Multi-mode travel light artistic effect generation method and system with body interaction

Abstract

The invention provides a method and a system for generating an artistic effect of multi-modal travel light with body interaction, and relates to the technical field of travel light art, and the method comprises the steps: collecting and analyzing user interaction data, and constructing a behavior feature sequence; an interaction space is divided, a space vitality map is generated, and a light effect dynamic mapping relation is established; generating an initial light effect and collecting feedback; the user preference features are extracted, the light control parameters are optimized, the optimal light control instruction is generated, accurate interaction between the light effect and the user behavior can be achieved, and the immersion and experience satisfaction of the text travel scene are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of cultural and tourism lighting art, and particularly to a method and system for generating cultural and tourism lighting art effects with multimodal embodied interaction. Background Art

[0002] As an important part of the modern tourism industry, cultural and tourism lighting art combines lighting devices with art design to provide tourists with immersive visual experiences, enhancing the attractiveness and sense of experience of cultural and tourism scenic spots. With the development of multimedia technology, cultural and tourism lighting art has evolved from a single static lighting display to a multimodal interactive experience, enabling tourists to not only be viewers but also participants, interacting with lighting devices in various ways such as body movements, expressions, and sounds to create personalized lighting art effects.

[0003] Currently, cultural and tourism lighting art designs are mainly based on designers pre-designing fixed lighting effects according to the characteristics of scenic spots, or using simple sensors to trigger preset lighting changes. With the development of the Internet of Things technology, lighting interaction systems based on user behavior analysis have begun to be applied in cultural and tourism scenarios, adjusting lighting effects by collecting tourist activity data to enhance tourists' sense of participation and experience. However, the existing technologies still suffer from the lack of comprehensive analysis of multimodal interaction data, being unable to comprehensively capture the behavior patterns and emotional changes of users in cultural and tourism scenarios, resulting in the disconnection between lighting art effects and users' actual experience needs, lacking the ability of dynamic perception and analysis of the overall spatial vitality distribution, being unable to generate corresponding lighting effects according to the differential changes in user activity levels and emotional intensities in different regions, making the presentation of lighting art effects lack spatial coherence and fluidity, and being unable to perform adaptive optimization based on users' real-time feedback on lighting effects, leading to a gap between lighting effects and user preferences and being difficult to meet the personalized needs of different user groups, etc. Therefore, there is an urgent need for a solution to address the problems existing in the prior art. Summary of the Invention

[0004] The embodiments of the present invention provide a method and system for generating cultural and tourism lighting art effects with multimodal embodied interaction, which can at least solve some of the problems existing in the prior art.

[0005] In the first aspect of the embodiments of the present invention, a method for generating cultural and tourism lighting art effects with multimodal embodied interaction is provided, including: Collecting the initial interaction data of users in a cultural and tourism scenario and analyzing the temporal correlation, extracting user behavior characteristics, constructing a user behavior characteristic sequence and calculating the change trend of the user behavior characteristic sequence, obtaining the light intensity, crowd density, and distribution of cultural element information in the cultural and tourism scenario, and generating a scene interaction data stream by combining the user behavior characteristic sequence and the corresponding change trend. Divide the interaction space based on the scenario interaction data stream, calculate the activity and emotional intensity of users in each space and generate a spatial vitality map, calculate the energy difference between adjacent spaces according to the spatial vitality map, determine the lighting fade coefficient, track the real-time change of spatial vitality based on the lighting fade coefficient, and establish a dynamic mapping relationship of lighting effects; Output and execute the lighting control instruction according to the dynamic mapping relationship, generate the initial lighting effect and collect the interactive feedback data of the user on the initial lighting effect; Based on the interactive feedback data, extract the user's preference characteristics for lighting effects, construct a lighting effect evaluation index, optimize and sort the lighting control parameters according to the lighting effect evaluation index, establish a parameter optimization strategy, and iteratively adjust the lighting control instruction based on the parameter optimization strategy to generate the optimal lighting control instruction; Output the optimal lighting control instruction, execute and record the optimal lighting effect.

[0006] In an alternative embodiment, Collect the initial interaction data of the user in the cultural and tourism scenario and analyze the temporal correlation, extract the user behavior characteristics, construct a user behavior characteristic sequence and calculate the change trend of the user behavior characteristic sequence, obtain the illumination intensity, pedestrian flow density and cultural element information distribution of the cultural and tourism scenario, and generate a scenario interaction data stream in combination with the user behavior characteristic sequence and the corresponding change trend, including: Collect the initial interaction data of the user in the cultural and tourism scenario, where the initial interaction data includes posture information, movement trajectory information, voice commands and emotional intonation information; Construct a correlation matrix in the time dimension and extract the temporal correlation of the initial interaction data, construct a user behavior characteristic sequence according to the temporal correlation, the user behavior characteristic sequence includes behavior duration, behavior occurrence frequency, behavior intensity change and behavior spatial position, and calculate the change trend of the user behavior characteristic sequence at different scales; Collect the illumination intensity, pedestrian flow density and cultural element information distribution of the cultural and tourism scenario, map the user behavior characteristic sequence to the corresponding scenario grid of the cultural and tourism scenario according to the spatial position, determine the behavior activity of each grid, calculate the activity difference value between different grids and combine the change trend to determine the scenario interaction data stream, where the scenario interaction data stream includes the real-time state and predicted state of the interaction data.

[0007] In an alternative embodiment, Divide the interaction space based on the scenario interaction data stream, calculate the activity and emotional intensity of users in each space and generate a spatial vitality map, calculate the energy difference between adjacent spaces according to the spatial vitality map, determine the lighting fade coefficient, and track the real-time change of spatial vitality based on the lighting fade coefficient. Establish the dynamic mapping relationship of the lighting effect, including: Divide the cultural and tourism scenario into multiple interaction spaces based on the scenario interaction data stream, establish a three-dimensional coordinate system in the interaction space, divide the interaction space into multiple grid cells, and calculate the ratio of the number of interaction events in each grid cell to the grid volume to obtain the interaction density; Calculate the weighted sum of the basic behavior intensity, movement trajectory complexity, and interaction density of users in each interaction space to obtain the user activity, calculate the weighted sum of the physiological characteristic emotion value, voice emotion value, and gesture emotion value of users in each interaction space to obtain the emotional intensity, and perform a linear weighted combination of the user activity and the emotional intensity to obtain the spatial vitality map; Determine the spatial vitality corresponding to each interaction space according to the spatial vitality map, calculate the absolute difference of the central point vitality values of adjacent interaction spaces to obtain the energy difference, and substitute the energy difference into an exponential function to obtain the lighting fade coefficient, where the input of the exponential function is the negative of the ratio of the energy difference to a preset energy threshold parameter; Track the real-time change of spatial vitality based on the lighting fade coefficient, extract crowd flow characteristics to calculate the dynamic partition boundary, construct a cross-region fade function to optimize the lighting fade coefficient, identify group behavior patterns to predict the vitality change trend and perform error compensation, and establish the dynamic mapping relationship between the real-time change of spatial vitality and the lighting effect.

[0008] In an alternative embodiment, Track the real-time change of spatial vitality based on the lighting fade coefficient, extract crowd flow characteristics to calculate the dynamic partition boundary, construct a cross-region fade function to optimize the lighting fade coefficient, identify group behavior patterns to predict the vitality change trend and perform error compensation, and establish the dynamic mapping relationship between the real-time change of spatial vitality and the lighting effect, including: Obtain the real-time change data of spatial vitality, generate a vitality mapping reference value according to the real-time change data of spatial vitality and a preset lighting fade coefficient, collect the crowd movement data in the interaction space, extract the crowd flow characteristics, and calculate the flow intensity by multiplying the crowd flow characteristics by the crowd density in the interaction space; Generate an adjustment factor for the dynamic partition boundary according to the difference in crowd flow characteristics and the difference in crowd density between adjacent interaction spaces, and adjust the boundary between adjacent interaction spaces according to the adjustment factor; Construct a cross - regional gradient function for the adjacent interaction space, substitute the flow intensity and spatial distance into the cross - regional gradient function, and fuse the calculation result with the preset lighting gradient coefficient to generate an optimized lighting gradient coefficient; Extract group behavior pattern features based on the flow intensity, input the group behavior pattern features and pre - collected historical vitality data into a prediction model to calculate the vitality change trend, and perform error compensation on the vitality change trend and the spatially - vital data collected in real - time, and output a compensated spatial vitality mapping value; Establish a dynamic mapping function according to the compensated spatial vitality mapping value, the optimized lighting gradient coefficient, and the pre - set environmental constraint parameters, and establish a dynamic mapping relationship between the real - time change of the spatial vitality and the lighting effect. The dynamic mapping relationship is dynamically adjusted according to the change rate of the compensated spatial vitality mapping value and the change rate of the optimized lighting gradient coefficient.

[0009] In an alternative implementation manner, Output and execute a lighting control instruction according to the dynamic mapping relationship to generate an initial lighting effect and collect interactive feedback data of the user on the initial lighting effect, including: Output and execute a lighting control instruction according to the dynamic mapping relationship. The lighting control instruction includes lighting brightness, color temperature, and gradient timing parameters; Generate an initial lighting effect according to the lighting control instruction, and the initial lighting effect is dynamically adjusted according to the real - time change of the spatial vitality; Collect interactive feedback data of the user on the initial lighting effect. The interactive feedback data includes user residence duration, interaction frequency, and pre - set emotional feedback indicators.

[0010] In an alternative implementation manner, Based on the interactive feedback data, extract the preference features of the user for the lighting effect, construct a lighting effect evaluation index, optimize and sort the lighting control parameters according to the lighting effect evaluation index, establish a parameter optimization strategy, and iteratively adjust the lighting control instruction based on the parameter optimization strategy to generate an optimal lighting control instruction, including: Collect interactive feedback data of the user on the lighting effect, construct a multi - dimensional feature vector according to the user residence duration, interaction frequency, and emotional feedback indicators in the interactive feedback data, and input the multi - dimensional feature vector into a pre - set user preference scoring module to obtain a preference score; Fuse the preference score with comfort indicators and spatial coordination indicators to generate a comprehensive evaluation value, and construct a time - series evaluation matrix based on the comprehensive evaluation value. The time - series evaluation matrix contains a sequence of evaluation values within a historical time window; Calculate the sensitivity of the comprehensive evaluation value to the lighting control parameters, construct a parameter sensitivity distribution module using kernel density estimation, suppress the influence of abnormal samples through iteratively reweighted least squares, determine the parameter priority based on the historical optimization effect and sort them; Based on the sorting result of the lighting control parameters, construct a parameter optimization objective function in sequence. Take the minimization of the mean square error of the evaluation value sequence in the time series evaluation matrix as the optimization objective, use the parameter change amplitude between adjacent moments as a constraint condition, and perform gradient iterative update on the lighting control parameters based on the optimization objective function. Stop the iteration when the difference in the comprehensive evaluation value between adjacent moments is less than a preset threshold, and output the optimized lighting control parameters; According to the optimized lighting control parameters and the pre-set environmental constraint parameters, generate the optimal lighting control instruction in combination with the pre-set dynamic gain matrix.

[0011] In an alternative implementation, Calculating the sensitivity of the comprehensive evaluation value to the lighting control parameters, constructing a parameter sensitivity distribution module using kernel density estimation, suppressing the influence of abnormal samples through iteratively reweighted least squares, and determining the parameter priority and sorting based on the historical optimization effect includes: Obtain the historical adjustment data of the lighting control parameters and the corresponding preference scores. Based on the historical adjustment data and the corresponding preference scores, calculate the first-order sensitivity of the comprehensive evaluation value to the lighting control parameters through partial derivatives, calculate the second-order interaction sensitivity through mixed partial derivatives, and construct a Jacobian matrix in combination with the central difference method to obtain the initial sensitivity sequence; Perform kernel density estimation on the initial sensitivity sequence using a Gaussian kernel function, adaptively calculate the optimal bandwidth parameter based on the sample standard deviation, transform the initial sensitivity sequence through the kernel function to obtain a continuous probability distribution, and output the distribution characteristics of the parameter sensitivity; Construct a Huber-type M-estimation loss function to perform anomaly detection on the distribution characteristics of the parameter sensitivity, adaptively calculate the sample weight matrix based on the residual size, apply the sample weight matrix to the iteratively reweighted least squares calculation, and suppress the influence of abnormal samples through multiple rounds of iterative optimization, and output the corrected sensitivity value; Calculate the temporal weight of the pre-obtained historical evaluation data based on the time decay exponential function, and perform weighted summation of the temporal weight and the historical comprehensive evaluation value to obtain the parameter historical optimization effect; Perform multi-objective weighted fusion on the corrected sensitivity value, the parameter historical optimization effect, and the parameter stability index to determine the parameter priority, sort the lighting control parameters in descending order according to the parameter priority, and generate a parameter optimization order table, where the parameter stability index includes the parameter fluctuation range, the convergence speed, and the disturbance recovery ability.

[0012] In a second aspect of the embodiments of the present invention, a system for generating a cultural and tourism lighting art effect for multimodal embodied interaction is provided, including: A first unit for collecting initial interaction data of a user in a cultural and tourism scenario, analyzing the temporal correlation, extracting user behavior characteristics, constructing a user behavior characteristic sequence, calculating the change trend of the user behavior characteristic sequence, obtaining the light intensity, crowd density, and cultural element information distribution of the cultural and tourism scenario, and generating a scenario interaction data stream in combination with the user behavior characteristic sequence and the corresponding change trend; A second unit for dividing the interaction space based on the scenario interaction data stream, calculating the activity and emotional intensity of users in each space, generating a space vitality map, calculating the energy difference between adjacent spaces according to the space vitality map, determining the light fading coefficient, tracking the real-time change of space vitality based on the light fading coefficient, and establishing a dynamic mapping relationship of the lighting effect; A third unit for outputting and executing a lighting control instruction according to the dynamic mapping relationship, generating an initial lighting effect, and collecting interactive feedback data of the user on the initial lighting effect; A fourth unit for extracting the preference characteristics of the user for the lighting effect based on the interactive feedback data, constructing a lighting effect evaluation index, optimizing and sorting the lighting control parameters according to the lighting effect evaluation index, establishing a parameter optimization strategy, and iteratively adjusting the lighting control instruction based on the parameter optimization strategy to generate an optimal lighting control instruction; A fifth unit for outputting the optimal lighting control instruction, executing it, and recording the optimal lighting effect.

[0013] In a third aspect of the embodiments of the present invention, an electronic device is provided, including: A processor and a memory for storing instructions executable by the processor, wherein the processor is configured to call the instructions stored in the memory to execute the method described above.

[0014] In a fourth aspect of the embodiments of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the method described above is implemented.

[0015] In the present invention, by collecting and analyzing the interaction data of users in the cultural and tourism scenarios, the personalized and intelligent control of the lighting art effects is realized, effectively enhancing the immersive experience of users, avoiding the disadvantages of single and static traditional lighting art effects, enabling the lighting art effects to be dynamically adjusted in real time according to user behaviors and emotions, enhancing the interactivity and attractiveness of the cultural and tourism scenarios, determining the lighting fade coefficient by calculating the energy difference between adjacent spaces, ensuring the continuity and harmony of the lighting effects in space, improving the overall aesthetic feeling of the lighting art and the ability to create the space atmosphere, forming a closed-loop multi-modal embodied interaction system by iteratively adjusting the lighting control instructions and continuously optimizing the lighting effects, enabling the lighting art effects to continuously learn and adapt to user preferences, improving the self-improving ability of the system and user satisfaction, and providing an intelligent solution for the lighting art design of the cultural and tourism scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a schematic flowchart of the method for generating the cultural and tourism lighting art effects of multi-modal embodied interaction according to an embodiment of the present invention; Figure 2 is a schematic structural diagram of the system for generating the cultural and tourism lighting art effects based on multi-modal embodied interaction; Figure 3 is a schematic diagram for comparing the system performances of the cultural and tourism lighting art effect system based on multi-modal embodied interaction. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are only a part rather than all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without any creative efforts shall fall within the protection scope of the present invention.

[0018] The technical solutions of the present invention will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0019] Figure 1 is a schematic flowchart of the method for generating the cultural and tourism lighting art effects of multi-modal embodied interaction according to an embodiment of the present invention, as Figure 1 shown, the method includes: Collect the initial interaction data of users in the cultural and tourism scenario, analyze the temporal correlation, extract user behavior characteristics, construct a user behavior characteristic sequence, calculate the change trend of the user behavior characteristic sequence, obtain the light intensity, crowd density and cultural element information distribution of the cultural and tourism scenario, and generate a scenario interaction data stream by combining the user behavior characteristic sequence and the corresponding change trend; Divide the interaction space based on the scenario interaction data stream, calculate the activity and emotional intensity of users in each space, generate a space vitality map, calculate the energy difference between adjacent spaces according to the space vitality map, determine the light fade coefficient, and track the real-time change of space vitality based on the light fade coefficient to establish a dynamic mapping relationship of the lighting effect; Output and execute the lighting control instruction according to the dynamic mapping relationship, generate the initial lighting effect, and collect the interactive feedback data of users on the initial lighting effect; Based on the interactive feedback data, extract the preference characteristics of users for the lighting effect, construct a lighting effect evaluation index, optimize and sort the lighting control parameters according to the lighting effect evaluation index, establish a parameter optimization strategy, and iteratively adjust the lighting control instruction based on the parameter optimization strategy to generate an optimal lighting control instruction; Output the optimal lighting control instruction, execute and record the optimal lighting effect.

[0020] In an alternative embodiment, Collect the initial interaction data of users in the cultural and tourism scenario, analyze the temporal correlation, extract user behavior characteristics, construct a user behavior characteristic sequence, calculate the change trend of the user behavior characteristic sequence, obtain the light intensity, crowd density and cultural element information distribution of the cultural and tourism scenario, and generating a scenario interaction data stream by combining the user behavior characteristic sequence and the corresponding change trend includes: Collect the initial interaction data of users in the cultural and tourism scenario, where the initial interaction data includes posture information, movement trajectory information, voice commands, emotional intonation information, heart rate data, and facial expression data; Construct a correlation matrix in the time dimension and extract the temporal correlation of the initial interaction data. Construct a user behavior characteristic sequence according to the temporal correlation. The user behavior characteristic sequence includes behavior duration, behavior occurrence frequency, behavior intensity change, and behavior spatial position, and calculate the change trend of the user behavior characteristic sequence at different scales; Collect the light intensity, crowd density and cultural element information distribution of the cultural and tourism scenario, map the user behavior characteristic sequence to the corresponding scenario grid of the cultural and tourism scenario according to the spatial position, determine the behavior activity of each grid, calculate the activity difference value between different grids, and combine the change trend to determine the scenario interaction data stream, where the scenario interaction data stream includes the real-time state and predicted state of the interaction data.

[0021] Collect the initial interaction data of users in the cultural and tourism scenario, including users' posture information, movement trajectory information, voice commands, and emotional intonation information.

[0022] After obtaining the initial interaction data, construct a correlation matrix in the time dimension and extract the temporal correlation of the initial interaction data. Divide the collected data according to time windows, with each time window set to 5 seconds and adjacent windows overlapping by 2 seconds. For the data within each time window, calculate the correlation coefficients between different data types and construct a 6×6 correlation matrix. For example, when analyzing a user watching an ancient building display, the correlation between facial expression data and heart rate data is 0.78, indicating a high consistency between the user's emotions and physiological responses; the correlation between posture information and movement trajectory is 0.65, indicating a moderate correlation between the user's posture changes and movement behaviors.

[0023] Based on the temporal correlation, construct a user behavior feature sequence, including behavior duration, behavior occurrence frequency, behavior intensity change, and behavior spatial location. Identify that the behavior duration of the user standing and watching in the cultural relic display area is on average 120 seconds; the frequency of the user issuing the "details" voice command is 5 times per hour; the maximum change in the user's heart rate occurs when watching a performance, with an average fluctuation of 15 times per minute; the main activity area of the user is concentrated in the central area of the exhibition hall, with the coordinate range from (35.2, 42.8) to (68.5, 75.3). Calculate the change trends of the user behavior feature sequence at different time scales, including 10-minute, 30-minute, and 60-minute scales. The analysis shows that the duration of the user standing and watching shows a downward trend at the 60-minute scale, dropping from the initial 120 seconds to 85 seconds; the frequency of voice commands shows a trend of rising first and then falling at the 30-minute scale; the heart rate change is highly correlated with the display content at all time scales.

[0024] Collect the environmental information of the cultural and tourism scenario, including light intensity, crowd density, and the distribution of cultural element information. The light intensity is collected by light sensors arranged throughout the scenario, with the numerical range from 0 to 1000 lux; the crowd density is calculated by a video analysis system, and the density levels are divided into low (0 - 0.2 people per square meter), medium (0.2 - 0.5 people per square meter), and high (>0.5 people per square meter); the distribution of cultural element information includes the type of exhibits, historical value level, and interactivity level, which are obtained through the preset information of the scenario.

[0025] Map the user behavior feature sequence to the corresponding scene grid in the cultural and tourism scene according to the spatial position. The scene is divided into grid cells of 10 meters × 10 meters, forming a 15×20 grid matrix in total. Calculate the behavior activity score (0 - 100 points) of each grid according to the stay time, interaction frequency, emotional change, and physiological index change of the user in each grid. For example, in the ancient clothing display area (grid coordinates [5, 8]), the user behavior activity is 85 points; in the rest area (grid coordinates [12, 3]), the user behavior activity is 25 points. Calculate the activity difference value between different grids. The average difference value between adjacent grids is 18 points, and the maximum difference value is 65 points (between the cultural relic display area and the aisle area).

[0026] Combine the change trend of the user behavior feature sequence and the grid activity difference value to generate a scene interaction data stream, including the real-time state and predicted state of the interaction data. The real-time state shows the active degree, emotional state, and interaction tendency of the current user in each grid; the predicted state predicts the possible behavior changes of the user within the next 30 minutes according to historical data and the current trend. For example, it is predicted that when the light intensity decreases by 10%, the stay time of the user in the display area will increase by 15%; when the pedestrian flow density increases by 0.2 people per square meter, the voice interaction frequency of the user will decrease by 20%.

[0027] In this embodiment, by collecting multi-dimensional initial interaction data, a comprehensive perception of the user interaction behavior is realized, a time-dimensional correlation matrix is constructed to extract the time-series correlation, and through the multi-scale analysis of the user behavior feature sequence, the dynamic changes of the user behavior pattern are effectively captured. By mapping the user behavior feature sequence to the scene grid and combining environmental factors, a refined scene-behavior association model is established, realizing the real-time state perception and predicted state evaluation of the scene interaction data.

[0028] In an alternative implementation manner, Divide the interaction space based on the scene interaction data stream, calculate the activity and emotional intensity of the user in each space and generate a spatial vitality map, calculate the energy difference between adjacent spaces according to the spatial vitality map, determine the light fade coefficient, and track the real-time change of the spatial vitality based on the light fade coefficient to establish a dynamic mapping relationship of the lighting effect, including: Divide the cultural and tourism scene into multiple interaction spaces based on the scene interaction data stream, establish a three-dimensional coordinate system in the interaction space, divide the interaction space into multiple grid cells, and calculate the ratio of the number of interaction events in each grid cell to the grid volume to obtain the interaction density; Calculate the weighted sum of the basic behavior intensity, movement trajectory complexity, and interaction density of users in each of the interaction spaces to obtain the user activity level. Calculate the weighted sum of the physiological characteristic emotion value, voice emotion value, and gesture emotion value of users in each of the interaction spaces to obtain the emotion intensity. Perform a linear weighted combination of the user activity level and the emotion intensity to obtain a spatial vitality map; Determine the spatial vitality corresponding to each interaction space according to the spatial vitality map. Calculate the absolute difference between the central point vitality values of adjacent interaction spaces to obtain an energy difference. Substitute the energy difference into an exponential function to obtain a lighting fade coefficient, where the input of the exponential function is the negative of the ratio of the energy difference to a pre-set energy threshold parameter; Track the real-time changes in spatial vitality based on the lighting fade coefficient, extract crowd flow characteristics to calculate dynamic partition boundaries, construct a cross-region fade function to optimize the lighting fade coefficient, identify group behavior patterns to predict the trend of vitality changes and perform error compensation, and establish a dynamic mapping relationship between the real-time changes in spatial vitality and lighting effects.

[0029] Figure 2 It is a schematic structural diagram of a cultural and tourism lighting art effect generation system based on multi-modal embodied interaction. As Figure 2 shown, obtain the pre-collected scene interaction data stream, use a spatial clustering algorithm to analyze the interaction data, and divide the entire cultural and tourism scene into multiple interaction spaces according to the spatial distribution characteristics of interaction events. In each interaction space, establish a three-dimensional rectangular coordinate system with the origin set at the geometric center of the space, and the X-axis, Y-axis, and Z-axis corresponding to the east-west, north-south, and vertical directions respectively. Further divide each interaction space into grid cells with a size of 0.5 m × 0.5 m × 0.5 m, calculate the number of interaction events in each grid cell and divide it by the grid volume (0.125 cubic meters) to obtain the interaction density. For example, if 15 interaction events are recorded in a certain grid cell, the interaction density of this grid cell is 15 / 0.125 = 120 times per cubic meter.

[0030] For the calculation of user activity in each interaction space, three factors are comprehensively considered: basic behavior intensity, movement trajectory complexity, and interaction density. The basic behavior intensity is obtained by analyzing the amplitude and frequency of the user's body movements. For example, a stationary state is assigned a value of 0.2, a walking state is assigned 0.5, a running state is assigned 0.8, and a jumping state is assigned 1.0. The movement trajectory complexity is obtained by calculating the curvature and turning frequency of the user's movement path. For example, the complexity of a straight-line movement is 0.3, the complexity of a curved movement is 0.6, and the complexity of frequent turning is 0.9. The system assigns weights of 0.4, 0.3, and 0.3 to the basic behavior intensity, movement trajectory complexity, and interaction density respectively, and performs a weighted sum to obtain the user activity. For example, if a user is in a walking state (0.5), the trajectory is a curved movement (0.6), and the interaction density of the grid cell where the user is located is 120 times per cubic meter (normalized to 0.7), then the activity of this user is 0.5×0.4 + 0.6×0.3 + 0.7×0.3 = 0.59.

[0031] The calculation of emotional intensity comprehensively considers physiological characteristic emotional value, voice emotional value, and gesture emotional value. The physiological characteristic emotional value is determined by the degree of preference feedback by the user. The voice emotional value is obtained by analyzing the pitch, volume, and speech rate of the user's voice. For example, a calm intonation is assigned 0.2, an excited intonation is assigned 0.7, and a cheering intonation is assigned 0.9. The gesture emotional value is obtained by recognizing the user's gesture actions. For example, a pointing gesture is assigned 0.4, a waving gesture is assigned 0.6, and a raising-hands-cheering gesture is assigned 0.9. The system assigns weights of 0.4, 0.3, and 0.3 to the physiological characteristic emotional value, voice emotional value, and gesture emotional value respectively, and performs a weighted sum to obtain the emotional intensity. For example, if a user's heart rate is 90 bpm (0.6), uses an excited intonation (0.7), and makes a waving gesture (0.6), then the emotional intensity of this user is 0.6×0.4 + 0.7×0.3 + 0.6×0.3 = 0.63.

[0032] The spatial vitality map is generated by linearly weighting and combining the user activity and emotional intensity, with weights of 0.5 and 0.5 respectively. For example, if the activity of the above user is 0.59 and the emotional intensity is 0.63, then the spatial vitality value of this user is 0.59×0.5 + 0.63×0.5 = 0.61. Calculate the average spatial vitality value of all users in the interaction space as the overall spatial vitality of this interaction space. For example, if there are 5 users in an interaction space, and their spatial vitality values are 0.61, 0.58, 0.72, 0.65, and 0.53 respectively, then the overall spatial vitality of this interaction space is (0.61 + 0.58 + 0.72 + 0.65 + 0.53) / 5 = 0.618.

[0033] Based on the generated spatial vitality map, calculate the energy difference between adjacent interaction spaces. The absolute difference in the vitality values of the central points of adjacent interaction spaces is used as the energy difference. For example, if the vitality value of the central point of space A is 0.618 and that of the adjacent space B is 0.452, then the energy difference between them is |0.618 - 0.452| = 0.166. The system substitutes the energy difference into the exponential function to calculate the lighting fade coefficient. The exponential function is defined as the negative power of e, and the power value is the ratio of the energy difference to the preset energy threshold parameter. If the energy threshold parameter is set to 0.2, then the lighting fade coefficient is e^(-(0.166 / 0.2)) = e^(-0.83) ≈ 0.436.

[0034] Use the calculated lighting fade coefficient to track the real-time changes in spatial vitality. By analyzing the crowd flow characteristics in the monitoring data, the system updates the dynamic partition boundary in real time. For example, when an increase in the flow of people is detected between two originally independent interaction areas, the system adjusts the boundary between these two areas to form a transition area. The system constructs a cross-region fade function to create a smooth lighting effect transition between adjacent regions. For example, if the lighting brightness of two adjacent regions is 80% and 50% respectively, and the fade coefficient is 0.436, then the lighting brightness in the transition area will smoothly change from 80% to 50%, and the change rate is controlled by the fade coefficient.

[0035] Table 1: Experimental results table of dynamic mapping of lighting effects;

[0036] As shown in Table 1, it systematically presents the mapping relationship between spatial vitality values and lighting parameters and the user experience evaluation. In the five main interaction spaces and two transition areas, there is an obvious positive correlation between spatial vitality values and lighting brightness, color temperature, and change rate.

[0037] Spaces with high vitality values (such as space 3 with a vitality value of 0.723) correspond to higher lighting brightness (85%), color temperature (4500K), and change rate (7.2 times per minute), and obtain the highest user satisfaction score (9.1 points). In contrast, spaces with low vitality values (such as space 4 with a vitality value of 0.385) use softer lighting parameters (brightness 50%, color temperature 3500K), and the change rate is reduced (3.8 times per minute).

[0038] Of particular note are the data for the transition areas, which demonstrate that the system has successfully achieved a smooth lighting transition effect. All parameters of transition area 1 - 2 are between those of space 1 and space 2, reflecting the effective application of the lighting fade coefficient. The user satisfaction score results (average 8.3 points) indicate that the dynamic mapping technology of lighting effects has a significant effect in enhancing the immersive experience in cultural and tourism scenarios.

[0039] Identify group behavior patterns through machine learning algorithms to predict the trend of vitality changes. For example, when it is identified that a group of tourists is approaching a certain scenic spot, it is predicted that the vitality value of this area will increase within a short period of time. At the same time, error compensation is carried out by comparing the predicted value with the actually collected data to continuously optimize the accuracy of the prediction model, and a dynamic mapping relationship between the real-time change of spatial vitality and the lighting effect is established, including parameters such as brightness, color temperature, color, and change rate, so that the lighting effect can adaptively respond to the change of spatial vitality and create an immersive lighting experience for the cultural and tourism scene.

[0040] In this embodiment, through the quantitative calculation of the interaction density of grid cells, the refined division of the interaction space is realized, providing a basic measurement standard for vitality assessment. Combining user activity, a spatial vitality map is constructed to comprehensively describe the spatial vitality characteristics. Through the calculation of energy difference and exponential function mapping, the lighting gradient coefficient is obtained, and a quantitative correlation between spatial vitality and lighting effect is established to achieve a smooth transition of the lighting effect. Based on the dynamic analysis of crowd flow characteristics and group behavior patterns, a cross-region gradient function and an error compensation mechanism are constructed to ensure the real-time response accuracy of the lighting effect to the change of spatial vitality.

[0041] In an alternative embodiment, Based on the lighting gradient coefficient, track the real-time change of spatial vitality, extract crowd flow characteristics to calculate the dynamic partition boundary, construct a cross-region gradient function to optimize the lighting gradient coefficient, identify group behavior patterns to predict the trend of vitality changes and perform error compensation, and establish a dynamic mapping relationship between the real-time change of spatial vitality and the lighting effect, including: Obtain the real-time change data of spatial vitality, generate a vitality mapping reference value according to the real-time change data of spatial vitality and the preset lighting gradient coefficient, collect the crowd movement data in the interaction space, extract the crowd flow characteristics, and multiply the crowd flow characteristics by the crowd density in the interaction space to calculate the flow intensity; Generate an adjustment factor for the dynamic partition boundary according to the difference in crowd flow characteristics and the difference in crowd density between adjacent interaction spaces, and adjust the boundary of adjacent interaction spaces according to the adjustment factor; Construct the cross-region gradient function of the adjacent interaction spaces, substitute the flow intensity and the spatial distance into the cross-region gradient function, and fuse the calculation result with the preset lighting gradient coefficient to generate an optimized lighting gradient coefficient; Extract group behavior pattern features based on the flow intensity, input the group behavior pattern features and the pre-collected historical vitality data into a prediction model to calculate the trend of vitality changes, and perform error compensation on the trend of vitality changes and the real-time collected spatial vitality data, and output a compensated spatial vitality mapping value; A dynamic mapping function is established based on the compensated space vitality mapping value, the optimized light gradient coefficient and pre-set environmental constraint parameters, and a dynamic mapping relationship is established between the real-time changes of the space vitality and the lighting effects. The dynamic mapping relationship is dynamically adjusted with the change rate of the compensated space vitality mapping value and the change rate of the optimized light gradient coefficient.

[0042] In the interactive space, the sensor network collects spatial vitality data in real time, including indicators such as traffic, dwell time, and movement speed. The preset initial light gradient coefficient matrix is [0.3, 0.5, 0.7, 0.9], corresponding to different vitality levels. After the collected vitality data is normalized, it is weighted averaged with the preset light gradient coefficient to generate the vitality mapping benchmark value. For example, the vitality value of a certain area is 0.65, the corresponding light gradient coefficient is 0.7, and the generated vitality mapping benchmark value is 0.675.

[0043] The camera array installed in the interactive space collects crowd movement data at a frequency of 25 frames per second. The crowd flow characteristics, including the flow direction vector and the flow speed scalar, are extracted through image processing algorithms. The area is divided into 10×10 grid cells, and the crowd density of each grid is calculated. 15 people are detected in a grid cell with an area of 25 square meters and a crowd density of 0.6 people / square meter. The crowd flow speed of 0.8 meters / second in this grid is multiplied by the crowd density of 0.6 people / square meter, and the flow intensity is 0.48.

[0044] The dynamic partition boundary adjustment of adjacent interactive spaces is based on the difference in crowd flow characteristics and crowd density. The flow characteristic difference between adjacent areas A and B is calculated with a period of 5 seconds. If the flow speed in area A is 0.9 m / s and that in area B is 0.6 m / s, the difference is 0.3 m / s; if the crowd density in area A is 0.7 people / m2 and that in area B is 0.5 people / m2, the difference is 0.2 people / m2. Multiply these two differences by the weight coefficient [0.6, 0.4] to obtain an adjustment factor of 0.26. When the adjustment factor is greater than the preset threshold of 0.25, the boundary is adjusted 2 meters toward area B with lower crowd density to achieve dynamic partitioning.

[0045] The cross-region gradient function is constructed using an exponential decay model. The flow intensities of adjacent regions A and B are 0.48 and 0.36 respectively, and the actual distance between the regions is 15 meters. Substituting the flow intensity and spatial distance into the cross-region gradient function, the attenuation coefficient is calculated to be 0.08. The original preset light gradient coefficient is [0.3, 0.5, 0.7, 0.9]. After being integrated with the attenuation coefficient, the optimized light gradient coefficient is adjusted to [0.324, 0.54, 0.756, 0.972].

[0046] The extraction of group behavior pattern features is based on the temporal variation of flow intensity. Taking 30 minutes as an observation window, the variation pattern of flow intensity is analyzed. The extracted features include peak intensity, fluctuation frequency, duration, etc. The temporal sequence data of the flow intensity in a certain area from 12:00 to 12:30 on a weekday is [0.25, 0.37, 0.52, 0.68, 0.72, 0.65, 0.48, 0.31], and the extracted features are: peak intensity 0.72, fluctuation frequency 0.0667Hz, and duration 30 minutes. The features and historical vitality data are jointly input into the prediction model to predict the vitality change trend in the next 30 minutes. The prediction result is [0.28, 0.42, 0.58, 0.65, 0.60, 0.52, 0.39, 0.27].

[0047] The real-time collected spatial vitality data is [0.30, 0.45, 0.60, 0.68, 0.63, 0.54, 0.40, 0.29]. Calculate the error with the prediction result to obtain the error sequence [0.02, 0.03, 0.02, 0.03, 0.03, 0.02, 0.01, 0.02]. Apply the error compensation algorithm to fuse the error sequence with the prediction result and output the compensated spatial vitality mapping value [0.29, 0.44, 0.59, 0.67, 0.62, 0.53, 0.40, 0.28].

[0048] During the establishment of the dynamic mapping function, the compensated spatial vitality mapping value, the optimized light fading coefficient, and the environmental constraint parameters are combined. The environmental constraint parameters include the maximum illuminance value of 300 lux, the minimum illuminance value of 50 lux, and the color temperature range of 3000K - 5000K. The dynamic mapping function converts the compensated spatial vitality mapping value of 0.67 into the corresponding light parameters: illuminance value of 218 lux, color temperature of 4200K, and light fading time of 1.5 seconds. When it is detected that the change rate of the spatial vitality mapping value exceeds 15%, the change rate of the light fading coefficient is dynamically adjusted. For example, when the vitality mapping value rapidly drops from 0.67 to 0.53 (change rate 20.9%), the change rate of the light fading coefficient is automatically increased from the standard 10% to 16% to make the light effect more sensitive to the change of spatial vitality.

[0049] In this embodiment, by introducing the comprehensive analysis of crowd flow characteristics and density, a dynamic boundary adjustment mechanism based on flow intensity is established, effectively solving the problem of boundary mutation caused by traditional fixed zoning. The dynamic mapping mechanism enables the light effect to be adaptively adjusted according to the change rate of spatial vitality and the change rate of the fading coefficient, ensuring both a sensitive response of the light effect to the change of spatial vitality and avoiding overly frequent light adjustment; In the prior art, the lighting effects of interactive spaces are usually adjusted using fixed mapping relationships or simple linear transformations, which cannot accurately respond to the dynamic changes in spatial vitality. In addition, sudden changes in lighting effects are likely to occur at the boundaries of spatial partitions, affecting the continuity of the lighting experience. There is a lack of in-depth analysis of the patterns of crowd activities, making it difficult to predict the changing trend of spatial vitality, resulting in a lag in lighting adjustment and an inability to adapt to the rapid changes in spatial vitality in a timely manner. This embodiment achieves a smooth transition of lighting effects between adjacent spaces by constructing a cross-region gradient function, thereby improving the continuity of the lighting experience. It combines real-time data for error compensation, significantly improving the real-time and accuracy of lighting adjustment. Through the fusion analysis and dynamic optimization of multi-dimensional data, it achieves intelligent linkage between lighting effects and spatial vitality, improves the lighting system's ability to adapt to changes in spatial vitality, and provides a new technical approach to improving the lighting experience of interactive spaces.

[0050] Figure 3 This is a performance comparison diagram of the cultural tourism lighting art effect system based on multimodal embodied interaction. This system shows significant advantages in all evaluation dimensions. In terms of boundary transition smoothness, this system reaches 78.2%, which is much higher than the 32.5% of traditional fixed mapping and 48.8% of simple dynamic adjustment, indicating that this system can effectively solve the problem of sudden changes in lighting effects at the boundaries of space. In terms of the vitality change response time indicator, the 84.3% of this system reflects its sensitive response ability to changes in spatial vitality, which is about 33 percentage points higher than the traditional method.

[0051] The prediction accuracy is particularly outstanding. By introducing group behavior pattern feature analysis and error compensation mechanism, this system has increased the prediction accuracy to 86.2%, which is more than twice the traditional fixed mapping method (39.6%). The final user satisfaction score also confirmed the success of converting technical advantages into improved experience. This system achieved a high satisfaction rate of 87.3%, significantly better than other solutions.

[0052] In an optional embodiment, Outputting and executing lighting control instructions according to the dynamic mapping relationship, generating initial lighting effects and collecting interactive feedback data of users on the initial lighting effects include: Outputting and executing lighting control instructions according to the dynamic mapping relationship, wherein the lighting control instructions include lighting brightness, color temperature and gradient timing parameters; generating an initial lighting effect according to the lighting control instruction, wherein the initial lighting effect is dynamically adjusted as the vitality of the space changes in real time; Collect user interaction feedback data on the initial lighting effect, wherein the interaction feedback data includes user stay time, interaction frequency, and pre-set emotional feedback indicators.

[0053] Obtain spatial vitality data, including indicators such as crowd density, personnel activity frequency, and environmental noise. For example, use an infrared sensor array to detect the change in the number of people in a 10-meter by 10-meter space at a sampling frequency of 4 times per second; record the number of times of personnel position changes per unit time (such as per minute) through motion sensors; at the same time, use noise sensors to measure the environmental decibel value, and the noise threshold is set in the range of 45 - 75 decibels. When the crowd density increases from 0.1 person per square meter to 0.3 person per square meter, it indicates an increase in spatial vitality; when the environmental noise rises from 45 decibels to 60 decibels, it also indicates an increase in spatial activity.

[0054] When establishing the dynamic mapping relationship, map the spatial vitality data to the lighting parameters. When the crowd density is 0.1 person per square meter, the lighting brightness is set to 300 lumens; when the crowd density increases to 0.3 person per square meter, the brightness is correspondingly increased to 500 lumens. Similarly, when the environmental noise is 45 decibels, the lighting color temperature is set to 2700K (warm white light); when the noise rises to 60 decibels, the color temperature is adjusted to 4000K (neutral white light). When the personnel activity frequency is 1 - 2 position changes per minute, the lighting fade-in period is set to 30 seconds; when the frequency increases to 5 - 6 times per minute, the fade-in period is shortened to 10 seconds, providing a more active light effect change.

[0055] In the stage of outputting and executing the lighting control instructions, according to the mapping relationship, generate lighting control instructions including brightness, color temperature, and fade-in timing in real time. Transmit them to the lighting control module through a wireless communication protocol (such as ZigBee, Wi-Fi, or Bluetooth 5.0). For example, when it is detected that the crowd density is 0.2 person per square meter, the environmental noise is 55 decibels, and the activity frequency is 3 position changes per minute, generate the instruction: brightness 400 lumens, color temperature 3500K, fade-in period 20 seconds. After receiving the instruction, the control module controls the LED driving circuit through a PWM (pulse width modulation) signal, precisely adjusts the output power to achieve brightness control; realizes color temperature change by adjusting the ratio of cold and warm light LEDs; and executes the light effect conversion according to the specified fade-in timing parameters.

[0056] In the link of generating the initial lighting effect, adjust the lighting effect in real time according to the control instructions, so that it responds dynamically to the change of spatial vitality. When it is detected that the spatial vitality changes from low to high (such as the crowd density increases from 0.1 to 0.25 person per square meter within 5 minutes), gradually transition the lighting brightness from 300 lumens to 450 lumens, the color temperature rises from 2700K to 3500K, and the fade-in period shortens from 30 seconds to 15 seconds. This gradual change process avoids sudden lighting changes and provides a comfortable light environment adaptation process for users. Identify specific activity patterns, such as the situation where people gather but the activity frequency decreases at the start of a meeting. At this time, maintain a medium brightness (such as 400 lumens) but adjust to a color temperature suitable for the meeting (such as 4000K), and extend the fade-in period to 45 seconds to reduce interference.

[0057] In the stage of collecting user interaction feedback data, the reactions of users to the lighting effects are recorded through a variety of sensing devices. The user residence duration is measured by a position sensor to record the time that users stay under specific lighting effects. For example, when the lighting is set to 400 lumens and a color temperature of 3500K, the average residence time of users is 15 minutes; while under the setting of 500 lumens and 4000K, the average residence time is shortened to 8 minutes, indicating that users have a greater preference for the former setting. The interaction frequency records the number of times users adjust the lighting through the control panel, voice or gestures. The experimental data shows that in the dynamic response mode, the number of times users adjust the lighting per hour is reduced from 7-8 times in traditional fixed lighting to 2-3 times, proving that the automatic adjustment of the system better meets the needs of users.

[0058] The emotional feedback index is collected by a smart questionnaire system to collect subjective scores after the user experience, and a 1-10 point system is used to evaluate comfort.

[0059] In this embodiment, by establishing a dynamic mapping relationship to output lighting control instructions, the lighting effect realizes real-time response to the change of spatial vitality, enabling the lighting effect to accurately reflect the dynamic change characteristics of spatial vitality. Through the dynamic adjustment mechanism of the initial lighting effect, the lighting system can continuously track the change trend of spatial vitality, ensuring the synchronization of the lighting effect and the spatial vitality state, improving the continuity and smoothness of lighting control. Based on the collection of multi-dimensional interaction data such as user residence duration, interaction frequency and emotional feedback, an evaluation and feedback mechanism for lighting effects is established, providing effective data support for the optimization of lighting control strategies.

[0060] In an alternative embodiment, Based on the interaction feedback data, extract the preference characteristics of users for lighting effects, construct lighting effect evaluation indicators, optimize and sort the lighting control parameters according to the lighting effect evaluation indicators, establish a parameter optimization strategy, and iteratively adjust the lighting control instructions based on the parameter optimization strategy to generate the optimal lighting control instructions, including: Collect the interaction feedback data of users on lighting effects, construct a multi-dimensional feature vector according to the user residence duration, interaction frequency and emotional feedback index in the interaction feedback data, and input the multi-dimensional feature vector into a pre-set user preference scoring module to obtain a preference score; Weightedly fuse the preference score with the comfort index and the spatial coordination index to generate a comprehensive evaluation value, and construct a time-series evaluation matrix based on the comprehensive evaluation value. The time-series evaluation matrix includes a sequence of evaluation values within a historical time window; Calculate the sensitivity of the comprehensive evaluation value to the lighting control parameters, construct a parameter sensitivity distribution module using kernel density estimation, suppress the influence of abnormal samples through iteratively reweighted least squares, determine the parameter priority and sort them in combination with the historical optimization effect; Based on the sorted results of the lighting control parameters, construct parameter optimization objective functions in sequence. Take the minimization of the mean square error of the evaluation value sequence in the time series evaluation matrix as the optimization objective, and use the parameter change amplitude between adjacent moments as the constraint condition. Gradient-iteratively update the lighting control parameters based on the optimization objective function. Stop the iteration when the difference in the comprehensive evaluation value between adjacent moments is less than the preset threshold, and output the optimized lighting control parameters; According to the optimized lighting control parameters and the pre-set environmental constraint parameters, generate the optimal lighting control instruction in combination with the pre-set dynamic gain matrix.

[0061] Collect interactive feedback data on users' feedback on the lighting effect through questionnaires, including user residence duration, interaction frequency, and emotional feedback indicators. For example, if the user adjusts the lighting less frequently under a certain lighting effect, it indicates that this lighting effect better meets the user's needs. For example, if the user feedback shows an increase in the smiling frequency under a specific lighting, a positive score is given.

[0062] Construct the collected data into a multi-dimensional feature vector. The residence duration is in seconds, ranging from 0 to 3600 seconds; the interaction frequency is the number of operations per hour, ranging from 0 to 100 times; the emotional feedback indicator ranges from -1 to 1, where 1 represents extreme satisfaction and -1 represents extreme dissatisfaction. In practical applications, for an office scenario, if the user's residence duration is 2 hours, the interaction frequency is 5 times per hour, and the emotional feedback indicator is 0.8 under a lighting environment with a color temperature of 5000K and a brightness of 500 lux, the constructed feature vector is [7200, 5, 0.8].

[0063] Input the constructed multi-dimensional feature vector into the pre-set user preference scoring module, and use a scoring algorithm based on support vector machines to map the input feature vector to a preference score interval of 0 - 100. In the training stage, use the labeled user feedback data set to train the model. For example, the feature vector [7200, 5, 0.8] may be mapped to a high preference score of 85 points.

[0064] After obtaining the preference score, it is further comprehensively evaluated by combining objective lighting effect evaluation indicators. The comfort index is calculated based on the color temperature, brightness, and the matching degree with the ambient brightness, ranging from 0 to 100; the spatial coordination index measures the uniformity and layering of the lighting in the space, also ranging from 0 to 100. These three indicators are weighted and fused, with the weight distribution being: preference score 0.5, comfort index 0.3, and spatial coordination index 0.2, to generate a comprehensive evaluation value. For example, if the preference score is 85, the comfort index is 90, and the spatial coordination index is 80, then the comprehensive evaluation value is 85×0.5 + 90×0.3 + 80×0.2 = 85.5.

[0065] Organize the comprehensive evaluation values within a continuous time period into a time-series evaluation matrix, recording the evaluation values every 30 minutes within the past 24 hours to form an evaluation sequence of 48 time points.

[0066] For the optimization of lighting control parameters, calculate the sensitivity of the comprehensive evaluation value to each control parameter. The control parameters include brightness (range 100 - 1000 lux), color temperature (range 2700 - 6500 K), irradiation angle (range 0 - 180 degrees), etc. By making small adjustments to the parameters and observing the change rate of the comprehensive evaluation value to determine the sensitivity. For example, when the brightness is adjusted from 500 lux to 550 lux, if the comprehensive evaluation value changes from 85.5 to 88.0, then the sensitivity is 0.05 (an increase in the evaluation value of approximately 0.05 for every 1% change).

[0067] Adopt the kernel density estimation method to construct a parameter sensitivity distribution model, assign weights to each parameter sensitivity sample in the historical data, and generate a continuous sensitivity distribution curve. To reduce the influence of abnormal samples, the system applies the iteratively reweighted least squares method to dynamically adjust the sample weights according to the deviation degree of the samples from the central tendency. For example, if most of the data shows that the brightness sensitivity is concentrated in the range of 0.04 - 0.06, while individual samples show 0.2, then the system will reduce the weights of these abnormal samples.

[0068] Based on the parameter sensitivity analysis and historical optimization effects, determine the parameter priorities and sort them. In practical applications, if the analysis results show that the color temperature sensitivity is 0.08, the brightness sensitivity is 0.05, and the irradiation angle sensitivity is 0.03, then the optimization order is color temperature, brightness, irradiation angle.

[0069] According to the sorting results, construct parameter optimization objective functions one by one. Taking the color temperature as an example, the objective function aims to minimize the mean square error of the evaluation value sequence in the time-series evaluation matrix, while restricting the change in color temperature between adjacent moments not to exceed 200 K to avoid abrupt changes affecting the user experience. Use the gradient descent method to iteratively update the parameter values, and stop the iteration when the difference in the comprehensive evaluation values between adjacent iteration times is less than 0.5.

[0070] In this embodiment, the efficient integration of user interaction data is achieved through the construction of multi-dimensional feature vectors, which can comprehensively capture the user's experience feedback on the lighting effects. By integrating the user preference scores with the comfort index and the spatial coordination index, a comprehensive evaluation system that takes into account various requirements is established, improving the scientificity and reliability of the evaluation results. Through the construction of the time-series evaluation matrix, the evaluation information in the historical optimization process is retained, providing a more complete decision-making basis for parameter optimization. The sensitivity analysis method based on kernel density estimation improves the accuracy of parameter importance evaluation. The interference effect of abnormal samples is effectively reduced through the iterative reweighting mechanism, making the parameter priority ranking more stable and reliable. The optimization strategy based on minimizing the mean square error of the evaluation value sequence ensures the stability of the lighting effects and avoids drastic fluctuations by restricting the change range of parameters at adjacent times.

[0071] In an alternative embodiment, Calculate the sensitivity of the comprehensive evaluation value to the lighting control parameters. Use kernel density estimation to construct a parameter sensitivity distribution module, suppress the influence of abnormal samples through iterative reweighted least squares, and determine and sort the parameter priorities in combination with the historical optimization effects, including: Obtain the historical adjustment data of the lighting control parameters and the corresponding preference scores. Based on the historical adjustment data and the corresponding preference scores, calculate the first-order sensitivity of the comprehensive evaluation value to the lighting control parameters through partial derivatives, calculate the second-order interaction sensitivity through mixed partial derivatives, and construct a Jacobian matrix in combination with the central difference method to obtain an initial sensitivity sequence; Use a Gaussian kernel function to perform kernel density estimation on the initial sensitivity sequence, adaptively calculate the optimal bandwidth parameter based on the sample standard deviation, transform the initial sensitivity sequence through the kernel function to obtain a continuous probability distribution, and output the distribution characteristics of the parameter sensitivity; Construct a Huber-type M-estimation loss function to perform anomaly detection on the distribution characteristics of the parameter sensitivity, adaptively calculate the sample weight matrix based on the residual size, apply the sample weight matrix to iterative reweighted least squares calculation, suppress the influence of abnormal samples through multiple rounds of iterative optimization, and output the corrected sensitivity value; Calculate the time-series weight of the pre-obtained historical evaluation data based on the time decay exponential function, and perform weighted summation of the time-series weight and the historical comprehensive evaluation value to obtain the historical optimization effect of the parameter; Perform multi-objective weighted fusion on the corrected sensitivity value, the historical optimization effect of the parameter, and the parameter stability index to determine the parameter priority, sort the lighting control parameters in descending order according to the parameter priority, and generate a parameter optimization order table, where the parameter stability index includes the parameter fluctuation range, the convergence speed, and the disturbance recovery ability.

[0072] Collect historical data on lighting control, including parameters such as brightness, color temperature, and illumination, and corresponding user preference scores. The scores are based on the subjective evaluation of the lighting effects under different parameter combinations by the experimental participants on a 1-10 scale.

[0073] Process the collected parameter data matrix P and the corresponding score matrix S, where P contains n groups of parameter records, each group contains m control parameters, such as {brightness: 80%, color temperature: 4500K, illumination: 650lux}. Calculate the first-order sensitivity for each parameter separately, and achieve it by calculating the partial derivative of the comprehensive evaluation value to the parameter. For example, when calculating the sensitivity of the brightness parameter, change the brightness from 70% to 90%, record the score from 7.2 to 8.4, and get a preliminary sensitivity of 0.06. Use the central difference method to make small positive and negative perturbations to the parameters and record the changes in the evaluation values. For example, if the brightness fluctuates by 5% at the 85% point, the score changes by ±0.3, and the sensitivity of the point is 0.03.

[0074] When calculating the second-order interaction sensitivity, adjust two parameters at the same time to observe the score changes. For example, adjust the brightness and color temperature at the same time, record the score changes, and obtain the interaction sensitivity matrix. Summarize all the calculation results and construct a complete Jacobian matrix. Each element in the matrix represents the sensitivity value of the corresponding parameter, such as brightness 0.042, color temperature 0.035, illuminance 0.028, etc., to form the initial sensitivity sequence.

[0075] In order to obtain more accurate sensitivity distribution characteristics, the kernel density estimation method is used. The Gaussian kernel function is selected as the basic kernel function to transform the initial sensitivity sequence. The bandwidth parameter h is adaptively calculated and determined based on the sample standard deviation. For example, if the sensitivity sample standard deviation is 0.018, the bandwidth parameter is set to 0.015. The kernel function transformation is applied to the sensitivity sequence of each parameter to convert the discrete sensitivity value into a continuous probability distribution. For example, the peak of the sensitivity distribution of the brightness parameter appears near 0.045, indicating that the parameter has the most significant effect near this value.

[0076] Identify and process the impact of abnormal samples on sensitivity calculation, and construct the Huber type M-estimation loss function. Set the threshold k to 1.345. When the absolute value of the residual is less than k, use the square loss, and when it is greater than k, use the linear loss. The initial fitting results in a brightness sensitivity of 0.042. The residual of an abnormal sample is 2.6. The system calculates the sample weight of 0.52 based on the Huber function. Construct the sample weight matrix W, and the diagonal elements are the weights of each sample, such as diag(1.0, 0.95, 0.52, ...).

[0077] Apply the iterative reweighted least squares method to update the parameter sensitivity estimate in each iteration. After the first iteration, the brightness sensitivity is adjusted from 0.042 to 0.039, and converges to 0.038 after 5 iterations, which is determined as the final correction value. Similarly, the corrected sensitivity values of other parameters are obtained, such as the color temperature of 0.033, the illuminance of 0.027, etc.

[0078] Considering the historical optimization effect, introduce the time decay exponential function to calculate the weights of historical data. The weights of recent data are high, and the weights of long-term data are low. For example, the weight of data in the most recent week is 0.8, the weight of data one month ago is 0.5, and the weight of data three months ago is 0.2. Combine the time series weights with the historical evaluation values to calculate the historical optimization scores of the brightness parameter as 8.2, the color temperature as 7.8, and the illuminance as 7.5.

[0079] Evaluate the parameter stability indicators, including the ability of the system to return to a stable state after parameter adjustment. Measure that the fluctuation range of the brightness parameter is ±3%, the convergence time is 1.2 seconds, and the disturbance recoverability is 0.92; the fluctuation range of the color temperature is ±150K, the convergence time is 1.5 seconds, and the disturbance recoverability is 0.88; the fluctuation range of the illuminance is ±50 lux, the convergence time is 1.8 seconds, and the disturbance recoverability is 0.85.

[0080] Perform multi-objective weighted fusion on the corrected sensitivity values, historical optimization effects, and stability indicators. Set the sensitivity weight to 0.5, the historical effect weight to 0.3, and the stability weight to 0.2. Calculate that the comprehensive priority of brightness is 0.81, the color temperature is 0.76, and the illuminance is 0.71. Sort the parameters in descending order according to the comprehensive priority to generate an optimization order table: brightness (0.81), color temperature (0.76), illuminance (0.71), color rendering index (0.68), projection angle (0.65). The lighting control system performs parameter optimization according to this order table, giving priority to adjusting high-priority parameters to improve the optimization efficiency and meet the user's light environment preferences.

[0081] In this embodiment, a joint analysis mechanism of first-order sensitivity and second-order interaction sensitivity is introduced. Through the construction of the Jacobian matrix, a comprehensive evaluation of the independent influence and interaction of parameters is realized. By introducing the Huber-type M-estimation loss function and the iterative reweighted least squares method, a robust outlier processing mechanism is established, effectively reducing the interference of outlier samples on the sensitivity calculation; In the prior art, the optimization ranking of lighting control parameters usually adopts a single sensitivity analysis method, only considering the simple linear relationship between parameters and evaluation indicators, ignoring the interaction effects between parameters, having limited ability to handle abnormal data, being easily affected by noise interference, resulting in unstable evaluation results of parameter importance, and at the same time being prone to ignoring the utilization of historical optimization experience and unable to effectively inherit the existing optimization results; In this embodiment, the kernel density estimation method is used to continuously process the sensitivity sequence, which improves the reliability of sensitivity analysis, makes the parameter importance evaluation more accurate. The design of the time-series weight makes full use of historical optimization experience, enabling the system to inherit and develop existing optimization results. By introducing stability indicators such as parameter fluctuation range, convergence speed, and disturbance recovery, a comprehensive parameter evaluation system is constructed, realizing the accurate evaluation and reliable ranking of the importance of lighting control parameters, providing a more scientific decision-making basis for subsequent parameter optimization, and improving the optimization efficiency and stability of the control effect of the lighting control system.

[0082] In the second aspect of the embodiment of the present invention, a multi-modal embodied interaction-based cultural and tourism lighting art effect generation system is provided, including: The first unit is used to collect the initial interaction data of users in the cultural and tourism scene, analyze the time-series correlation, extract user behavior characteristics, construct a user behavior characteristic sequence, calculate the change trend of the user behavior characteristic sequence, obtain the light intensity, crowd density, and cultural element information distribution of the cultural and tourism scene, and generate a scene interaction data stream in combination with the user behavior characteristic sequence and the corresponding change trend; The second unit is used to divide the interaction space based on the scene interaction data stream, calculate the activity and emotional intensity of users in each space, generate a space vitality map, calculate the energy difference between adjacent spaces according to the space vitality map, determine the light fade coefficient, track the real-time change of space vitality based on the light fade coefficient, and establish a dynamic mapping relationship of the light effect; The third unit is used to output and execute the lighting control instruction according to the dynamic mapping relationship, generate an initial lighting effect, and collect the interactive feedback data of users on the initial lighting effect; The fourth unit is used to extract the preference characteristics of users for the lighting effect based on the interactive feedback data, construct a lighting effect evaluation index, optimize and rank the lighting control parameters according to the lighting effect evaluation index, establish a parameter optimization strategy, and iteratively adjust the lighting control instruction based on the parameter optimization strategy to generate an optimal lighting control instruction; The fifth unit is used to output the optimal lighting control instruction, execute it, and record the optimal lighting effect.

[0083] In the third aspect of the embodiment of the present invention, an electronic device is provided, including: A processor and a memory for storing instructions executable by the processor, wherein the processor is configured to call the instructions stored in the memory to execute the method described above.

[0084] In the fourth aspect of the embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the method described above is implemented.

[0085] The present invention may be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having thereon computer-readable program instructions for performing various aspects of the present invention.

[0086] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for generating cultural and tourism lighting art effects of multimodal embodied interaction, characterized in that, Including: Collect the initial interaction data of users in the cultural and tourism scenario, analyze the temporal correlation, extract the user behavior characteristics, construct the user behavior characteristic sequence, calculate the change trend of the user behavior characteristic sequence, obtain the light intensity, crowd density and cultural element information distribution of the cultural and tourism scenario, and generate the scenario interaction data stream by combining the user behavior characteristic sequence and the corresponding change trend; Divide the interaction space based on the scenario interaction data stream, calculate the activity and emotional intensity of users in each space, generate the spatial vitality map, calculate the energy difference between adjacent spaces according to the spatial vitality map, determine the light fade coefficient, track the real-time change of spatial vitality based on the light fade coefficient, and establish the dynamic mapping relationship of the light effect; Output the light control instruction according to the dynamic mapping relationship and execute it to generate the initial light effect and collect the interactive feedback data of users on the initial light effect; Based on the interactive feedback data, extract the preference characteristics of users for the light effect, construct the light effect evaluation index, optimize and sort the light control parameters according to the light effect evaluation index, establish the parameter optimization strategy, and iteratively adjust the light control instruction based on the parameter optimization strategy to generate the optimal light control instruction; Output the optimal light control instruction, execute it and record the optimal light effect.

2. The method according to claim 1, characterized in that Collect the initial interaction data of users in the cultural and tourism scenario, analyze the temporal correlation, extract the user behavior characteristics, construct the user behavior characteristic sequence, calculate the change trend of the user behavior characteristic sequence, obtain the light intensity, crowd density and cultural element information distribution of the cultural and tourism scenario, and generate the scenario interaction data stream including: Collect the initial interaction data of users in the cultural and tourism scenario, where the initial interaction data includes posture information, movement trajectory information, voice commands and emotional intonation information data; Construct the correlation matrix in the time dimension and extract the temporal correlation of the initial interaction data. According to the temporal correlation, construct the user behavior characteristic sequence, where the user behavior characteristic sequence includes behavior duration, behavior occurrence frequency, behavior intensity change and behavior spatial position, and calculate the change trend of the user behavior characteristic sequence at different scales; Collect the light intensity, crowd density and cultural element information distribution of the cultural and tourism scenario, map the user behavior characteristic sequence to the corresponding scenario grid of the cultural and tourism scenario according to the spatial position, determine the behavior activity of each grid, calculate the activity difference value between different grids, and combine the change trend to determine the scenario interaction data stream, where the scenario interaction data stream includes the real-time state and predicted state of the interaction data.

3. The method according to claim 1, wherein Divide the interaction space based on the scenario interaction data stream, calculate the activity and emotional intensity of users in each space, generate the spatial vitality map, calculate the energy difference between adjacent spaces according to the spatial vitality map, determine the light fade coefficient, track the real-time change of spatial vitality based on the light fade coefficient, and establish the dynamic mapping relationship of the light effect including: Divide the cultural and tourism scene into multiple interactive spaces based on the scenario interaction data stream, establish a three-dimensional coordinate system in the interactive space, divide the interactive space into multiple grid cells, and calculate the ratio of the number of interaction events in each grid cell to the grid volume to obtain the interaction density; Calculate the weighted sum of the basic behavior intensity, movement trajectory complexity, and interaction density of users in each interactive space to obtain the user activity, calculate the weighted sum of the physiological characteristic emotion value, voice emotion value, and gesture emotion value of users in each interactive space to obtain the emotion intensity, and perform a linear weighted combination of the user activity and the emotion intensity to obtain the spatial vitality map; Determine the spatial vitality corresponding to each interactive space according to the spatial vitality map, calculate the absolute difference between the central point vitality values of adjacent interactive spaces to obtain the energy difference, and substitute the energy difference into an exponential function to obtain the lighting fade coefficient, where the input of the exponential function is the negative of the ratio of the energy difference to a preset energy threshold parameter; Track the real-time changes in spatial vitality based on the lighting fade coefficient, extract crowd flow characteristics to calculate the dynamic partition boundary, construct a cross-region fade function to optimize the lighting fade coefficient, identify group behavior patterns to predict the vitality change trend and perform error compensation, and establish a dynamic mapping relationship between the real-time changes in spatial vitality and the lighting effect.

4. The method according to claim 3, characterized in that Tracking the real-time changes in spatial vitality based on the lighting fade coefficient, extracting crowd flow characteristics to calculate the dynamic partition boundary, constructing a cross-region fade function to optimize the lighting fade coefficient, identifying group behavior patterns to predict the vitality change trend and performing error compensation, and establishing a dynamic mapping relationship between the real-time changes in spatial vitality and the lighting effect includes: Obtain the real-time change data of spatial vitality, generate a vitality mapping reference value according to the real-time change data of spatial vitality and the preset lighting fade coefficient, collect the crowd movement data in the interactive space, extract the crowd flow characteristics, and multiply the crowd flow characteristics by the crowd density in the interactive space to calculate the flow intensity; Generate an adjustment factor for the dynamic partition boundary according to the difference in crowd flow characteristics and the difference in crowd density between adjacent interactive spaces, and adjust the boundary between adjacent interactive spaces according to the adjustment factor; Construct a cross-region fade function for the adjacent interactive spaces, substitute the flow intensity and the spatial distance into the cross-region fade function, and fuse the calculation result with the preset lighting fade coefficient to generate an optimized lighting fade coefficient; Extract group behavior pattern features based on the flow intensity, input the group behavior pattern features and the previously collected historical vitality data into a prediction model to calculate the vitality change trend, and perform error compensation on the vitality change trend and the real-time collected spatial vitality data, and output the compensated spatial vitality mapping value; A dynamic mapping function is established based on the compensated spatial vitality mapping value, the optimized lighting fade coefficient, and the pre-set environmental constraint parameters, to establish a dynamic mapping relationship between the real-time changes of the spatial vitality and the lighting effects. The dynamic mapping relationship is dynamically adjusted according to the change rate of the compensated spatial vitality mapping value and the change rate of the optimized lighting fade coefficient.

5. The method according to claim 1, characterized in that Output and execute a lighting control instruction according to the dynamic mapping relationship to generate an initial lighting effect and collect interactive feedback data of the user on the initial lighting effect, including: Output and execute a lighting control instruction according to the dynamic mapping relationship. The lighting control instruction includes lighting brightness, color temperature, and fade timing parameters. Generate an initial lighting effect according to the lighting control instruction. The initial lighting effect is dynamically adjusted according to the real-time changes of the spatial vitality. Collect interactive feedback data of the user on the initial lighting effect. The interactive feedback data includes user residence duration, interaction frequency, and pre-set emotional feedback indicators.

6. The method according to claim 1, characterized in that Based on the interactive feedback data, extract the preference characteristics of the user for the lighting effect, construct a lighting effect evaluation index, optimize and sort the lighting control parameters according to the lighting effect evaluation index, establish a parameter optimization strategy, and iteratively adjust the lighting control instruction based on the parameter optimization strategy to generate an optimal lighting control instruction, including: Collect interactive feedback data of the user on the lighting effect, construct a multi-dimensional feature vector according to the user residence duration, interaction frequency, and emotional feedback indicators in the interactive feedback data, and input the multi-dimensional feature vector into a pre-set user preference scoring module to obtain a preference score. Perform weighted fusion on the preference score, comfort index, and spatial coordination index to generate a comprehensive evaluation value, and construct a time series evaluation matrix based on the comprehensive evaluation value. The time series evaluation matrix includes a sequence of evaluation values within a historical time window. Calculate the sensitivity of the comprehensive evaluation value to the lighting control parameters, construct a parameter sensitivity distribution module using kernel density estimation, suppress the influence of abnormal samples through iteratively re-weighted least squares, and determine the parameter priority and sort according to the historical optimization effect. Based on the sorting result of the lighting control parameters, construct a parameter optimization objective function in sequence. Take the minimization of the mean square error of the sequence of evaluation values in the time series evaluation matrix as the optimization objective, use the change amplitude of the parameters at adjacent times as the constraint condition, and perform gradient iterative update on the lighting control parameters based on the optimization objective function. Stop the iteration when the difference between the comprehensive evaluation values at adjacent times is less than a preset threshold, and output the optimized lighting control parameters. Generate an optimal lighting control instruction according to the optimized lighting control parameters and the pre-set environmental constraint parameters, in combination with a pre-set dynamic gain matrix.

7. The method according to claim 6, wherein Calculate the sensitivity of the comprehensive evaluation value to the lighting control parameters, construct a parameter sensitivity distribution module using kernel density estimation, suppress the influence of abnormal samples through iteratively re-weighted least squares, and determine the parameter priority and sort according to the historical optimization effect, including: Obtain the historical adjustment data of the lighting control parameters and the corresponding preference scores. Based on the historical adjustment data and the corresponding preference scores, calculate the first-order sensitivity of the comprehensive evaluation value to the lighting control parameters through partial derivatives, calculate the second-order interaction sensitivity through mixed partial derivatives, and combine the central difference method to construct a Jacobian matrix to obtain the initial sensitivity sequence; Use the Gaussian kernel function to perform kernel density estimation on the initial sensitivity sequence, adaptively calculate the optimal bandwidth parameter based on the sample standard deviation, transform the initial sensitivity sequence through the kernel function to obtain a continuous probability distribution, and output the distribution characteristics of the parameter sensitivity; Construct a Huber-type M-estimation loss function to perform anomaly detection on the distribution characteristics of the parameter sensitivity, adaptively calculate the sample weight matrix based on the residual size, apply the sample weight matrix to the iteratively reweighted least squares calculation, and suppress the influence of abnormal samples through multiple rounds of iterative optimization, and output the corrected sensitivity value; Calculate the temporal weight of the pre-obtained historical evaluation data based on the time decay exponential function, and perform weighted summation of the temporal weight and the historical comprehensive evaluation value to obtain the historical optimization effect of the parameter; Perform multi-objective weighted fusion on the corrected sensitivity value, the historical optimization effect of the parameter, and the parameter stability index to determine the parameter priority, sort the lighting control parameters in descending order according to the parameter priority, and generate a parameter optimization order table, where the parameter stability index includes the parameter fluctuation range, the convergence speed, and the disturbance recovery ability.

8. A cultural and tourism lighting art effect generation system for multi-modal embodied interaction, which is used to implement the method described in any one of the foregoing claims 1-7, and is characterized in that, Comprising: The first unit is used to collect the initial interaction data of the user in the cultural and tourism scenario and analyze the temporal correlation, extract the user behavior characteristics, construct the user behavior characteristic sequence and calculate the change trend of the user behavior characteristic sequence, obtain the light intensity, the crowd density, and the distribution of cultural element information in the cultural and tourism scenario, and generate the scenario interaction data stream in combination with the user behavior characteristic sequence and the corresponding change trend; The second unit is used to divide the interaction space based on the scenario interaction data stream, calculate the activity and emotional intensity of the users in each space and generate a spatial vitality map, calculate the energy difference between adjacent spaces according to the spatial vitality map, determine the light fading coefficient, track the real-time change of the spatial vitality based on the light fading coefficient, and establish a dynamic mapping relationship of the light effect; The third unit is used to output and execute the lighting control instruction according to the dynamic mapping relationship, generate the initial lighting effect and collect the interactive feedback data of the user on the initial lighting effect; The fourth unit is used to extract the preference characteristics of the user for the lighting effect based on the interactive feedback data, construct the lighting effect evaluation index, optimize and sort the lighting control parameters according to the lighting effect evaluation index, establish a parameter optimization strategy, and iteratively adjust the lighting control instruction based on the parameter optimization strategy to generate the optimal lighting control instruction; The fifth unit is used to output the optimal lighting control instruction, execute and record the optimal lighting effect.

9. An electronic device, characterized in that, Comprising: A processor; A memory for storing instructions executable by the processor; Wherein, the processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images