Holographic display layout optimization method based on improved genetic algorithm and neural network

Through improved genetic algorithms and neural network technology, combined with multiple sensors and data analysis, the holographic display layout is optimized in real time, and the problems of low efficiency and insufficient optimization in the existing technology are solved, achieving efficient user interaction experience and visual effect improvement.

CN118447204BActive Publication Date: 2024-11-12JIANGSU SHENGJI CREATIVE TECH CO LTD
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202410637838.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-22
Publication Date
2024-11-12
Estimated Expiration
2044-05-22

AI Technical Summary

Technical Problem

The existing holographic display technology is difficult to optimize the layout in real time to adapt to different display environments and audience needs. The traditional methods are inefficient and cannot reflect audience changes in real time. There is a lack of effective mechanisms to evaluate and optimize holographic display effects, and the algorithm efficiency and optimization quality are insufficient.

Method used

Using improved genetic algorithms and neural network technology, the layout is dynamically adjusted to improve vision by monitoring holographic display environment changes in real time, building parameter data sets, initializing layout configuration, iterative optimization, combining stereo cameras, depth perception, emotion recognition and heart rate monitoring. Effects and interactive experience.

Benefits of technology

It realizes holographic displays instantly respond to audience behavior and environmental changes, greatly improves user interaction experience and visual effects, improves layout optimization speed and efficiency, and provides personalized display content.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118447204B_ABST
    Figure CN118447204B_ABST
Patent Text Reader

Abstract

The present invention discloses a holographic display layout optimization method based on an improved genetic algorithm and a neural network, including S1, constructing a holographic display environment parameter data set; S2, initializing a set of layout configurations of holographic display elements through an improved genetic algorithm; S3, using a neural network model to evaluate each layout configuration generated in step S2; S4, adjusting the layout configuration in the improved genetic algorithm according to the evaluation result of the neural network model, iteratively optimizing to gradually improve the visual effect and interactive experience of the layout scheme; S5, repeating steps S3 and S4 until a preset number of iterations is reached or the performance index of layout optimization is met; S6, outputting the final optimized holographic display layout configuration for use in actual holographic display. The present invention greatly improves the quality of user interactive experience and visual effects by real-time monitoring and analyzing changes in the holographic display environment.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of holographic display, and in particular to a holographic display layout optimization method based on improved genetic algorithm and neural network. Background Art

[0002] In modern holographic display technology, how to effectively optimize the layout of holographic images to adapt to different display environments and audience needs is a key technical challenge. Traditional holographic display layout methods mainly rely on manual settings or simple calculation models to adjust the position and direction of display elements. This method is not only time-consuming and labor-intensive, but also difficult to achieve accurate and dynamic optimization.

[0003] First, traditional methods show obvious deficiencies in dealing with complex and changing display environments. Due to the lack of real-time response capabilities to environmental variables such as the number of audience members, audience positions, ambient light, and sound levels, these methods often fail to automatically adjust the layout to maximize the visual effect and audience interaction experience. For example, when the audience position changes, traditional methods require manual re-evaluation and adjustment of the layout, which is not only inefficient but also often fails to instantly reflect the audience's latest needs.

[0004] Secondly, existing technologies lack an effective mechanism to evaluate and optimize the effect of holographic display layout in real time. Although some existing systems try to improve the layout through basic image processing technology, these technologies can usually only provide static optimization suggestions and cannot capture the dynamic changes caused by audience interaction. In addition, these systems usually cannot fully consider the interactivity of holographic displays, resulting in a gap between the final display effect and the actual experience of the audience.

[0005] In addition, existing layout optimization methods also have limitations in algorithm efficiency and optimization quality. Although traditional optimization algorithms such as genetic algorithms perform well in some application scenarios, in the layout optimization of holographic displays, due to the lack of targeted improvements and adaptive adjustments, these algorithms often take a long time to converge to an acceptable solution. At the same time, these algorithms also lack effective integration with advanced data analysis technologies such as neural networks, which makes them perform unsatisfactory when processing complex data.

[0006] Therefore, how to provide a holographic display layout optimization method based on an improved genetic algorithm and a neural network is an urgent problem that needs to be solved by those skilled in the art. Summary of the invention

[0007] One object of the present invention is to propose a holographic display layout optimization method based on an improved genetic algorithm and a neural network. The present invention monitors and analyzes changes in the holographic display environment in real time, and dynamically adjusts the layout of holographic display elements based on these data, so that the holographic display can instantly respond to the audience's behavior and changes in the display environment, thereby greatly improving the quality of user interaction experience and visual effects.

[0008] A holographic display layout optimization method based on an improved genetic algorithm and a neural network according to an embodiment of the present invention comprises the following steps:

[0009] S1. Using input devices to receive the number of audience members, dynamic position changes of audience members, real-time environmental sound levels, changes in ambient light intensity, and real-time audience emotional feedback about the holographic display environment, and construct a holographic display environment parameter data set;

[0010] S2. Initializing a layout configuration of a set of holographic display elements through an improved genetic algorithm according to the received holographic display environment parameter data set;

[0011] S3, using a neural network model to evaluate each layout configuration generated in step S2, wherein the neural network model is pre-trained and can evaluate the effect of the layout configuration based on the holographic display effect and audience interaction data;

[0012] S4. According to the evaluation results of the neural network model, adjust the layout configuration in the improved genetic algorithm, and iterate and optimize to gradually improve the visual effect and interactive experience of the layout solution;

[0013] S5, repeating step S3 and step S4 until a preset number of iterations is reached or the performance index of layout optimization is met;

[0014] S6. Output the final optimized holographic display layout configuration for use in actual holographic display.

[0015] Optionally, S1 includes the following steps:

[0016] S11. Deploy stereo cameras at key locations in the holographic display area. The stereo cameras use depth perception technology to capture the audience's three-dimensional spatial position and facial expressions in real time, and provide the audience's dynamic position data D(x,y) and facial expression data F(t) by generating depth maps and texture mapping:

[0017]

[0018] Where D(x,y) represents the depth value of the coordinate (x,y) on the image plane, f is the focal length of the camera, B is the baseline distance between the cameras, and d(x,y) is the corresponding disparity value;

[0019]

[0020] Among them, F(t) represents the facial expression feature value at time t, N f is the number of feature points, f i is the measured value of the i-th feature point, μ i and σ i are the mean and standard deviation of the i-th feature point, w i is the weight of the i-th feature point;

[0021] S12. Using the photoelectric volumetric pulse wave sensor integrated in the seats or wearable devices in the holographic display area, the changes in blood flow are analyzed through a comprehensive heart rate monitoring model to monitor the heart rate change data HR(t) of each audience member in real time;

[0022] S13. Continuously collect the sound level and light intensity in the holographic display area through the environmental monitoring system, and use the sensor array to independently measure the sound in multiple frequency bands and spectral regions. and light Then adjust the audio-visual environment of the holographic display;

[0023] S14. Using an emotion recognition algorithm to convert the audience's facial expression data and heart rate change data into an emotion score E(t). The emotion recognition algorithm is based on a large-scale sentiment analysis training set to identify multiple emotional states and quantify the emotional states into actionable values ​​to guide the content and layout adjustment of the holographic display.

[0024] S15. Collect the data from steps S11-S14 to construct a holographic display environment parameter data set:

[0025]

[0026] Optionally, the comprehensive heart rate monitoring model combines direct pulse wave peak counts with the integral information of the photoplethysmogram obtained through blood flow changes to measure heart rate:

[0027]

[0028] Where HR(t) represents the heart rate at time t, α is a weight parameter used to balance the contribution of the two measurement methods, and N p is the number of pulse wave peaks detected within the time interval Δt, K is a constant used to adjust the dimension of A(t) to match the heart rate unit, and A(t) represents the value calculated based on the integral of the light absorption change caused by blood flow within a certain time window Δt:

[0029]

[0030] Among them, Imax (t) and I min (t) are the maximum and minimum light intensity values ​​measured by the photoplethysmography sensor within period t, reflecting the changes in blood flow.

[0031] Optionally, for sound, the average sound pressure level of each frequency band is measured in multiple frequency bands and calculated, and the sound band is k different frequency bands {f1,f2,...,f k}, the sound level of each frequency band is expressed as:

[0032]

[0033] in, is the sound level of the ith frequency band at time t, is the sound pressure level in this frequency band, T i It is a time window;

[0034] For illumination, measurements are made in different spectral regions, where the spectral region is j different wavelength intervals λ1,λ2,...,λ j}, the light intensity of each interval is expressed as:

[0035]

[0036] in, is the light intensity of the jth spectral interval at time t, L λ (t) is the instantaneous light intensity at wavelength λ, V(λ) is the human visual response function corresponding to the wavelength, λ j,min and λ j,max are the minimum and maximum wavelengths of the jth interval.

[0037] Optionally, the S14 specifically includes:

[0038] S141, using facial expression recognition technology, collect the audience's facial expression data F(t), analyze and identify the position of the eyebrows, the degree of eye opening and closing, and the position of the mouth corners in the expression, and encode each feature point to generate a facial expression feature vector

[0039] S142, real-time monitoring of each audience's heart rate change data HR(t), extracting the average heart rate and heart rate variability and converting the data into a heart rate feature vector

[0040] S143, using the emotion recognition neural network model M to accept facial expression feature vector and the heart rate feature vector As input, the emotion recognition neural network model M is pre-trained to recognize and quantify various emotional states, and the output is the probability distribution P of emotional states.emo,k (t);

[0041] S144. According to the probability distribution P of the emotional state emo,k (t) Combined with the time weighting and emotional state intensity, the comprehensive emotional score E(t) is calculated:

[0042]

[0043] Among them, K is the number of emotional states that model M can recognize, and P emo,k (t) is the recognition probability of the kth emotional state at time t, w k is the weight of the kth emotional state, which is set based on the expected impact of the emotional state on the audience's behavior and interactive experience. k is the duration from the most recent time of the emotional state to the current time, τ k is the time decay constant of the kth emotional state, indicating the duration of the influence of different emotional states, and is used to simulate the natural process of emotional states weakening over time.

[0044] Optionally, the emotional state probability distribution P emo,k (t) Through the multi-layer perceptron model, the multi-layer perceptron model integrates a variety of physiological and behavioral data to predict emotional states. The specific calculation formula is as follows:

[0045] The facial expression feature vector and the heart rate feature vector Perform data fusion to generate a comprehensive feature vector

[0046]

[0047] Among them, ⊕ represents the merging operation of feature vectors;

[0048] The comprehensive feature vector The neural network model M is input into the neural network model. The neural network model M consists of multiple hidden layers. Except for the last layer, which uses the Softmax function to output the probability of each emotional state, each of the remaining layers uses the ReLU activation function. The output of the neural network model M The probability distribution P of each emotional state emo (t), the calculation process is expressed as:

[0049]

[0050] Among them, W i and are the weight matrix and bias vector of the i-th layer, n is the number of hidden layers, the ReLU function is a linear rectification function used to increase the nonlinear ability of the network, and the Softmax function is used to convert the original output value of the output layer into a probability distribution.

[0051] Optionally, S2 includes the following steps:

[0052] S21, based on the received holographic display environment parameter data set Data set Define the initial population, each individual Represents a possible holographic display element layout configuration, including the position coordinates x of each display element i ,y i ,z i and the rotation angle θ i

[0053] S22. Applying adaptive fitness function Each individual is scored, where:

[0054]

[0055] in, It is the evaluation function of visual coverage, which is calculated based on the element position and the audience's perspective. It is an evaluation function of the interaction effect, which is based on the audience's dynamic position and expression parameters to calculate the degree of interaction with the holographic display content. is the resource utilization rate, λ is the coefficient for balancing visual effects and resource utilization, and w1 and w2 are preset weights;

[0056] The evaluation function of visual coverage is calculated based on the element position, the viewer's perspective, and the visual quality of the displayed content:

[0057]

[0058] Where n is the number of holographic elements, m is the number of viewers, and v ij represents the visibility of the i-th holographic element to the j-th viewer, α ij is the sight angle between the jth viewer and the ith holographic element, d ij is the distance between viewer j and holographic element i, σ is the attenuation coefficient, which controls the effect of distance on the visual effect;

[0059] The evaluation function of the interactive effect is based on the audience's dynamic position, expression, and heart rate data to calculate the degree of interaction with the holographic display content:

[0060]

[0061] Where m is the number of viewers, Rk is the response coefficient of audience k, reflecting the interaction willingness of each audience, β, γ, δ are weight factors used to balance the influence of spatial interaction, physiological interaction and emotional interaction;

[0062] S k is the spatial interaction index calculated based on the position and facial expression data of viewer k, and is defined as:

[0063]

[0064] Among them, s ki is the spatial sensitivity of viewer k to holographic element i, L i is the position of element i;

[0065] H k It is a physiological interaction index calculated based on the heart rate change data HR(t):

[0066]

[0067] Among them, HR rest is the resting heart rate, a is the adjustment factor;

[0068] E k is the impact of the sentiment score E(t) on the interaction:

[0069]

[0070] Normalize the sentiment scores between 0 and 1;

[0071] S23. Use the tournament selection method to randomly select individuals from the population, compare the fitness of the individuals, and select the individual with the highest fitness to copy to the next generation:

[0072]

[0073] in, and are two individuals selected at random;

[0074] S24, implement uniform crossover operation to perform equal exchange of genes on the selected fathers. Each gene position determines which father its genetic information comes from. The gene vectors of the two father individuals are defined as and Generate a new individual Its gene position is and Random and equal selection, uniform crossover is expressed as:

[0075]

[0076] Among them, ri is a random number generated for the ith gene position, satisfying r i ~U(0,1), indicating uniform distribution;

[0077] S25. Introduce a gradient-based mutation method to select the parameter x in the gene i ,y i ,z i ,θ i Make adjustments using the local gradient information of the objective function F Guide the mutation process and achieve directional optimization adjustment:

[0078]

[0079] Among them, σ is the parameter that controls the step length of the variable. is The gradient of the objective function calculated at the point, including the partial derivatives for each gene position;

[0080] The gradient is computed using finite differences:

[0081]

[0082] Here, δ is a small positive number used to calculate x i The gradient of the direction, the same method is used for y i ,z i ,θ i Gradient calculation of ;

[0083] S26. Repeat steps S22 to S25 until a population that meets a specific fitness standard is generated.

[0084] Optionally, S3 includes the following steps:

[0085] S31, using a pre-trained neural network model M to evaluate the visual effect and interactive experience of the holographic display element layout configuration, inputting each layout configuration generated in step S2 And from the holographic display environment parameter dataset Data set :

[0086]

[0087] in, Indicates flattening the dynamic position data into a one-dimensional vector. and are the sound level and light intensity vectors, respectively, integrating data from all measured frequencies and wavelengths;

[0088] S32. The structure of the neural network model M includes a multi-layer perceptron, and each layer uses a nonlinear activation function;

[0089] S33, the output layer uses the Softmax function to convert the evaluation results of the neural network into a probability distribution It indicates the suitability of each layout configuration for the holographic display effect and audience interactive experience. The calculation formula is:

[0090]

[0091] Where the exponential function and sum operation ensure that the output is a probability distribution, k is the number of possible layout configurations;

[0092] S34. Configure for each layout Generate a rating Based on probability distribution Converted into numerical values ​​that are easy to optimize, the scoring formula is:

[0093]

[0094] By the probability of each configuration Applying logarithms and negative signs, probabilities are converted to scores so that high probabilities correspond to high scores;

[0095] S35. Scoring of all layout configurations The results are summarized and fed back to the improved genetic algorithm to guide the iterative optimization process of the layout.

[0096] Optionally, S4 includes the following steps:

[0097] S41, receiving all layout configurations of step S35 Rating The ratings reflect the visual and interactive experience of each layout configuration;

[0098] S42. According to the rating Adjusting the layout configuration in the improved genetic algorithm using the fitness function Updated to include evaluation results of neural network models;

[0099] S43. Use updated fitness function Select individuals in the next generation population, using the tournament selection method to ensure that individuals with high fitness have a higher probability of being selected;

[0100] S44, performing uniform crossover and mutation operations of the genetic algorithm on the selected individuals, wherein the mutation operation adjusts the layout parameters according to the gradient information to introduce new genetic mutations;

[0101] S45, repeating steps S42 to S44 until a preset number of iterations is reached or the layout optimization meets a predetermined performance indicator, and the system continuously learns and adapts to optimize the layout configuration of the holographic display;

[0102] S46. After each iteration, evaluate the average fitness of the current population and adjust the crossover rate and mutation rate of the genetic algorithm as needed.

[0103] Optionally, the layout configuration process in the adjusted improved genetic algorithm includes:

[0104] Define the hybrid fitness function Combined with the original fitness of the improved genetic algorithm and the neural network score

[0105]

[0106] Among them, λ is an adjustment coefficient, which is used to balance the weight of traditional fitness evaluation and neural network-based evaluation in genetic algorithms;

[0107] Update the layout configuration selection strategy, introduce a probability distribution adjustment mechanism to respond to score changes, and adjust the score-based Dynamic range, calculate the selection probability of each layout configuration

[0108]

[0109] Among them, κ is a positive parameter used to control the size of the selection pressure, so that the layout configuration with a high score has a higher probability of being selected, thereby accelerating the convergence of the algorithm;

[0110] Introducing a mutation strategy based on dynamic score adjustment to adjust layout parameters in a targeted manner, mutation probability Depending on the current score of the layout configuration, low-scoring configurations receive higher probability of mutation, increasing the chances of exploring new configurations:

[0111]

[0112] Among them, p base is the basic mutation rate, p max is the maximum allowed mutation rate, δ and θ are parameters for adjusting the mutation sensitivity, which are used to adjust the mutation rate according to the comprehensive score of the layout configuration.

[0113] The beneficial effects of the present invention are:

[0114] (1) The present invention monitors and analyzes the changes in the holographic display environment in real time and dynamically adjusts the layout of the holographic display elements based on these data, so that the holographic display can respond to the audience's behavior and the changes in the display environment in real time, greatly improving the quality of user interaction experience and visual effects.

[0115] (2) The present invention combines the improved genetic algorithm and neural network technology, which not only speeds up the optimization speed of the holographic layout configuration, but also improves the efficiency of the optimization process. The improved genetic algorithm accelerates the convergence speed of the holographic layout by introducing a fitness function based on neural network evaluation, making the iterative process more efficient.

[0116] (3) The present invention allows accurate evaluation of the visual effects and interactive experience of the holographic display layout through the application of a neural network model. By feeding back these evaluation results, the improved genetic algorithm can more accurately adjust the layout configuration, thereby providing more personalized display content while ensuring high visual quality and optimizing the user interaction experience.

[0117] (4) The present invention can be widely applied to different holographic display application scenarios, including education, entertainment, advertising, scientific research and other fields. In addition, due to its design flexibility, it can be easily integrated into existing holographic display systems or used as a basic technical platform for new systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0118] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0119] Figure 1 The present invention provides a flow chart of a holographic display layout optimization method based on an improved genetic algorithm and a neural network. DETAILED DESCRIPTION

[0120] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0121] refer to Figure 1 , a holographic display layout optimization method based on improved genetic algorithm and neural network, comprising the following steps:

[0122] S1. Using input devices to receive the number of audience members, dynamic position changes of audience members, real-time environmental sound levels, changes in ambient light intensity, and real-time audience emotional feedback about the holographic display environment, and construct a holographic display environment parameter data set;

[0123] In this implementation, S1 includes the following steps:

[0124] S11. Deploy stereo cameras at key locations in the holographic display area. The stereo cameras use depth perception technology to capture the audience's three-dimensional spatial position and facial expressions in real time, and provide the audience's dynamic position data D(x,y) and facial expression data F(t) by generating depth maps and texture mapping:

[0125]

[0126] Where D(x,y) represents the depth value of the coordinate (x,y) on the image plane, f is the focal length of the camera, B is the baseline distance between the cameras, and d(x,y) is the corresponding disparity value;

[0127]

[0128] Among them, F(t) represents the facial expression feature value at time t, N f is the number of feature points, f i is the measured value of the i-th feature point, μ i and σ i are the mean and standard deviation of the i-th feature point, w i is the weight of the i-th feature point;

[0129] S12. Using the photoelectric volumetric pulse wave sensor integrated in the seats or wearable devices in the holographic display area, the changes in blood flow are analyzed through a comprehensive heart rate monitoring model to monitor the heart rate change data HR(t) of each audience member in real time;

[0130] In this embodiment, the comprehensive heart rate monitoring model combines direct pulse wave peak counts and the integral information of the photoplethysmogram obtained through blood flow changes to measure heart rate:

[0131]

[0132] Where HR(t) represents the heart rate at time t, α is a weight parameter used to balance the contribution of the two measurement methods, and N p is the number of pulse wave peaks detected within the time interval Δt, K is a constant used to adjust the dimension of A(t) to match the heart rate unit, and A(t) represents the value calculated based on the integral of the light absorption change caused by blood flow within a certain time window Δt:

[0133]

[0134] Among them, I max (t) and I min (t) are the maximum and minimum light intensity values ​​measured by the photoplethysmography sensor within period t, reflecting the changes in blood flow.

[0135] Monitoring the heart rate change data of each audience member is a key link in the optimization of holographic display layout, which is crucial for evaluating and enhancing the audience's interactive experience. In the present invention, the heart rate data monitoring method is improved, and combined with the improved genetic algorithm and neural network technology, it is more effectively applied to the optimization of holographic display layout.

[0136] In this embodiment, a comprehensive heart rate monitoring model is applied, which not only calculates the pulse wave peak, but also combines the changes in the photoplethysmogram to jointly evaluate the heart rate through direct and indirect measurement methods. The comprehensive heart rate monitoring model can more accurately capture heart rate data in various environments, such as music and scenes with drastic changes in light, and conduct a more detailed analysis of the audience's physiological responses. Heart rate change data is directly used to adjust the layout configuration of the holographic display, especially when used in conjunction with audience emotional feedback. By analyzing the relationship between heart rate data and the audience's emotional state in real time, the optimization algorithm can dynamically adjust the holographic image to adapt to the audience's actual feelings.

[0137] S13. Continuously collect the sound level and light intensity in the holographic display area through the environmental monitoring system, and use the sensor array to independently measure the sound in multiple frequency bands and spectral regions. and light Then adjust the audio-visual environment of the holographic display;

[0138] In this implementation, for sound, the average sound pressure level of each frequency band is measured in multiple frequency bands and calculated, and the sound frequency band is k different frequency bands {f1, f2, ..., f k}, the sound level of each frequency band is expressed as:

[0139]

[0140] in, is the sound level of the ith frequency band at time t, is the sound pressure level in this frequency band, T i It is a time window;

[0141] For illumination, measurements are made in different spectral regions, where the spectral region is j different wavelength intervals λ1,λ2,...,λ j}, the light intensity of each interval is expressed as:

[0142]

[0143] in, is the light intensity of the jth spectral interval at time t, L λ (t) is the instantaneous light intensity at wavelength λ, V(λ) is the human visual response function corresponding to the wavelength, λ j,min and λj,max are the minimum and maximum wavelengths of the jth interval.

[0144] S14. Using an emotion recognition algorithm to convert the audience's facial expression data and heart rate change data into an emotion score E(t). The emotion recognition algorithm is based on a large-scale sentiment analysis training set to identify multiple emotional states and quantify the emotional states into actionable values ​​to guide the content and layout adjustment of the holographic display.

[0145] In this implementation, S14 specifically includes:

[0146] S141, using facial expression recognition technology, collect the audience's facial expression data F(t), analyze and identify the position of the eyebrows, the degree of eye opening and closing, and the position of the mouth corners in the expression, and encode each feature point to generate a facial expression feature vector

[0147] S142, real-time monitoring of each audience's heart rate change data HR(t), extracting the average heart rate and heart rate variability and converting the data into a heart rate feature vector

[0148] S143, using the emotion recognition neural network model M to accept facial expression feature vector and the heart rate feature vector As input, the emotion recognition neural network model M is pre-trained to recognize and quantify various emotional states, and the output is the probability distribution P of emotional states. emo,k (t);

[0149] S144. According to the probability distribution P of the emotional state emo,k (t) Combined with the time weighting and emotional state intensity, the comprehensive emotional score E(t) is calculated:

[0150]

[0151] Among them, K is the number of emotional states that model M can recognize, and P emo,k (t) is the recognition probability of the kth emotional state at time t, w k is the weight of the kth emotional state, which is set based on the expected impact of the emotional state on the audience's behavior and interactive experience. k is the duration from the most recent time of the emotional state to the current time, τ k is the time decay constant of the kth emotional state, indicating the duration of the influence of different emotional states, and is used to simulate the natural process of emotional states weakening over time.

[0152] In this embodiment, the emotional state probability distribution P emo,k(t) Through the multi-layer perceptron model, the multi-layer perceptron model integrates a variety of physiological and behavioral data to predict emotional states. The specific calculation formula is as follows:

[0153] The facial expression feature vector and the heart rate feature vector Perform data fusion to generate a comprehensive feature vector

[0154]

[0155] Among them, ⊕ represents the merging operation of feature vectors;

[0156] The comprehensive feature vector The neural network model M is input into the neural network model. The neural network model M consists of multiple hidden layers. Except for the last layer, which uses the Softmax function to output the probability of each emotional state, each of the remaining layers uses the ReLU activation function. The output of the neural network model M The probability distribution P of each emotional state emo (t), the calculation process is expressed as:

[0157]

[0158] Among them, W i and are the weight matrix and bias vector of the i-th layer, n is the number of hidden layers, the ReLU function is a linear rectification function used to increase the nonlinear ability of the network, and the Softmax function is used to convert the original output value of the output layer into a probability distribution.

[0159] Steps S11 to S14 in this embodiment can accurately capture the three-dimensional position and emotional state of the audience in a changing display environment by introducing stereo cameras and multi-sensor technology, thereby improving the accuracy and practicality of the data. However, relying solely on position and expression data may not be sufficient to fully understand the audience's emotions and reactions. Therefore, this method enhances the ability to capture and analyze audience emotional changes by introducing heart rate monitoring and comprehensive emotion scoring algorithms, thereby providing a more refined basis for layout adjustment.

[0160] S15. Collect the data from steps S11-S14 to construct a holographic display environment parameter data set:

[0161]

[0162] S2. Initializing a layout configuration of a set of holographic display elements through an improved genetic algorithm according to the received holographic display environment parameter data set;

[0163] In this implementation, S2 includes the following steps:

[0164] S21, based on the received holographic display environment parameter data set Data set Define the initial population, each individual Represents a possible holographic display element layout configuration, including the position coordinates x of each display element i ,y i ,z i and the rotation angle θ i

[0165] S22. Applying adaptive fitness function Each individual is scored, where:

[0166]

[0167] in, It is the evaluation function of visual coverage, which is calculated based on the element position and the audience's perspective. It is an evaluation function of the interaction effect, which is based on the audience's dynamic position and expression parameters to calculate the degree of interaction with the holographic display content. is the resource utilization rate, λ is the coefficient for balancing visual effects and resource utilization, and w1 and w2 are preset weights;

[0168] Adaptive fitness function Combining the optimization capability of genetic algorithms with the evaluation accuracy of neural networks achieves a more dynamic and accurate layout optimization. By adjusting the value of λ, the influence ratio of genetic algorithms and neural network evaluation can be flexibly controlled, which improves the adaptability and efficiency of the optimization process.

[0169] The evaluation function of visual coverage is calculated based on the element position, the viewer's perspective, and the visual quality of the displayed content:

[0170]

[0171] Where n is the number of holographic elements, m is the number of viewers, and v ij represents the visibility of the i-th holographic element to the j-th viewer, α ij is the sight angle between the jth viewer and the ith holographic element, d ij is the distance between viewer j and holographic element i, σ is the attenuation coefficient, which controls the effect of distance on the visual effect;

[0172] This function effectively simulates the impact of audience perspective on visual coverage by combining distance attenuation and angle cosine adjustment, thereby improving the accuracy and visual effect of layout optimization.

[0173] The evaluation function of the interactive effect is based on the audience's dynamic position, expression, and heart rate data to calculate the degree of interaction with the holographic display content:

[0174]

[0175] Where m is the number of viewers, R k is the response coefficient of audience k, reflecting the interaction willingness of each audience, β, γ, δ are weight factors used to balance the influence of spatial interaction, physiological interaction and emotional interaction;

[0176] S k is the spatial interaction index calculated based on the position and facial expression data of viewer k, and is defined as:

[0177]

[0178] Among them, s ki is the spatial sensitivity of viewer k to holographic element i, L i is the position of element i;

[0179] H k It is a physiological interaction index calculated based on the heart rate change data HR(t):

[0180]

[0181] Among them, HR rest is the resting heart rate, a is the adjustment factor;

[0182] E k is the impact of the sentiment score E(t) on the interaction:

[0183]

[0184] Normalize the sentiment scores between 0 and 1;

[0185] The evaluation function of the interactive effect comprehensively considers the interactive effects of spatial position, physiological signals and emotional responses, making the holographic display more finely adapted to the audience's actual experience and enhancing the interactivity and attractiveness of the layout.

[0186] The improved genetic algorithm takes into account comprehensive environmental and audience data when initializing the holographic layout configuration, which maintains high efficiency and adaptability under dynamic and changing display conditions. A single genetic algorithm may be inefficient when processing complex environmental data and easily fall into local optimal solutions. Therefore, the present invention ensures the global search capability of the algorithm and the diversity of solutions by introducing the tournament selection method and uniform crossover operation.

[0187] S23. Use the tournament selection method to randomly select individuals from the population, compare the fitness of the individuals, and select the individual with the highest fitness to copy to the next generation:

[0188]

[0189] in, and are two individuals selected at random;

[0190] S24, implement uniform crossover operation to perform equal exchange of genes on the selected fathers. Each gene position determines which father its genetic information comes from. The gene vectors of the two father individuals are defined as and Generate a new individual Its gene position is and Random and equal selection, uniform crossover is expressed as:

[0191]

[0192] Among them, r i is a random number generated for the ith gene position, satisfying r i ~U(0,1), indicating uniform distribution;

[0193] S25. Introduce a gradient-based mutation method to select the parameter x in the gene i ,y i ,z i ,θ i Make adjustments using the local gradient information of the objective function F Guide the mutation process and achieve directional optimization adjustment:

[0194]

[0195] Among them, σ is the parameter that controls the step length of the variable. is The gradient of the objective function calculated at the point, including the partial derivatives for each gene position;

[0196] The gradient is computed using finite differences:

[0197]

[0198] Here, δ is a small positive number used to calculate x i The gradient of the direction, the same method is used for y i ,z i ,θ i Gradient calculation of ;

[0199] S26. Repeat steps S22 to S25 until a population that meets a specific fitness standard is generated.

[0200] At the same time, step S2 introduces an adaptive fitness function, combines neural network scoring and traditional fitness evaluation to form a hybrid fitness function Not only does it take into account the layout effect based on physical parameters, but it also incorporates audience interaction data and emotional feedback to make the fitness evaluation more comprehensive. By dynamically adjusting the weight coefficient λ in the fitness function, it can automatically adjust the focus of fitness evaluation according to real-time data, thereby improving the personalization and scene adaptability of layout configuration.

[0201] Dynamic selection pressure adjustment mechanism, introduce probability distribution adjustment mechanism, through calculation To dynamically adjust the probability of each individual being selected, based on the score and mixed fitness To optimize the selection process of the population, the parameter κ is used to regulate the selection pressure to ensure that high-scoring configurations have a higher selection probability, thereby accelerating the optimization process and avoiding local optimality.

[0202] The mutation strategy of dynamically adjusting the score is based on the score of each layout configuration. Adjust the mutation probability and realize dynamic adjustment of the mutation rate. Configurations with low scores will have higher mutation rates, increasing the chances of exploring new configurations. Flexible adjustment of the mutation strategy is performed based on the comprehensive score of the layout configuration, increasing the algorithm's adaptability to complex environments and the diversity of optimization.

[0203] Uniform crossover and gradient-based mutation operations. Uniform crossover technology is used to ensure balanced information exchange, and gradient-based mutation operations are introduced to use the local gradient information of the objective function F. Guiding the mutation process, gradient-guided mutation operations enable genetic algorithms to be more effective in local search, using gradient information to directly influence the mutation steps, improving solution accuracy and algorithm convergence speed.

[0204] S3, using a neural network model to evaluate each layout configuration generated in step S2, where the neural network model is pre-trained and can evaluate the effect of the layout configuration based on the holographic display effect and audience interaction data;

[0205] In this implementation, S3 includes the following steps:

[0206] S31, using a pre-trained neural network model M to evaluate the visual effect and interactive experience of the holographic display element layout configuration, inputting each layout configuration generated in step S2 And from the holographic display environment parameter dataset Data set :

[0207]

[0208] in, Indicates flattening the dynamic position data into a one-dimensional vector. and are the sound level and light intensity vectors, respectively, integrating data from all measured frequencies and wavelengths;

[0209] S32. The structure of the neural network model M includes a multi-layer perceptron, and each layer uses a nonlinear activation function;

[0210] S33, the output layer uses the Softmax function to convert the evaluation results of the neural network into a probability distribution It indicates the suitability of each layout configuration for the holographic display effect and audience interactive experience. The calculation formula is:

[0211]

[0212] Where the exponential function and sum operation ensure that the output is a probability distribution, k is the number of possible layout configurations;

[0213] S34. Configure for each layout Generate a rating Based on probability distribution Converted into numerical values ​​that are easy to optimize, the scoring formula is:

[0214]

[0215] By the probability of each configuration Applying logarithms and negative signs, probabilities are converted to scores so that high probabilities correspond to high scores;

[0216] S35. Scoring of all layout configurations The results are summarized and fed back to the improved genetic algorithm to guide the iterative optimization process of the layout.

[0217] This embodiment enhances the accuracy and timeliness of the evaluation by using a pre-trained neural network model to evaluate the layout configuration. However, a single visual evaluation may not be sufficient to fully optimize the interactive experience of the holographic display. Therefore, the present invention combines the audience's behavioral and physiological data and uses Softmax to output probability distribution to ensure the comprehensiveness and scientificity of the layout optimization decision.

[0218] S4. According to the evaluation results of the neural network model, adjust the layout configuration in the improved genetic algorithm, and iterate and optimize to gradually improve the visual effect and interactive experience of the layout solution;

[0219] In this implementation, S4 includes the following steps:

[0220] S41, receiving all layout configurations of step S35 Rating The ratings reflect the visual and interactive experience of each layout configuration;

[0221] S42. According to the rating Adjusting the layout configuration in the improved genetic algorithm using the fitness function Updated to include evaluation results of neural network models;

[0222] In this implementation manner, the layout configuration process in adjusting the improved genetic algorithm includes:

[0223] Define the hybrid fitness function Combined with the original fitness of the improved genetic algorithm and the neural network score

[0224]

[0225] Among them, λ is an adjustment coefficient, which is used to balance the weight of traditional fitness evaluation and neural network-based evaluation in genetic algorithms;

[0226] Update the layout configuration selection strategy, introduce a probability distribution adjustment mechanism to respond to score changes, and adjust the score-based Dynamic range, calculate the selection probability of each layout configuration

[0227]

[0228] Among them, κ is a positive parameter used to control the size of the selection pressure, so that the layout configuration with a high score has a higher probability of being selected, thereby accelerating the convergence of the algorithm;

[0229] Introducing a mutation strategy based on dynamic score adjustment to adjust layout parameters in a targeted manner, mutation probability Depending on the current score of the layout configuration, low-scoring configurations receive higher probability of mutation, increasing the chances of exploring new configurations:

[0230]

[0231] Among them, p base is the basic mutation rate, p max is the maximum allowed mutation rate, δ and θ are parameters for adjusting the mutation sensitivity, which are used to adjust the mutation rate according to the comprehensive score of the layout configuration.

[0232] S43. Use updated fitness function Select individuals in the next generation population, using the tournament selection method to ensure that individuals with high fitness have a higher probability of being selected;

[0233] S44, performing uniform crossover and mutation operations of the genetic algorithm on the selected individuals, wherein the mutation operation adjusts the layout parameters according to the gradient information to introduce new genetic mutations;

[0234] S45, repeating steps S42 to S44 until a preset number of iterations is reached or the layout optimization meets a predetermined performance indicator, and the system continuously learns and adapts to optimize the layout configuration of the holographic display;

[0235] S46. After each iteration, evaluate the average fitness of the current population and adjust the crossover rate and mutation rate of the genetic algorithm as needed.

[0236] S5, repeating step S3 and step S4 until a preset number of iterations is reached or the performance index of layout optimization is met;

[0237] S6. Output the final optimized holographic display layout configuration for use in actual holographic display.

[0238] In response to the problems of the prior art, this embodiment integrates an improved genetic algorithm and a pre-trained neural network model, which can not only collect and process complex data about the holographic display environment (such as the dynamic position of the audience, the ambient sound level, the light intensity and the audience's emotional feedback) in real time, but also dynamically optimize the layout of the holographic display based on these data.

[0239] The specific implementation of this method includes: first, using an input device to collect the parameters of the holographic display environment in real time and constructing a holographic display environment parameter data set; then, initializing the layout configuration of the holographic display elements through an improved genetic algorithm; next, using a neural network model to evaluate the effect of each layout configuration, and adjusting the layout configuration according to the evaluation results, iteratively optimizing to gradually improve the visual effect and interactive experience of the layout scheme; finally, outputting the final optimized holographic display layout configuration.

[0240] The present invention monitors and analyzes changes in the holographic display environment (such as audience position, ambient light and sound level, etc.) in real time, and dynamically adjusts the layout of holographic display elements based on these data, so that the holographic display can instantly respond to audience behavior and changes in the display environment, greatly improving the quality of user interaction experience and visual effects.

[0241] The present invention not only accelerates the optimization speed of holographic layout configuration but also improves the efficiency of the optimization process by combining the improved genetic algorithm and neural network technology. The improved genetic algorithm accelerates the convergence speed of the holographic layout by introducing a fitness function based on neural network evaluation, making the iterative process more efficient.

[0242] The present invention allows accurate evaluation of the visual effects and interactive experience of the holographic display layout through the application of the neural network model. By feeding back these evaluation results, the improved genetic algorithm can more accurately adjust the layout configuration, thereby providing more personalized display content while ensuring high visual quality and optimizing the user interaction experience.

[0243] The present invention can be widely applied to different holographic display application scenarios, including education, entertainment, advertising, scientific research and other fields. In addition, due to its design flexibility, it can be easily integrated into the existing holographic display system or used as the basic technical platform of the new system.

[0244] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A holographic display layout optimization method based on improved genetic algorithm and neural network, It is characterized in that The steps include: S1. Using input devices to receive the number of audience members, dynamic position changes of audience members, real-time environmental sound levels, changes in ambient light intensity, and real-time audience emotional feedback about the holographic display environment, and construct a holographic display environment parameter data set; S2. Initializing a layout configuration of a set of holographic display elements through an improved genetic algorithm according to the received holographic display environment parameter data set; S3, using a neural network model to evaluate each layout configuration generated in step S2, wherein the neural network model is pre-trained to evaluate the effect of the layout configuration based on the holographic display effect and audience interaction data; S4. According to the evaluation results of the neural network model, adjust the layout configuration in the improved genetic algorithm, and iterate and optimize to gradually improve the visual effect and interactive experience of the layout solution; S5, repeating step S3 and step S4 until a preset number of iterations is reached or the performance index of layout optimization is met; S6. Outputting the final optimized holographic display layout configuration for use in actual holographic display; The S1 comprises the following steps: S11. Deploy stereo cameras at key locations in the holographic display area. The stereo cameras use depth perception technology to capture the audience's three-dimensional spatial position and facial expressions in real time, and provide the audience's dynamic position data D(x,y) and facial expression data F(t) by generating depth maps and texture mapping: Where D(x,y) represents the depth value of the coordinate (x,y) on the image plane, f is the focal length of the camera, B is the baseline distance between the cameras, and d(x,y) is the corresponding disparity value; Among them, F(t) represents the facial expression feature value at time t, N f is the number of feature points, f i is the measured value of the i-th feature point, μ i and σ i are the mean and standard deviation of the i-th feature point, w i is the weight of the i-th feature point; S12. Using the photoelectric volumetric pulse wave sensor integrated in the seats or wearable devices in the holographic display area, the changes in blood flow are analyzed through a comprehensive heart rate monitoring model to monitor the heart rate change data HR(t) of each audience member in real time; S13. Continuously collect the sound level and light intensity in the holographic display area through the environmental monitoring system, and use the sensor array to independently measure the sound in multiple frequency bands and spectral regions. and light Then adjust the audio-visual environment of the holographic display; S14. Using an emotion recognition algorithm to convert the audience's facial expression data and heart rate change data into an emotion score E(t). The emotion recognition algorithm is based on a large-scale sentiment analysis training set to identify multiple emotional states and quantify the emotional states into actionable values ​​to guide the content and layout adjustment of the holographic display. S15. Collect the data from steps S11-S14 to construct a holographic display environment parameter data set: The S2 comprises the following steps: S21, based on the received holographic display environment parameter data set Data set Define the initial population, each individual Represents a possible holographic display element layout configuration, including the position coordinates x of each display element i ,y i ,z i and the rotation angle θ i S22. Applying adaptive fitness function Each individual is scored, where: in, It is the evaluation function of visual coverage, which is calculated based on the element position and the audience's perspective. It is an evaluation function of the interaction effect, which is based on the audience's dynamic position and expression parameters to calculate the degree of interaction with the holographic display content. is the resource utilization rate, λ is the coefficient that balances visual effects and resource utilization, and w 1 and w 2 is the preset weight; The evaluation function of visual coverage is calculated based on the element position, the viewer's perspective, and the visual quality of the displayed content: Where n is the number of holographic elements, m is the number of viewers, and v ij represents the visibility of the i-th holographic element to the j-th viewer, α ij is the sight angle between the jth viewer and the ith holographic element, d ij is the distance between viewer j and holographic element i, σ is the attenuation coefficient, which controls the effect of distance on the visual effect; The evaluation function of the interactive effect is based on the audience's dynamic position, expression, and heart rate data to calculate the degree of interaction with the holographic display content: Where m is the number of viewers, R k is the response coefficient of audience k, reflecting the interaction willingness of each audience, β, γ, δ are weight factors used to balance the influence of spatial interaction, physiological interaction and emotional interaction; S k is the spatial interaction index calculated based on the position and facial expression data of viewer k, and is defined as: Among them, s ki is the spatial sensitivity of viewer k to holographic element i, L i is the position of element i; H k It is a physiological interaction index calculated based on the heart rate change data HR(t): Among them, HR rest is the resting heart rate, a is the regulating factor; E k is the impact of the sentiment score E(t) on the interaction: Normalize the sentiment scores between 0 and 1; S23. Use the tournament selection method to randomly select individuals from the population, compare the fitness of the individuals, and select the individual with the highest fitness to copy to the next generation: in, and are two individuals selected at random; S24, implement uniform crossover operation to perform equal exchange of genes on the selected fathers. Each gene position determines which father its genetic information comes from. The gene vectors of the two father individuals are defined as and Generate a new individual Its gene position is and Random and equal selection, uniform crossover is expressed as: Among them, r i is a random number generated for the ith gene position, satisfying r i ~U(0,1), indicating uniform distribution; S25. Introduce a gradient-based mutation method to select the parameter x in the gene i ,y i ,z i ,θ i Make adjustments using the local gradient information of the objective function F Guide the mutation process and achieve directional optimization adjustment: Among them, σ is the parameter that controls the step length of the variable. is The gradient of the objective function calculated at the point, including the partial derivatives for each gene position; The gradient is computed using finite differences: Here, δ is a small positive number used to calculate x i The gradient of the direction, the same method is used for y i ,z i ,θ i Gradient calculation of ; S26, repeating steps S22 to S25 until a population that meets the fitness standard is generated; The S3 comprises the following steps: S31, using a pre-trained neural network model M to evaluate the visual effect and interactive experience of the holographic display element layout configuration, inputting each layout configuration generated in step S2 And from the holographic display environment parameter dataset Data set : in, Indicates flattening the dynamic position data into a one-dimensional vector. and are the sound level and light intensity vectors, respectively, integrating data from all measured frequencies and wavelengths; S32. The structure of the neural network model M includes a multi-layer perceptron, and each layer uses a nonlinear activation function; S33, the output layer uses the Softmax function to convert the evaluation results of the neural network into a probability distribution It indicates the suitability of each layout configuration for the holographic display effect and audience interactive experience. The calculation formula is: Where the exponential function and sum operation ensure that the output is a probability distribution, k is the number of possible layout configurations; S34. Configure for each layout Generate a rating Based on probability distribution Converted into numerical values ​​that are easy to optimize, the scoring formula is: By the probability of each configuration Applying logarithms and negative signs, probabilities are converted to scores so that high probabilities correspond to high scores; S35. Scoring of all layout configurations Summarize and feed back to the improved genetic algorithm to guide the iterative optimization process of the layout; The S4 comprises the following steps: S41, receiving all layout configurations of step S35 Rating The ratings reflect the visual and interactive experience of each layout configuration; S42. According to the rating Adjusting the layout configuration in the improved genetic algorithm using the fitness function Updated to include evaluation results of neural network models; S43. Use updated fitness function Select individuals in the next generation population, using the tournament selection method to ensure that individuals with high fitness have a higher probability of being selected; S44, performing uniform crossover and mutation operations of the genetic algorithm on the selected individuals, wherein the mutation operation adjusts the layout parameters according to the gradient information to introduce new genetic mutations; S45, repeating steps S42 to S44 until a preset number of iterations is reached or the layout optimization meets a predetermined performance indicator, and the system continuously learns and adapts to optimize the layout configuration of the holographic display; S46. After each iteration, evaluate the average fitness of the current population and adjust the crossover rate and mutation rate of the genetic algorithm as needed.

2. According to claim 1, a holographic display layout optimization method based on improved genetic algorithm and neural network, It is characterized in that The comprehensive heart rate monitoring model combines direct pulse wave peak counts with the integrated information of the photoplethysmogram obtained through blood flow changes to measure heart rate: Where HR(t) represents the heart rate at time t, α is a weight parameter used to balance the contribution of the two measurement methods, and N p is the number of pulse wave peaks detected within the time interval Δt, K is a constant used to adjust the dimension of A(t) to match the heart rate unit, and A(t) represents the value calculated based on the integral of the light absorption change caused by blood flow within a certain time window Δt: Among them, I max (t) and I min (t) are the maximum and minimum light intensity values ​​measured by the photoplethysmography sensor within period t, reflecting the changes in blood flow.

3. According to claim 2, a holographic display layout optimization method based on improved genetic algorithm and neural network, It is characterized in that For sound, the average sound pressure level of each frequency band is measured in multiple frequency bands and calculated. The sound band is k different frequency bands {f 1 ,f 2 ,...,f k }, the sound level of each frequency band is expressed as: in, is the sound level of the ith frequency band at time t, is the sound pressure level of the ith frequency band, T i It is a time window; For illumination, measurements are made in different spectral regions, where the spectral regions are j different wavelength intervals {λ 1 ,λ 2 ,...,λ j }, the light intensity of each interval is expressed as: Among them, I λj (t) is the light intensity of the jth spectral interval at time t, L λ (t) is the instantaneous light intensity at wavelength λ, V(λ) is the human visual response function corresponding to the wavelength, λ j,min and λ j,max are the minimum and maximum wavelengths of the jth interval.

4. According to claim 3, a holographic display layout optimization method based on improved genetic algorithm and neural network, It is characterized in that The S14 specifically includes: S141, using facial expression recognition technology, collect the audience's facial expression data F(t), analyze and identify the position of the eyebrows, the degree of eye opening and closing, and the position of the mouth corners in the expression, and encode each feature point to generate a facial expression feature vector S142, real-time monitoring of each audience's heart rate change data HR(t), extracting the average heart rate and heart rate variability and converting the data into a heart rate feature vector S143, using the emotion recognition neural network model M to accept facial expression feature vector and the heart rate feature vector As input, the emotion recognition neural network model M is pre-trained to recognize and quantify various emotional states, and the output is the probability distribution P of emotional states. emo,k (t); S144. According to the probability distribution P of the emotional state emo,k (t) Combined with the time weighting and emotional state intensity, the comprehensive emotional score E(t) is calculated: Among them, K is the number of emotional states that model M can recognize, and P emo,k (t) is the recognition probability of the kth emotional state at time t, w k is the weight of the kth emotional state, which is set based on the expected impact of the emotional state on the audience's behavior and interactive experience. k is the duration from the most recent time of the emotional state to the current time, τ k is the time decay constant of the kth emotional state, indicating the duration of the influence of different emotional states, and is used to simulate the natural process of emotional states weakening over time.

5. According to claim 4, a holographic display layout optimization method based on improved genetic algorithm and neural network, It is characterized in that The emotional state probability distribution P emo,k (t) Through the multi-layer perceptron model, the multi-layer perceptron model integrates a variety of physiological and behavioral data to predict emotional states. The specific calculation formula is as follows: The facial expression feature vector and the heart rate feature vector Perform data fusion to generate a comprehensive feature vector in, represents the merging operation of feature vectors; The comprehensive feature vector The neural network model M is input into the neural network model. The neural network model M consists of multiple hidden layers. Except for the last layer, which uses the Softmax function to output the probability of each emotional state, each of the remaining layers uses the ReLU activation function. The output of the neural network model M The probability distribution P of each emotional state emo (t), the calculation process is expressed as: Among them, W i and are the weight matrix and bias vector of the i-th layer, n is the number of hidden layers, the ReLU function is a linear rectification function used to increase the nonlinear ability of the network, and the Softmax function is used to convert the original output value of the output layer into a probability distribution.

6. According to claim 1, a holographic display layout optimization method based on improved genetic algorithm and neural network, It is characterized in that The layout configuration process in the adjusted and improved genetic algorithm includes: Define the hybrid fitness function Combined with the original fitness of the improved genetic algorithm and the neural network score Among them, λ is an adjustment coefficient, which is used to balance the weight of traditional fitness evaluation and neural network-based evaluation in genetic algorithms; Update the layout configuration selection strategy, introduce a probability distribution adjustment mechanism to respond to score changes, and adjust the score-based Dynamic range, calculate the selection probability of each layout configuration Among them, κ is a positive parameter used to control the size of the selection pressure, so that the layout configuration with a high score has a higher probability of being selected, thereby accelerating the convergence of the algorithm; Introducing a mutation strategy based on dynamic score adjustment to adjust layout parameters in a targeted manner, mutation probability Depending on the current score of the layout configuration, low-scoring configurations receive higher probability of mutation, increasing the chances of exploring new configurations: Among them, p base is the basic mutation rate, p max is the maximum allowed mutation rate, δ and θ are parameters for adjusting the mutation sensitivity, which are used to adjust the mutation rate according to the comprehensive score of the layout configuration.

Citation Information

Patent Citations

  • Dairy cow feed intake evaluation method based on genetic algorithm optimized BP neural network

    CN112116145A

  • Meteorological data prediction method for optimizing BP neural network based on improved genetic algorithm

    CN113850420A

  • RBF neural network training method and prediction system based on gene regulation genetic algorithm

    CN115310664A

  • Active control holographic method and holographic element for free-form surface super-structure surface

    CN116125774A

  • Urban green space automatic layout method based on artificial intelligence

    CN116401736A