid three-dimensional reconstruction device based on the neural fluid field can further include a fifth processing module configured to, after training the neural fluid field model based on the target loss function, acquire an initial fluid image corresponding to the petrochemical gas in the target space within a target time period.

[0203] input the initial fluid image into the trained neural fluid field model to acquire a target fluid image corresponding to the petrochemical gas output by the trained neural fluid field model.

[0204] The VOCs fluid three-dimensional reconstruction device based on the neural fluid field in the embodiments of the present application can be an electronic device or a component in an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or other devices other than a terminal. For example, the electronic device can be a mobile phone, a tablet computer, a notebook computer, a palm computer, a vehicle-mounted electronic device, a mobile Internet device (MID), an augmented reality (AR) / virtual reality (VR) device, a robot, a wearable device, an ultra-mobile personal computer (UMPC), a netbook, or a personal digital assistant (PDA), and the like, and can also be a server, a network attached storage (NAS), a personal computer (PC), a television (TV), a teller machine, or a self-service machine, and the like, and the embodiments of the present application are not limited specifically.

[0205] The VOCs fluid three-dimensional reconstruction device based on the neural fluid field in the embodiments of the present application can be a device with an operating system. The operating system can be an Android operating system, an IOS operating system, or other possible operating systems, and the embodiments of the present application are not limited specifically.

[0206] The VOCs fluid three-dimensional reconstruction device based on the neural fluid field provided in the embodiments of the present application can realize Figures 1 to 3 The processes realized by the method embodiments are not repeated here to avoid repetition.

[0207] In some embodiments, as Figure 5As shown, the electronic device 500 according to the embodiment of the present application further includes a processor 501, a memory 502, and a computer program stored in the memory 502 and executable on the processor 501. The computer program is executed by the processor 501 to implement each process of the above-mentioned embodiment of the method for reconstructing VOCs fluid in three dimensions based on neural fluid field, and achieve the same technical effects. To avoid repetition, details are not described herein.

[0208] It should be noted that the electronic device in the embodiment of the present application includes the mobile electronic device and the non-mobile electronic device described above.

[0209] On the other hand, the present application further provides a computer program product, which includes a computer program stored in a non-transitory computer readable storage medium. The computer program includes program instructions, and when the program instructions are executed by a computer, the computer can execute each process of the above-mentioned embodiment of the method for reconstructing VOCs fluid in three dimensions based on neural fluid field, and achieve the same technical effects. To avoid repetition, details are not described herein.

[0210] In yet another aspect, the present application further provides a non-transitory computer readable storage medium having a computer program stored thereon. The computer program is executed by a processor to implement each process of the above-mentioned embodiment of the method for reconstructing VOCs fluid in three dimensions based on neural fluid field, and achieve the same technical effects. To avoid repetition, details are not described herein.

[0211] In yet another aspect, the present application further provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is configured to run a program or instructions to implement each process of the above-mentioned embodiment of the method for reconstructing VOCs fluid in three dimensions based on neural fluid field, and achieve the same technical effects. To avoid repetition, details are not described herein.

[0212] It should be understood that the chip mentioned in the embodiment of the present application can also be referred to as a system-level chip, a system chip, a chip system, or a system-on-chip, etc.

[0213] The device embodiments described above are only schematic, and the units shown as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiment of the present application. Those skilled in the art can understand and implement it without creative labor.

[0214] Those skilled in the art can clearly understand the implementation of the various embodiments by means of software and the necessary general hardware platform from the above description of the embodiments, and of course, the embodiments can also be implemented by hardware. Based on such understanding, the above technical solutions or the part that contributes to the related art can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, and the like, and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in the various embodiments or some parts of the embodiments.

[0215] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement for some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for VOCs fluid 3D reconstruction based on neural fluid field, characterized in that, The method comprises the steps of: inputting the sample fluid image corresponding to the petrochemical gas in the target space within the sample period and the sample label corresponding to the sample fluid image into a neural fluid field model, to obtain a first fluid image corresponding to the petrochemical gas output by the neural fluid field model; the sample label comprises at least one of the actual density field and the actual velocity field of the petrochemical gas in the target space; constructing a target loss function based on at least one of the first fluid image, the sample fluid image and the sample label; training the neural fluid field model based on the target loss function; the neural fluid field model is used to reconstruct at least one of the target density field and the target velocity field of the petrochemical gas in the target space, and the at least one of the target density field and the target velocity field is used to perform three-dimensional reconstruction on the fluid corresponding to the petrochemical gas; the neural fluid field model is constructed based on the following steps: constructing a base neural velocity field based on a neural graph primitive, and constructing a vortex particle driven residual turbulent velocity field based on a target motion equation; constructing the neural fluid field model based on the base neural velocity field and the vortex particle driven residual turbulent velocity field.

2. The neural fluid field based VOCs fluid 3D reconstruction method of claim 1, wherein, The method comprises the steps of: constructing at least one of a density loss function, a projection loss function, a laminar flow regularization loss function and a rendering loss function based on at least one of the first fluid image, the sample fluid image and the sample label; fusing at least one of the density loss function, the projection loss function, the laminar flow regularization loss function and the rendering loss function to obtain the target loss function.

3. The neural fluid field based VOCs fluid 3D reconstruction method of claim 2, wherein, The method comprises the steps of: constructing the density loss function based on the product of the diffusion rate of the density of the petrochemical gas along the target dimension in the plurality of spatial dimensions and the velocity of the petrochemical gas along the target dimension, and the density variation rate of the petrochemical gas; constructing the projection loss function based on the difference between the actual velocity field and the divergence corresponding to the actual velocity field; constructing the laminar flow regularization loss function based on the actual velocity field and the actual density field; constructing the rendering loss function based on the first fluid image and the sample fluid image.

4. The neural fluid field based VOCs fluid 3D reconstruction method of claim 3, wherein, The method comprises the steps of: constructing a hinge loss function based on the actual velocity field and the actual density field; constructing the laminar flow regularization loss function based on the hinge loss function.

5. The neural fluid field based VOCs fluid 3D reconstruction method of claim 3, wherein, The method comprises the steps of: obtaining the rendering brightness of the first fluid image under the target camera position and the target ray direction, and the observation brightness of the sample fluid image under the target camera position and the target ray direction; construct the rendering loss function based on a difference between the rendering luminance and the observation luminance.

6. The neural fluid field based VOCs fluid 3D reconstruction method of claim 2, wherein, The fusion of at least one of the density loss function, the projection loss function, the laminar flow regularization loss function and the rendering loss function obtains the target loss function, comprising: obtaining the loss weight corresponding to the density loss function, the loss weight corresponding to the projection loss function, the loss weight corresponding to the laminar flow regularization loss function and the loss weight corresponding to the rendering loss function; Based on the loss weight corresponding to the density loss function, the loss weight corresponding to the projection loss function, the loss weight corresponding to the laminar flow regularization loss function and the loss weight corresponding to the rendering loss function, the density loss function, the projection loss function, the laminar flow regularization loss function and the rendering loss function are weighted and summed to obtain the target loss function.

7. The neural fluid field based VOCs fluid 3D reconstruction method according to any one of claims 1-6, characterized in that, After the neural fluid field model is trained based on the target loss function, the method further comprises: obtaining the initial fluid image corresponding to the petrochemical gas in the target space within the target period; inputting the initial fluid image into the trained neural fluid field model to obtain the target fluid image corresponding to the petrochemical gas output by the trained neural fluid field model.

8. A device for 3D reconstruction of VOCs fluid based on neural fluid field, characterized in that, comprising: The first processing module is configured to input the obtained sample fluid image corresponding to the petrochemical gas in the target space within the sample period and the sample label corresponding to the sample fluid image into the neural fluid field model to obtain the first fluid image corresponding to the petrochemical gas output by the neural fluid field model; the sample label comprises at least one of the actual density field and the actual velocity field of the petrochemical gas in the target space; The second processing module is configured to construct a target loss function based on at least one of the first fluid image, the sample fluid image and the sample label; The third processing module is configured to train the neural fluid field model based on the target loss function; the neural fluid field model is used to reconstruct at least one of the target density field and the target velocity field of the petrochemical gas in the target space, and at least one of the target density field and the target velocity field is used to perform three-dimensional reconstruction on the fluid corresponding to the petrochemical gas; The fourth processing module is configured to construct the neural fluid field model based on the following steps: construct a base neural velocity field based on a neural graph primitive, and construct a vortex particle driven residual turbulent velocity field based on a target motion equation; construct the neural fluid field model based on the base neural velocity field and the vortex particle driven residual turbulent velocity field.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the VOCs fluid three-dimensional reconstruction method based on the neural fluid field as claimed in any one of claims 1-7.

10. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the VOCs fluid three-dimensional reconstruction method based on the neural fluid field as claimed in any one of claims 1-7.

11. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the VOCs fluid three-dimensional reconstruction method based on the neural fluid field as claimed in any one of claims 1-7.

Images