Activity scene arrangement scheme generation method and system

Abstract

The invention relates to an activity scene arrangement scheme generation method and system, and the method comprises the following steps: obtaining activity scene data, and carrying out the analysis to obtain an activity theme and the characteristics of participants; objective condition parameters of the target site are obtained, and the candidate arrangement elements are screened; constructing a multi-dimensional attribute vector according to the screened arrangement elements; clustering the multi-dimensional attribute vectors, and performing spatial layout planning on layout elements; performing combined optimization on the generated spatial layout plan, color scheme and light effect setting; the time period of the activity is extracted, and the recommended arrangement scheme is adjusted according to the relevance; and performing visual rendering on the adjusted recommended arrangement scheme, and optimizing and determining an activity scene arrangement scheme. Intelligentization and individuation of activity scene arrangement are achieved, integrity, creativity and coordination of the arrangement effect are improved, meanwhile, influences of actual site conditions and activity time are considered, and a more optimized scene arrangement solution is provided for a user.

Description

Technical Field

[0001] The present invention relates to the technical field of activity scene arrangement, and in particular to a method and system for generating an activity scene arrangement scheme. Background Art

[0002] In the process of generating the scene layout plan for an event, we are faced with the technical contradiction of how to automatically generate a scene layout plan that meets both individual needs and overall coordination based on multiple factors such as the theme of the event, participants, and venue conditions. On the one hand, the scene layout plan needs to reflect the characteristics and style of the event and fit the preferences of the participants, while also considering the limitations of objective conditions such as the size, shape, and lighting of the venue; on the other hand, how to achieve both creative highlights and harmony and unity in terms of spatial layout, color matching, and lighting effects for each layout element in the plan, and avoid being too complicated, messy, or monotonous, is a technical problem that requires a multi-dimensional balance. Furthermore, there is an influencing relationship between the scene layout plan and the time of the event, so it is a particularly critical step to explore the correlation between the scene layout plan and the event and adjust the layout plan based on the correlation. In addition, the generation of the scene layout plan requires screening and combining in a massive scene material library. How to use algorithms to achieve intelligent matching and optimized arrangement of various layout elements and improve the efficiency and quality of the generated plan is also a technical challenge that needs to be overcome. Summary of the invention

[0003] In order to solve the problems existing in the above-mentioned prior art, the present invention aims to provide a method and system for generating an activity scene arrangement plan. The method for generating an activity scene arrangement plan realizes the intelligentization and personalization of the activity scene arrangement, improves the integrity, creativity and coordination of the arrangement effect, and takes into account the influence of the actual site conditions and the activity time, providing users with a more optimized scene arrangement solution.

[0004] The method for generating an activity scene arrangement scheme of the present invention comprises the following steps:

[0005] S1. Acquire activity scene data, process the activity scene data, analyze to obtain the activity theme and the characteristics of the participants, and retrieve several candidate layout elements related to the theme and meeting the characteristics of the participants according to a pre-established layout element material library;

[0006] S2. Obtain objective condition parameters of the target site, use the objective condition parameters as constraints, screen the candidate layout elements, and eliminate layout elements that exceed the site carrying capacity or conflict with lighting conditions;

[0007] S3, extracting the color and texture visual attribute characteristics of the target scene based on the screened layout elements, and combining the material and style semantic attributes of the layout elements to construct a multidimensional attribute vector;

[0008] S4, clustering the multidimensional attribute vectors, classifying the visually and semantically similar layout elements into one category, obtaining a plurality of layout element combinations, and planning the spatial layout of the layout elements in combination with the size, shape, and placement of the elements, and generating a color scheme and lighting effect setting that matches the layout;

[0009] S5. Combine and optimize the generated space layout plan, color scheme and lighting effect settings, score the overallness, creativity and coordination of the layout plan through a preset scoring mechanism, and select the plan with the highest score as the recommended layout plan;

[0010] S6, extracting the time period of the activity, analyzing the influence relationship between the time period of the activity and the layout elements in the layout plan, obtaining the correlation between the scene layout plan and the time period of the activity, and adjusting the recommended layout plan according to the correlation;

[0011] S7. Visually render the adjusted recommended layout plan to generate a three-dimensional scene model, and optimize and determine the activity scene layout plan.

[0012] Preferably, the step S1 specifically includes:

[0013] Acquire activity scene data, and use the TF-IDF algorithm to process text data in the activity scene data to obtain keyword features of the activity scene data;

[0014] According to the keyword features, principal component analysis is used to reduce the dimension of the numerical features in the activity scene data to obtain the activity theme feature vector and the participant group feature vector;

[0015] According to each arrangement element in the pre-established arrangement element material library, the cosine similarity between the arrangement element and the activity theme feature vector is calculated to obtain the theme relevance score of the arrangement element, and the cosine similarity between the arrangement element and the participant group feature vector is calculated to obtain the personnel matching score of the arrangement element;

[0016] The optimal weights are determined through regression analysis of historical data, and the weighted sum of the subject relevance score and the person matching score is taken to obtain the comprehensive score of the layout elements;

[0017] Sort the layout elements in descending order according to the comprehensive scores, and select the layout elements whose scores exceed a preset threshold as a set of candidate layout elements;

[0018] Acquire the color, shape, and material attribute features of each element in the candidate layout element set, and construct a layout element feature matrix;

[0019] A hierarchical clustering algorithm is used to analyze the arrangement element feature matrix, and clustering results of a suitable level are intercepted according to a preset number of scene areas to obtain a plurality of arrangement element clusters.

[0020] Preferably, in step S1, the establishment of the layout element library includes:

[0021] Classify according to the activity theme and the characteristics of the participants, mark relevant tags, and when matching, based on keyword retrieval and feature semantic matching, determine whether the selected elements are in line with the theme and meet the preferences of the participants, and generate several candidate layout plans.

[0022] Preferably, the step S2 specifically includes:

[0023] Acquire point cloud data of the target site, and generate a three-dimensional model of the site based on the point cloud data, wherein the three-dimensional model includes the area size, shape and height space parameters of the site;

[0024] Measure the lighting intensity at different times and locations to obtain a heat map of the lighting conditions at the site;

[0025] Establishing a site constraint model according to the three-dimensional model and the lighting condition heat map, wherein the site constraint model includes a space utilization rate threshold and a lighting conflict degree threshold;

[0026] The candidate layout elements are preliminarily screened by a rule-based filter, and the rule-based filter eliminates the layout elements that exceed the site carrying capacity or seriously conflict with the lighting conditions according to the site constraint model.

[0027] Preferably, the step S3 specifically includes:

[0028] After eliminating the layout elements that exceed the site's carrying capacity or conflict with lighting conditions, the color histogram analysis is used to obtain the color features of the target scene image, and the main color tone, color saturation and brightness parameters in the color features are extracted;

[0029] According to the color features, the texture features of the scene are extracted using gray-level co-occurrence matrix analysis to obtain texture parameters including roughness, directionality and complexity.

[0030] Extracting material characteristic parameters of the layout elements from a pre-established material database, wherein the material characteristic parameters include reflectivity, transparency and texture;

[0031] The word vector model is used to process the description text of the layout elements, obtain style-related keywords and semantic tags, and generate the style semantic vector of the layout elements;

[0032] Through principal component analysis, color features, texture parameters, material characteristic parameters and style semantic vectors are integrated through dimensionality reduction to construct a multidimensional attribute vector containing color, texture, material and style information.

[0033] Preferably, the step S4 specifically includes:

[0034] Using the DBSCAN clustering algorithm to perform cluster analysis on the multidimensional attribute vectors, and obtaining a plurality of arrangement element combinations by calculating the Euclidean distance between the vectors;

[0035] According to the combination of layout elements, a simulated annealing algorithm is used to perform spatial layout planning, and the size, shape and placement of the layout elements are encoded as states;

[0036] According to the status, the objective function is set to evaluate the space utilization, aesthetics and functionality of the layout, and the layout solution is obtained through iterative optimization;

[0037] Based on the layout plan, according to user preferences or design style guidance, adjust the size and shape of layout elements, determine the proportional relationship and visual balance between elements, generate a color scheme that matches the layout plan, calculate the main color of the layout elements, and use color contrast and coordination principles to determine the overall color scheme of the scene;

[0038] According to the color scheme and layout plan, the scene effects under different light source positions and intensities are simulated, and the light source parameters are iteratively adjusted to obtain a lighting effect setting with a harmonious and unified overall visual effect.

[0039] Preferably, the step S5 specifically includes:

[0040] Receiving particle swarm optimization algorithm input information carrying space layout planning, color scheme and lighting effect setting parameters, the particle swarm optimization algorithm input information is generated by the initialization process;

[0041] Iterative optimization is performed according to the input information of the particle swarm optimization algorithm, and the iterative optimization generates multiple candidate layout solutions through speed update and position update operations;

[0042] Obtaining scoring criteria for the three dimensions of integrity, creativity, and coordination, the scoring criteria being used to evaluate the candidate layout plans;

[0043] A multi-layer perceptron is used to quantitatively evaluate each candidate layout scheme in terms of integrity, creativity, and coordination, and obtain scores of the candidate layout scheme in the three dimensions;

[0044] After obtaining the scores of the candidate layout solutions in the three dimensions, a weighted sum method is used to calculate the total score of the candidate layout solutions, wherein the weights are dynamically adjusted according to user feedback through a reinforcement learning algorithm;

[0045] The candidate arrangement scheme with the highest total score is determined as the recommended arrangement scheme;

[0046] A simulated annealing algorithm is executed on the recommended arrangement scheme, the simulated annealing algorithm is used to fine-tune and optimize the recommended arrangement scheme, and the recommended arrangement scheme after fine-tune and optimization by the simulated annealing algorithm is output.

[0047] Preferably, the step S6 specifically includes:

[0048] Obtaining time period data in the activity information, and converting the time period data into a Unix timestamp;

[0049] Assigning a time attribute tag to the layout elements in the layout plan according to the Unix timestamp, wherein the time attribute tag includes an applicable period, lighting requirements and functional characteristics;

[0050] Analyze the correlation between time periods and layout elements to obtain the time and element adaptability matrix;

[0051] Genetic algorithm is used to optimize the layout scheme. The chromosome encoding of genetic algorithm is a binary string of element position and attribute. The fitness function of genetic algorithm is the time and element adaptability matrix.

[0052] Dynamically adjust the position, lighting, and functional settings of elements based on the optimized layout plan. Dynamic adjustments include moving furniture positions, adjusting light brightness and color temperature, and switching device function modes.

[0053] Preferably, the step S7 specifically includes:

[0054] Constructing a scene basic model, wherein the scene basic model includes a spatial structure, geometric shapes and position information of layout elements;

[0055] Extract the corresponding material map from the hierarchically organized material library and apply it to the surface of the scene base model to obtain a preliminary visualized 3D scene;

[0056] Perform real-time lighting rendering on 3D scenes, calculate direct lighting, indirect lighting and shadows, and obtain realistic lighting effects;

[0057] Set up multiple virtual cameras to capture 3D scene views from different angles and heights, and combine and process the scene views captured by the virtual cameras to achieve smooth perspective switching and roaming effects;

[0058] Dynamically optimize the 3D scene, automatically adjust the rendering accuracy according to the scene complexity, and obtain a smooth interactive scene preview;

[0059] receiving a real-time adjustment instruction through a user interaction interface, the real-time adjustment instruction including adjustment information of an object position, material, and lighting parameters;

[0060] The 3D scene is updated according to the real-time adjustment instructions, while the frame rate and rendering time are monitored, and the one that meets the preset performance threshold is determined as the final activity scene layout plan.

[0061] A system for generating an activity scene arrangement scheme adopts the above-mentioned method for generating an activity scene arrangement scheme, comprising:

[0062] A data processing and analysis module, which is used to obtain activity scene data and analyze activity themes and participant characteristics;

[0063] A layout element screening module, which is used to retrieve candidate layout elements from a material library according to themes and features;

[0064] A site constraint processing module, the site constraint processing module is used to select layout elements according to site condition parameters;

[0065] A layout scheme generation module, which is used to construct attribute vectors, cluster, space layout planning and generate color lighting schemes;

[0066] A scheme optimization scoring module, which is used to combine and optimize layout schemes and select the best scheme through a scoring mechanism;

[0067] A visualization rendering module is used to generate a three-dimensional scene model and preview the optimization layout effect.

[0068] The method and system for generating an activity scene arrangement scheme of the present invention have the following advantages:

[0069] The method for generating an activity scene arrangement scheme of the present invention can accurately retrieve arrangement elements related to the theme and in line with the characteristics of the participants by analyzing the activity theme and the characteristics of the participants, thereby meeting the personalized needs of the activity; clustering by using multidimensional attribute vectors can quickly classify similar arrangement elements into one category, which is convenient for subsequent layout planning and effect setting; by constructing multidimensional attribute vectors, comprehensively considering factors such as color, texture, material, style, etc. of the arrangement elements, ensuring the visual and semantic coordination and unity of the arrangement scheme; extracting the activity time period and analyzing its influence relationship with the arrangement elements, adjusting the recommended arrangement scheme according to the correlation, so that the arrangement scheme is more in line with the actual needs and time rhythm of the activity; by optimizing and determining the final activity scene arrangement scheme, ensuring that the arrangement effect meets the expected requirements, and improving the success rate and satisfaction of the activity. The present invention realizes the intelligence and personalization of the activity scene arrangement, improves the integrity, creativity and coordination of the arrangement effect, and at the same time considers the influence of the actual site conditions and the activity time, providing users with a more optimized scene arrangement solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 It is a flow chart of a method for generating an activity scene arrangement plan described in the present invention. DETAILED DESCRIPTION

[0071] like Figure 1 As shown, a method for generating an activity scene arrangement scheme according to the present invention comprises the following steps:

[0072] S1. Acquire activity scene data, process the activity scene data, analyze the activity theme and the characteristics of the participants, and retrieve several candidate layout elements that are related to the theme and meet the characteristics of the participants based on a pre-established layout element material library;

[0073] S2. Obtain objective condition parameters of the target site, use the objective condition parameters as constraints, screen candidate layout elements, and eliminate layout elements that exceed the site carrying capacity or conflict with lighting conditions;

[0074] S3, extracting the color and texture visual attribute characteristics of the target scene based on the screened layout elements, and combining the material and style semantic attributes of the layout elements to construct a multidimensional attribute vector;

[0075] S4, clustering the multidimensional attribute vectors, classifying the layout elements that are visually and semantically similar into one category, obtaining a number of layout element combinations, and planning the spatial layout of the layout elements in combination with the size, shape, and placement of the elements, generating a color scheme and lighting effect settings that match the layout;

[0076] S5. Combine and optimize the generated space layout plan, color scheme and lighting effect settings, score the overallness, creativity and coordination of the layout plan through a preset scoring mechanism, and select the plan with the highest score as the recommended layout plan;

[0077] S6, extracting the time period of the activity, analyzing its influence relationship with the layout elements in the layout plan, obtaining the correlation between the scene layout plan and the time period of the activity, and adjusting the recommended layout plan according to the correlation;

[0078] S7. Visually render the adjusted recommended layout plan, generate a three-dimensional scene model, and optimize and determine the activity scene layout plan.

[0079] Furthermore, in this embodiment, step S1 specifically includes:

[0080] Acquire activity scene data, use TF-IDF algorithm to process text data in the activity scene data, and obtain keyword features of the activity scene data;

[0081] According to the keyword features, principal component analysis is used to reduce the dimension of the numerical features in the activity scene data to obtain the activity theme feature vector and the participant group feature vector;

[0082] According to each arrangement element in the pre-established arrangement element material library, the cosine similarity between the arrangement element and the activity theme feature vector is calculated to obtain the theme relevance score of the arrangement element, and the cosine similarity between the arrangement element and the participant group feature vector is calculated to obtain the personnel matching score of the arrangement element;

[0083] The optimal weights are determined through regression analysis of historical data, and the weighted sum of the subject relevance score and the person matching score is taken to obtain the comprehensive score of the layout elements;

[0084] Sort the layout elements in descending order according to the comprehensive scores, and select the layout elements whose scores exceed a preset threshold as a set of candidate layout elements;

[0085] Obtaining the color, shape, and material attribute features of each element in the candidate layout element set, and constructing a layout element feature matrix;

[0086] A hierarchical clustering algorithm is used to analyze the layout element feature matrix, and the clustering results of the appropriate level are intercepted according to the preset number of scene areas to obtain multiple layout element clusters;

[0087] Specifically, the activity scene data is obtained and the TF-IDF algorithm is used to process the text data to extract keyword features. At the same time, principal component analysis is used to reduce the dimension of numerical features to obtain the activity theme feature vector and the participant group feature vector;

[0088] According to the two feature vectors, the pre-established layout element library is searched and screened, and the cosine similarity between each layout element in the layout element library and the activity theme feature vector is calculated to obtain the theme relevance score, and the cosine similarity between the layout element and the participant group feature vector is calculated to obtain the participant matching score;

[0089] The optimal weight is determined by regression analysis of historical data, and the weighted sum of the two scores is used to obtain the comprehensive score of the layout element;

[0090] Sort the layout elements in descending order according to their comprehensive scores, select the layout elements whose scores exceed a preset threshold as a set of candidate layout elements, extract attribute features such as color, shape, and material for each element in the set of candidate layout elements, and construct a layout element feature matrix;

[0091] A hierarchical clustering algorithm is used to analyze the layout element feature matrix, and the clustering results of the appropriate level are intercepted according to the preset number of scene areas to obtain multiple layout element clusters;

[0092] Select the layout elements with the highest comprehensive score from each cluster to form a preliminary scene layout plan, which not only meets the theme of the activity but also meets the characteristics of the participants;

[0093] Here is an example:

[0094] When obtaining activity scene data, extract records containing activity descriptions, participant information, and historical preferences from the database. Use the TF-IDF algorithm to process the activity description text and extract keyword features, such as "music festival", "outdoor", "rock", etc., to form an activity theme feature vector;

[0095] Principal component analysis was applied to the numerical characteristics of the participants, such as age and occupation, and the dimensions were reduced to 10 principal components to form the characteristic vector of the participant group;

[0096] In the pre-established material library containing 1,000 layout elements, the cosine similarity between each element and the feature vector is calculated. For example, the cosine similarity between a stage setting element and the feature vector of the activity theme is 0.85, and the cosine similarity between a stage setting element and the feature vector of the participant group is 0.72.

[0097] Through regression analysis of the data of the past 100 events, the optimal weights were determined to be 0.6 and 0.4, respectively, and the overall score of the set element was calculated to be 0.798;

[0098] Sort all layout elements in descending order, and select elements with a score greater than 0.7 as candidate sets;

[0099] Extract the color (RGB value), shape (circle, square), and material (metal, wood) attributes of the candidate elements and construct a feature matrix;

[0100] The feature matrix is ​​analyzed using a hierarchical clustering algorithm. Assuming that the number of preset scene areas is 5, the clustering results of the corresponding level are intercepted to obtain 5 layout element clusters;

[0101] Elements with the highest comprehensive scores were selected from each cluster, such as the main stage backdrop, side decorative lights, audience area seats, entrance arches and rest area awnings, to form the final scene layout plan, which not only reflects the theme characteristics of the "outdoor music festival" but also meets the preferences of young working people.

[0102] Furthermore, in this embodiment, in step S1, establishing the layout element material library includes:

[0103] Classify according to the activity theme and the characteristics of the participants, mark relevant tags, and when matching, determine whether the selected elements are in line with the theme and meet the preferences of the participants based on keyword retrieval and feature semantic matching, and generate several candidate layout plans;

[0104] Specifically, a tree-structured material library is constructed for historical activity data, and each node in the material library represents a category or subcategory;

[0105] The TF-IDF algorithm is used to extract the keyword features of each material item in the material library and generate a feature vector. The relevant tags are marked according to the activity theme and participant characteristics of the material item, and an inverted index is constructed for fast retrieval.

[0106] Get the current activity's theme description and participant information, extract activity theme keywords through the TF-IDF algorithm and generate an activity theme feature vector, and generate a participant group feature vector based on the participant's attribute information;

[0107] Use cosine similarity to calculate the similarity between the activity theme feature vector and the feature vector of each material item in the material library, and calculate the matching degree between the feature vector of the participant group and the material label. Combine the two scores to calculate the comprehensive score of the material.

[0108] Sort the comprehensive scoring results in descending order, select the top N high-scoring materials and group them, select representative materials from each group, and combine them to form multiple candidate layout solutions;

[0109] Calculate the diversity and balance scores of the candidate layout plans, and select the most suitable candidate layout plans based on the comprehensive scores of diversity and balance;

[0110] Here is an example:

[0111] Build a material library based on historical activity data, organize materials in a tree structure, with each node representing a category or subcategory. Use the TF-IDF algorithm to extract keyword features and generate feature vectors for each material item. Label each material with labels related to the activity theme and the characteristics of the participants, and build an inverted index for fast retrieval.

[0112] Get the current activity's theme description and participant information, extract activity theme keywords using the TF-IDF algorithm, generate activity theme feature vectors using word embedding tools, and generate participant group feature vectors based on participant attribute information;

[0113] Use cosine similarity to calculate the similarity between the activity theme feature vector and the feature vector of each material item in the material library. At the same time, calculate the matching degree between the feature vector of the participant group and the material label. Combine the two scores and apply the weights obtained through historical data regression analysis to calculate the comprehensive score of the material.

[0114] The scoring results are sorted in descending order, and the top N high-scoring materials are selected. These materials are grouped using the K-means clustering algorithm. Representative materials are selected from each group to form multiple candidate layout schemes. Finally, the diversity of the schemes is calculated using the entropy value, and the balance is calculated using the variance. Based on the comprehensive scores of diversity and balance, the most suitable candidate layout schemes are screened out.

[0115] When constructing the material library, a three-layer tree structure is used, with the root node being "activity layout", the second layer including categories such as "stage design", "lighting layout", and "seating arrangement", and the third layer being specific material items;

[0116] For the "background board" material under "stage design", use the TF-IDF algorithm to extract keywords such as "modern", "simple", and "sense of technology", generate a feature vector [0.5, 0.3, 0.2], and annotate the labels "young people", "technology company", etc., and establish a mapping in the inverted index;

[0117] Get the activity theme "Technology Innovation Summit" and the participant information "25-40 years old IT practitioners", extract the keywords "technology", "innovation", and "summit" through TF-IDF, and use the word embedding tool to generate the activity theme feature vector [0.6, 0.3, 0.1];

[0118] Generate a feature vector [0.4, 0.5, 0.1] based on the attributes of the participants, representing age, occupation, and other characteristics. Calculate the cosine similarity between the activity theme feature vector and the "background board" material feature vector as 0.85, and the matching degree between the participant feature vector and the material label is 0.75. Apply weights of 0.6 and 0.4, and obtain the weights through historical data regression analysis to calculate the comprehensive score of 0.81;

[0119] Sort all materials in descending order, select the top 50 high-scoring materials, use the K-means algorithm with K=5 for clustering, and select the highest-scoring materials from each cluster to form candidate solutions;

[0120] The diversity (entropy) and balance (variance) of the schemes were calculated. For example, the diversity of scheme A was 0.9, the balance was 0.8, and the comprehensive score was 0.85. Finally, the three candidate layout schemes with the highest scores were selected.

[0121] Furthermore, in this embodiment, step S2 specifically includes:

[0122] Obtain the point cloud data of the target site and generate a three-dimensional model of the site based on the point cloud data. The three-dimensional model includes the area, shape and height space parameters of the site. The point cloud data is collected by a laser scanner.

[0123] Measure the lighting intensity at different times and locations to obtain a heat map of the lighting conditions at the site; measure the lighting intensity at different times and locations using a illuminance meter;

[0124] A site constraint model is established based on the three-dimensional model and the lighting condition heat map. The site constraint model includes a space utilization rate threshold and a lighting conflict degree threshold.

[0125] The candidate layout elements are preliminarily screened through rule-based filters. The rule-based filters eliminate the layout elements that exceed the site's carrying capacity or seriously conflict with lighting conditions according to the site constraint model;

[0126] Furthermore, a genetic algorithm is used to generate multiple layout schemes. The encoding method of the genetic algorithm is the position and orientation of the layout elements. The fitness function of the genetic algorithm is the weighted sum of space utilization and lighting conflict degree.

[0127] If the space utilization rate is lower than the preset threshold or the lighting conflict degree is higher than the preset threshold, the layout plan is determined to be infeasible;

[0128] For feasible layout plans, calculate the environmental compatibility score of each layout element. The environmental compatibility score is obtained by combining shape matching, weight distribution balance and light coordination through weighted average method, and select the plan with the highest environmental compatibility score as the final layout plan;

[0129] Specifically, the point cloud data of the target site is obtained through a laser scanner, and a three-dimensional model of the site is generated using a point cloud processing tool to obtain the spatial parameters such as the size, shape and height of the site. At the same time, a illuminance meter is used to measure the lighting intensity at different times and locations to generate a lighting condition heat map, thus forming an objective condition parameter data set of the site;

[0130] According to the objective condition parameters of the site, a site constraint model is established, including the space utilization rate threshold and the lighting conflict degree threshold. The candidate layout elements are preliminarily screened through rule-based filters to eliminate the layout elements that obviously exceed the site carrying capacity or seriously conflict with the lighting conditions.

[0131] The spatial layout of the layout elements that have passed the preliminary screening is optimized, and a genetic algorithm is used to generate multiple layout schemes, in which the encoding method is the position and orientation of the layout elements, and the fitness function is the weighted sum of the space utilization rate and the lighting conflict degree. New schemes are generated through crossover and mutation operations;

[0132] If the space utilization rate is lower than the preset threshold or the lighting conflict degree is higher than the preset threshold, the solution is determined to be infeasible;

[0133] For feasible layout schemes, calculate the environmental compatibility score of each layout element, and use the weighted average method to comprehensively consider shape matching, weight distribution balance and light coordination. The shape matching is calculated by the outline overlap rate, the weight distribution balance is calculated by the center of gravity offset, and the light coordination is calculated by the lighting intensity difference. The scheme with the highest environmental compatibility score is selected as the final layout scheme;

[0134] Use 3D rendering tools to visualize the final layout plan and generate scene renderings to intuitively evaluate the feasibility of the plan;

[0135] Here is an example:

[0136] A FaroFocus S350 laser scanner was used to scan the 300-square-meter indoor venue, obtaining 100 million points of point cloud data. The 3D model of the venue was generated through CloudCompare software, and the venue size was accurately measured to be 20 meters × 15 meters × 4 meters.

[0137] A KonicaMino ltaT-10A illuminance meter was used to set up 100 measurement points in the venue, and measurements were taken every 2 hours for 24 hours to generate a heat map of lighting conditions. The light intensity ranged from 50 to 500 lux.

[0138] Based on the acquired data, the space utilization threshold is set to 80%, the lighting conflict threshold is set to 20%, and the rule-based filter is applied to 1000 candidate layout elements to eliminate 450 elements that do not meet the requirements. The genetic algorithm is used to optimize the spatial layout of the remaining 550 elements, with a population size of 100, 500 iterations, a crossover rate of 0.8, a mutation rate of 0.1, and the encoding method is the x, y coordinates and rotation angle of each element. The fitness function is 0.7×space utilization+0.3×(1-lighting conflict);

[0139] Generate 50 feasible solutions. When calculating the environmental compatibility score, the shape matching weight is 0.4, the weight distribution balance weight is 0.3, and the light coordination weight is 0.3.

[0140] The calculation showed that the highest-scoring solution had a contour overlap of 92%, a center of gravity offset of 0.5 meters, and a lighting intensity difference of 30 lux;

[0141] Finally, 3dsMax software was used to perform three-dimensional rendering of the best solution and generate a 4K resolution scene rendering for final evaluation.

[0142] Furthermore, in this embodiment, step S3 specifically includes:

[0143] After eliminating the layout elements that exceed the site's carrying capacity or conflict with lighting conditions, the color histogram analysis is used to obtain the color features of the target scene image, and the main color tone, color saturation and brightness parameters in the color features are extracted;

[0144] According to the color features, the texture features of the scene are extracted using gray-level co-occurrence matrix analysis to obtain texture parameters including roughness, directionality and complexity.

[0145] Extracting material characteristic parameters of the layout elements from a pre-established material database, the material characteristic parameters including reflectivity, transparency and texture;

[0146] The word vector model is used to process the description text of the layout elements, obtain style-related keywords and semantic tags, and generate the style semantic vector of the layout elements;

[0147] The color features, texture parameters, material characteristic parameters and style semantic vectors are integrated through principal component analysis to construct a multidimensional attribute vector containing color, texture, material and style information.

[0148] Specifically, the color histogram analysis is used to analyze the target scene image to extract the main color tone, color saturation and brightness features, and the gray-level co-occurrence matrix analysis is used to extract the texture features of the scene, including roughness, directionality and complexity parameters, to form the visual attribute feature vector of the scene;

[0149] Extract attributes of the selected layout elements, calculate the similarity between each element and the scene visual attributes, and ensure the coordination between the layout elements and the scene features;

[0150] Extracting material characteristic parameters of each layout element from a pre-established material database, which includes attribute fields such as reflectivity, transparency and texture, and constructing a material feature vector of the layout element;

[0151] Use the word vector model to perform semantic analysis on the description text of the layout elements, extract style-related keywords and semantic tags, such as "modern", "classical", and "simple", and generate the style semantic vector of the layout elements;

[0152] The principal component analysis is used to reduce the dimension of the scene visual attribute feature vector, the layout element material feature vector and the style semantic vector to construct a multi-dimensional attribute vector. This vector contains information on multiple dimensions of color, texture, material and style, which is used in the subsequent matching and optimization process.

[0153] Here is an example:

[0154] The color histogram of the 1920x1080 pixel high-definition image of the target scene was analyzed, and the 256-level grayscale value distribution of the RGB three channels was extracted. The main color was RGB (120, 180, 210), the color saturation was 0.65, and the brightness was 0.75. The texture features were extracted using the gray-level co-occurrence matrix analysis, and the distance parameter was set to 1, and the directions were 0°, 45°, 90°, and 135°. The roughness was calculated to be 0.82, the directionality was 0.56, and the complexity was 0.73;

[0155] Attributes of the 50 selected layout elements were extracted, and the cosine similarity between each element and the scene visual attribute was calculated. The threshold was set to 0.8, and 40 coordinated elements were retained. The material parameters of these 40 elements were extracted from a material database containing 1,000 records. The database fields included reflectivity 0-1, transparency 0-1, and texture 1-10, with 1 being the roughest and 10 being the smoothest;

[0156] Using the pre-trained Word2Vec model, the vector dimension is set to 100, and semantic analysis is performed on the element description text to extract style labels such as "modern", "classical", and "simple" to generate style semantic vectors;

[0157] Finally, the three types of feature vectors are reduced to 20 dimensions through principal component analysis to obtain a multidimensional attribute vector containing color, texture, material and style information for subsequent optimization;

[0158] Furthermore, by analyzing the design style and material of the layout elements, extracting description information, obtaining style semantic attributes, and constructing a multidimensional attribute vector, the color, texture, material and style of the layout elements are input, and the color, texture, material and style of the layout elements are standardized to form a multidimensional attribute vector, which includes color features, texture features, material features and style features;

[0159] Specifically, an image of a layout element is obtained, and the main hue, color saturation and brightness value of the layout element in RGB color space are extracted through an image processing tool; a grayscale co-occurrence matrix and a local binary pattern are calculated according to the image of the layout element to obtain a texture feature vector, and the texture feature vector includes contrast, correlation, entropy value and texture pattern distribution;

[0160] Retrieving material information of the layout elements from a pre-established material database, and after retrieving matching items, extracting reflectivity, transparency and texture parameters of the layout elements to construct a material feature vector;

[0161] If there is no match in the material database, a convolutional neural network is used to analyze the surface of the layout element and estimate the material property value;

[0162] The Word2Vec word vector model is used to perform semantic analysis on the design description text of the layout elements, extract style-related keywords, map the keywords to a high-dimensional space, and generate a style semantic feature vector;

[0163] The maximum and minimum normalization method is used to standardize the color, texture, material and style feature vectors, and the values ​​of each dimension are mapped to intervals;

[0164] The dimension of the standardized feature vector is reduced by principal component analysis to obtain the reduced-dimensional feature vector;

[0165] Perform feature concatenation on the feature vectors after dimensionality reduction to form a unified multi-dimensional attribute vector;

[0166] Here is an example:

[0167] The color analysis of the layout elements is performed through image processing tools to extract the main color tone, color saturation and brightness value of the RGB color space. At the same time, the gray-level co-occurrence matrix and local binary pattern are calculated to obtain texture features, including contrast, correlation, entropy value and texture pattern distribution, and obtain color and texture feature vectors;

[0168] According to the pre-established material database, the material information of the layout elements is retrieved, the reflectivity, transparency and texture parameters are extracted, and the material feature vector is constructed. If there is no match in the database, the surface of the layout elements is analyzed by the surface material recognition method based on the convolutional neural network to estimate the material attribute value;

[0169] The Word2Vec word vector model is used to perform semantic analysis on the design description text of the layout elements, extract style-related keywords, map the keywords to a high-dimensional space, generate style semantic feature vectors, and quantify the style semantic attributes through sentiment analysis tools;

[0170] The color, texture, material and style feature vectors are standardized using the maximum and minimum normalization method, and the values ​​of each dimension are mapped to the range of 0-1. Then, the principal component analysis is used to reduce the dimension of the features of different dimensions. Then, the four standardized feature vectors are combined by feature splicing to form a unified multi-dimensional attribute vector.

[0171] Finally, the random forest algorithm is used to analyze the importance of features, and the weights of different types of features in the multidimensional attribute vector are adjusted according to the analysis results;

[0172] For example, to analyze a modern sofa, we first use the OpenCV library to extract the RGB color space information, and get the main color R:200, G:180, B:160, color saturation 0.75, and brightness 0.85. The grayscale co-occurrence matrix is ​​calculated to get a contrast of 0.68, a correlation of 0.72, and an entropy of 5.3. At the same time, the LBP algorithm is used to get the texture pattern distribution histogram;

[0173] The leather material information of the sofa is retrieved from the material database, with a reflectivity of 0.3, a transparency of 0.05, and a texture parameter of 8.5 on a 10-point scale;

[0174] The design description of the sofa, "simple modern style, comfortable and durable", is processed using the pre-trained Word2Vec model to generate a 100-dimensional word vector, where the "simple" vector is [0.2, 0.5, ..., 0.3] and the "modern" vector is [0.4, 0.6, ..., 0.1];

[0175] The style semantics was quantified using sentiment analysis tools, and the results were 0.8 for positivity, 0.9 for modernity, and 0.85 for simplicity;

[0176] All features are normalized using the MinMaxScaler function of the sk learn library, and then PCA is applied to reduce the dimension to 20 dimensions;

[0177] Finally, the random forest regression algorithm is used to calculate the feature importance, with the color feature weight of 0.3, the texture feature weight of 0.2, the material feature weight of 0.25, and the style feature weight of 0.25 to generate the final 80-dimensional multidimensional attribute vector.

[0178] Furthermore, in this embodiment, step S4 specifically includes:

[0179] The DBSCAN clustering algorithm is used to perform cluster analysis on multidimensional attribute vectors, and several layout element combinations are obtained by calculating the Euclidean distance between vectors.

[0180] According to the combination of layout elements, the simulated annealing algorithm is used to carry out spatial layout planning, and the size, shape and placement of the layout elements are encoded as states;

[0181] According to the status, the objective function is set to evaluate the space utilization, aesthetics and functionality of the layout, and the layout solution is obtained through iterative optimization;

[0182] Based on the layout plan, according to user preferences or design style guidance, adjust the size and shape of layout elements, determine the proportional relationship and visual balance between elements, generate a color scheme that matches the layout plan, calculate the main color of the layout elements, and use color contrast and coordination principles to determine the overall color scheme of the scene;

[0183] According to the color scheme and layout scheme, simulate the scene effects under different light source positions and intensities, and obtain the lighting effect settings with harmonious and unified overall visual effects by iteratively adjusting the light source parameters;

[0184] Specifically, the DBSCAN clustering algorithm is used to perform cluster analysis on the multidimensional attribute vectors, and the visually and semantically similar layout elements are grouped into one category by calculating the Euclidean distance between the vectors, thus obtaining several layout element combinations;

[0185] For each combination of layout elements, simulated annealing algorithm is used for spatial layout planning. The size, shape and placement of elements are encoded as states. The objective function is set to evaluate the spatial utilization, aesthetics and functionality of the layout, and the layout plan is optimized through iteration.

[0186] Adjust the size and shape of elements according to user preferences or design style guidelines, consider the proportional relationship and visual balance between elements, and use the golden ratio to optimize the size of elements;

[0187] Based on the planned spatial layout, the Munsell color system is used to generate a color scheme that matches the layout, calculate the main color of the layout elements, and use the principles of color contrast and coordination to determine the overall color scheme of the scene;

[0188] Based on the generated color scheme and space layout, real-time global illumination technology is used to simulate the scene effects under different light source positions and intensities. By iteratively adjusting the light source parameters and optimizing the lighting effect settings, the overall visual effect is harmoniously unified.

[0189] Here is an example:

[0190] Perform DBSCAN clustering on a multidimensional attribute vector set containing 50 layout elements, set the neighborhood parameter ε to 0.5, and the minimum number of samples to 3, and obtain 7 layout element combinations;

[0191] The largest combination containing 12 elements was selected, and the simulated annealing algorithm was used for spatial layout planning. The initial temperature was set to 100, the cooling coefficient was 0.95, and 500 iterations were performed;

[0192] The objective function weights were set to space utilization 0.4, aesthetics 0.3, and functionality 0.3. Based on the user preference of "modern simplicity", the element sizes were adjusted, the length-to-width ratio of the largest element (sofa) was set to 1.618 (golden ratio), and other elements were scaled proportionally;

[0193] The color scheme was generated using the Munsell color system, with 5Y7 / 4 (light yellow-brown) as the main color, 10BG5 / 6 (blue-green) as an embellishment, and the color contrast was set to 3.5;

[0194] Using real-time global illumination technology, three main light sources are set, including ceiling lights with a color temperature of 6500K and a brightness of 800lm, floor lamps with a color temperature of 3000K and a brightness of 400lm, and table lamps with a color temperature of 4000K and a brightness of 200lm;

[0195] After 50 iterative adjustments, the main light source position was finally offset by 0.5m from the original position, and the brightness was increased by 10%, achieving a harmonious overall visual effect.

[0196] Furthermore, in this embodiment, step S5 specifically includes:

[0197] Receiving particle swarm optimization algorithm input information carrying space layout planning, color scheme and lighting effect setting parameters, the particle swarm optimization algorithm input information is generated by the initialization process;

[0198] Iterative optimization is performed according to the input information of the particle swarm optimization algorithm, and the iterative optimization generates multiple candidate layout solutions through speed update and position update operations;

[0199] Obtain the scoring criteria for the three dimensions of integrity, creativity, and coordination, which are used to evaluate candidate layout plans;

[0200] A multi-layer perceptron is used to quantitatively evaluate each candidate layout scheme in terms of integrity, creativity, and coordination, and the scores of the candidate layout scheme in the three dimensions are obtained;

[0201] After obtaining the scores of the candidate layout solutions in the three dimensions, the total score of the candidate layout solutions is calculated using a weighted summation method, where the weights are dynamically adjusted based on user feedback through a reinforcement learning algorithm;

[0202] The candidate arrangement scheme with the highest total score is determined as the recommended arrangement scheme;

[0203] Execute a simulated annealing algorithm on the recommended layout scheme, the simulated annealing algorithm is used to fine-tune and optimize the recommended layout scheme, and output the recommended layout scheme after fine-tune and optimization by the simulated annealing algorithm;

[0204] Specifically, the particle swarm optimization algorithm is used to perform combined optimization of spatial layout planning, color scheme and lighting effect settings, the parameters of the three are encoded as particle positions, the group size and number of iterations are set, and multiple candidate layout schemes are generated through speed update and position update operations;

[0205] Construct a preset scoring mechanism and design scoring criteria for the three dimensions of integrity, creativity and coordination. Integrity considers the spatial relationship between elements, creativity is calculated by similarity with historical schemes, and coordination is evaluated by color harmony and light and shadow balance. Multi-layer perceptron is used to quantitatively evaluate each dimension to obtain scores for the three dimensions.

[0206] The scores of the three dimensions are combined using a weighted summation method to calculate the total score of each candidate layout solution, where the weights are dynamically adjusted based on user feedback using a reinforcement learning algorithm to obtain the ranking results of the candidate solutions;

[0207] The solution with the highest score is selected as the recommended layout solution, while the suboptimal solution is retained. The recommended solution is fine-tuned through the simulated annealing algorithm to further optimize the layout effect, and finally the optimized recommended layout solution is output;

[0208] Here is an example:

[0209] In the optimization of the layout of a 100-square-meter living room space, the particle swarm optimization algorithm is used for combinatorial optimization, with the number of particles set to 50 and the maximum number of iterations set to 1000;

[0210] Each particle contains 30 dimensions, corresponding to the positions and orientations of 10 pieces of furniture, 5 main colors, and the positions and intensities of 3 main light sources. The algorithm converged after 900 iterations and obtained 20 candidate solutions;

[0211] The scoring mechanism uses a three-layer neural network, with an input layer containing 50 nodes corresponding to the indicators of space utilization, color coordination, and light and shadow uniformity, a hidden layer of 100 nodes, and an output layer of 3 nodes corresponding to the scores of integrity, creativity, and coordination;

[0212] After training with 5,000 sets of historical data, the network achieved 95% accuracy;

[0213] After scoring the 20 candidate solutions, we used the weights [0.4, 0.3, 0.3] to perform a weighted sum, and the solution with the highest total score was 92.5 points;

[0214] The Q-learning algorithm with ε-greedy strategy was used to adjust the weights according to user feedback, with the learning rate set to 0.1 and the discount factor to 0.9;

[0215] Finally, the simulated annealing algorithm was used to fine-tune the optimal solution, with an initial temperature of 100, a cooling coefficient of 0.95, and 500 iterations. The final optimization solution score was improved to 93.2 points.

[0216] Furthermore, in this embodiment, step S6 specifically includes:

[0217] Get the time period data in the activity information and convert the time period data into Unix timestamp;

[0218] Assign time attribute tags to the layout elements in the layout plan according to the Unix timestamp. The time attribute tags include applicable time periods, lighting requirements and functional characteristics.

[0219] Analyze the correlation between time periods and layout elements to obtain the time and element adaptability matrix;

[0220] Genetic algorithm is used to optimize the layout scheme. The chromosome encoding of genetic algorithm is a binary string of element position and attribute. The fitness function of genetic algorithm is the time and element adaptability matrix.

[0221] Dynamically adjust the position, lighting and function settings of elements according to the optimized layout plan. Dynamic adjustments include moving furniture positions, adjusting light brightness and color temperature, and switching equipment function modes.

[0222] Specifically, the time period data is extracted from the activity information and converted into Unix timestamps. At the same time, a time attribute label is assigned to each layout element in the layout plan, including the applicable period, lighting requirements and functional characteristics. The label assignment method is based on the expert knowledge base and historical data statistics.

[0223] The long short-term memory network is used to analyze the correlation between time periods and layout elements, predict the element adaptability in different time periods, and directly generate the time and element adaptability matrix;

[0224] According to the obtained time and element adaptability matrix, the importance weights of different time features are determined by the hierarchical analysis method, and the adaptability scores of the overall layout scheme in different time periods are calculated using the weighted summation method;

[0225] Genetic algorithm is used to optimize the layout scheme. The chromosome is encoded as a binary string of element position and attribute. The crossover operation uses single-point crossover, and the mutation operation randomly changes the element attribute. The time-element adaptability matrix is ​​used as the fitness function. It is iterated and optimized until the layout scheme with the highest adaptability score within the target time period is obtained.

[0226] According to the optimized plan, dynamically adjust the position, lighting and function settings of elements, including moving furniture, adjusting light brightness and color temperature, and switching equipment function modes;

[0227] Here is an example:

[0228] For a party event that lasts from 8 pm to 2 am the next day, the system converts the time period into Unix timestamps 1634324400 to 1634346000;

[0229] Assign time attribute labels to the 10 layout elements. For example, the sofa is suitable for use throughout the day, has a light requirement of 200-300 lux, and is a rest area.

[0230] Use a 3-layer LSTM network with 64 neurons in each layer. The input is the time series data of the past 100 events, and the output is a 10x24 time-element adaptability matrix.

[0231] The weights were determined by the hierarchical analysis method, where the weight of time period was 0.5, the weight of light demand was 0.3, and the weight of functional characteristics was 0.2;

[0232] The genetic algorithm sets the population size to 100, the number of iterations to 500, the crossover rate to 0.8, and the mutation rate to 0.1. The chromosome code length is 200 bits, the first 100 bits represent the two-dimensional coordinates of 10 elements, and the last 100 bits represent the element attributes;

[0233] Single-point crossover was performed at position 100, and mutation randomly changed 1 to 5 genes. The fitness function was the weighted time-element fitness score;

[0234] The final optimization solution moves the sofa to coordinates (3,4), adjusts the main light brightness from 800 lux to 500 lux, adjusts the color temperature from 6500K to 3000K, and converts the multi-functional table from meeting mode to dining table mode. The overall adaptability score is improved from the initial 0.72 to 0.89.

[0235] Furthermore, in terms of the time period of the activity, the lighting conditions in different time periods are correlated with the lighting selection of the layout plan. According to the different time periods of the activity, the selection of lighting and the brightness of the lighting in the layout plan are dynamically adjusted, including reducing the use of lighting during the daytime period, increasing the lighting arrangement during the night time period or selecting lighting with high illumination;

[0236] Specifically, a light sensor is used to monitor the natural light intensity of the site in real time, obtain weather API and sunrise and sunset time data, predict future light intensity changes, and obtain the predicted light intensity value for each time period;

[0237] Based on the predicted light intensity value, a deep reinforcement learning model is constructed to handle the lighting selection problem. The input parameters of the deep reinforcement learning model include time period, predicted light intensity, and activity type, and the output lighting selection scheme and initial brightness setting;

[0238] The initial brightness setting is adjusted in real time using a fuzzy controller, which takes the natural light intensity, activity progress, and number of participants as input variables and the light brightness as output variable;

[0239] Establish a lighting effect evaluation model, which calculates the illumination uniformity, color rendering index, and glare index, and conducts real-time evaluation of the lighting selection scheme. If the evaluation result does not meet the preset conditions, the lighting reselection and adjustment process is triggered;

[0240] Set the threshold for switching between day and night modes. If the natural light intensity is higher than the daytime threshold, the use of artificial lighting will be automatically reduced.

[0241] If the natural light intensity is below the nighttime threshold, increase the lighting arrangement and lighting intensity;

[0242] Here is an example:

[0243] Use light sensors to monitor the natural light intensity of the site in real time, combine weather API and sunrise and sunset time data, and use long short-term memory networks to predict future light intensity changes and generate predicted light intensity values ​​for each time period;

[0244] Build a deep reinforcement learning model and use the DQN algorithm to handle the lighting selection problem. The input parameters include time period, predicted light intensity, and activity type. The output lighting selection plan and initial brightness setting are used.

[0245] The model training process includes collecting historical lighting usage data, extracting time and environmental characteristics, designing reward functions, and optimizing lighting selection strategies through trial and error. The fuzzy controller is used to adjust the lighting brightness in real time, and the fuzzy rule set is designed by combining expert knowledge and data-driven methods, with natural light intensity, activity progress, and number of participants as input variables and lighting brightness as output variable;

[0246] Establish a lighting effect evaluation model, calculate the illumination uniformity, color rendering index, and glare index, conduct real-time evaluation of the current lighting solution, and trigger the lighting reselection and adjustment process;

[0247] Set the threshold for switching between day and night modes, automatically reduce the use of artificial lighting during daytime hours, and increase the lighting arrangement and lighting intensity during nighttime hours;

[0248] During a three-day indoor exhibition, the system deployed 10 light sensors to collect data every 5 minutes;

[0249] Combining the cloud cover data provided by the local weather API and the sunrise and sunset times accurate to the minute, a 3-layer 128-neuron LSTM network is used to predict the changes in light intensity in the next 72 hours with a prediction accuracy of 90%.

[0250] The deep reinforcement learning model uses the DQN algorithm. The state space includes 24 time periods, 10 levels of light intensity, and 5 types of activities. The action space includes 20 light combination schemes.

[0251] Using the ε-greedy strategy, the initial ε value is 0.9, decaying by 0.1% every 1000 steps. After 1 million interactions, the model converges;

[0252] The fuzzy controller was designed with 27 rules, such as “if the natural light intensity is low and the number of participants is high, the light brightness is high”;

[0253] The evaluation model calculates the illumination uniformity, color rendering index and glare index, and obtains the illumination uniformity target value>0.7, the color rendering index target value>80 and the glare index target value<19;

[0254] The system is set up with day mode 6:00-18:00 and night mode 18:00-6:00. In day mode, when the natural light illumination exceeds 500l ux, 50% of the artificial lighting is automatically turned off. In night mode, the light brightness is increased by 30% and additional decorative lighting is activated. The lighting energy consumption is reduced by 25% compared with the traditional solution, and the visitor comfort score is improved by 15%.

[0255] Furthermore, in this embodiment, step S7 specifically includes:

[0256] Construct a basic scene model, which includes the spatial structure, geometric shapes and position information of the layout elements;

[0257] Extract the corresponding material map from the hierarchically organized material library and apply it to the surface of the scene base model to obtain a preliminary visualized 3D scene;

[0258] Perform real-time lighting rendering on 3D scenes, calculate direct lighting, indirect lighting and shadows, and obtain realistic lighting effects;

[0259] Set up multiple virtual cameras to capture 3D scene views from different angles and heights, and combine and process the scene views captured by the virtual cameras to achieve smooth perspective switching and roaming effects;

[0260] Dynamically optimize the 3D scene, automatically adjust the rendering accuracy according to the scene complexity, and obtain a smooth interactive scene preview;

[0261] receiving a real-time adjustment instruction through a user interaction interface, the real-time adjustment instruction including adjustment information of an object position, material, and lighting parameters;

[0262] Update the 3D scene according to the real-time adjustment instructions, monitor the frame rate and rendering time, and determine the final activity scene layout plan that meets the preset performance threshold;

[0263] Specifically, according to the adjusted recommendation, Autodesk Maya was used to build the basic model of the scene, including the spatial structure, the geometric shape and position information of the layout elements. At the same time, the corresponding material maps were extracted from the hierarchically organized material library and applied to the model surface to achieve preliminary visualization of the scene;

[0264] Use Screen Space Global Illumination technology to perform real-time lighting rendering on 3D scenes, simulate the effects of natural and artificial light sources, calculate direct lighting, indirect lighting and shadows, generate realistic lighting effects, and improve the visual realism of the scene;

[0265] Set up multiple virtual cameras to capture scene views from different angles and heights, achieve smooth perspective switching and roaming effects through the Catmull-Rom spline interpolation algorithm, and generate a multi-perspective preview sequence of the scene;

[0266] Use GPU acceleration and LevelDetail technology to dynamically optimize 3D scenes, automatically adjust rendering accuracy according to scene complexity, improve rendering speed while ensuring image quality, and achieve smooth interactive scene preview;

[0267] The user interface allows real-time adjustment of object positions, materials, and lighting parameters, while monitoring frame rate and rendering time to ensure a smooth interactive experience, optimize and determine the final activity scene layout plan;

[0268] Here is an example:

[0269] In a 300-square-meter exhibition hall layout project, Autodesk Maya 2023 version was used to build a 3D scene containing 50 independent layout elements, each of which consisted of an average of 2,000 polygons;

[0270] Match the most suitable material for each element from a layered material library containing 1,000 materials, and apply 4K resolution PBR texture maps;

[0271] Using Screen Space Globin technology, 3 main light sources and 5 auxiliary light sources are set, the lighting calculation resolution is 1024x1024, the number of indirect light bounces is set to 2, and 30fps real-time rendering is achieved;

[0272] Set up 8 virtual cameras with a field of view of 60 degrees, and use the Catmu l l-Rom spline interpolation algorithm to generate a 200-frame smooth transition sequence;

[0273] Apply Level Detail technology to create 3-level detailed models for each complex object, automatically switching to low-poly models when the distance exceeds 10 meters, and using high-poly models when the distance is less than 2 meters;

[0274] The user interface supports dragging to adjust the position of objects with an accuracy of 0.1 meters. Material adjustments include color, glossiness, and bumpiness, each with 10 levels of adjustment. Lighting parameters can be adjusted within the color temperature range of 2000K to 10000K.

[0275] The system continuously monitors performance, keeps the frame rate above 60fps, and controls the single-frame rendering time within 16.7ms to ensure a smooth real-time interactive experience. The final optimized scene layout solution reaches professional standards in both visual effects and interactive experience.

[0276] A system for generating an activity scene arrangement scheme adopts the above-mentioned method for generating an activity scene arrangement scheme, comprising:

[0277] Data processing and analysis module: The data processing and analysis module is used to obtain activity scene data and analyze activity themes and participant characteristics;

[0278] A layout element screening module, which is used to retrieve candidate layout elements from a material library according to themes and features;

[0279] A site constraint processing module is used to select layout elements according to site condition parameters;

[0280] Layout plan generation module, which is used to construct attribute vectors, clustering, spatial layout planning and generate color lighting plans;

[0281] Scheme optimization scoring module: The scheme optimization scoring module is used to combine and optimize layout schemes and select the best scheme through a scoring mechanism;

[0282] Visualization rendering module, which is used to generate a three-dimensional scene model and preview the optimization layout effect.

[0283] In the description of the present invention, it is necessary to understand that the directions or positional relationships indicated by directional words such as "front, back, up, down, left, right", "lateral, vertical, horizontal" and "top, bottom" are usually based on the directions or positional relationships shown in the drawings. They are only for the convenience of describing the present invention and simplifying the description. Unless otherwise specified, these directional words do not indicate or imply that the device or element referred to must have a specific direction or be constructed and operated in a specific direction. Therefore, they cannot be understood as limiting the scope of protection of the present invention.

[0284] For those skilled in the art, various other corresponding changes and deformations can be made according to the technical solutions and concepts described above, and all of these changes and deformations should fall within the protection scope of the claims of the present invention.

Claims

1. A method for generating an activity scene arrangement scheme, characterized in that: The following steps are involved: S1. Acquire activity scene data, process the activity scene data, analyze to obtain the activity theme and the characteristics of the participants, and retrieve several candidate layout elements related to the theme and meeting the characteristics of the participants according to a pre-established layout element material library; S2. Obtain objective condition parameters of the target site, use the objective condition parameters as constraints, screen the candidate layout elements, and eliminate layout elements that exceed the site carrying capacity or conflict with lighting conditions; S3, extracting the color and texture visual attribute characteristics of the target scene based on the screened layout elements, and combining the material and style semantic attributes of the layout elements to construct a multidimensional attribute vector; S4, clustering the multidimensional attribute vectors, classifying the visually and semantically similar layout elements into one category, obtaining a plurality of layout element combinations, and planning the spatial layout of the layout elements in combination with the size, shape, and placement of the elements, and generating a color scheme and lighting effect setting that matches the layout; S5. Combine and optimize the generated space layout plan, color scheme and lighting effect settings, score the overallness, creativity and coordination of the layout plan through a preset scoring mechanism, and select the plan with the highest score as the recommended layout plan; S6, extracting the time period of the activity, analyzing the influence relationship between the time period of the activity and the layout elements in the layout plan, obtaining the correlation between the scene layout plan and the time period of the activity, and adjusting the recommended layout plan according to the correlation; S7. Visually render the adjusted recommended layout plan to generate a three-dimensional scene model, and optimize and determine the activity scene layout plan.

2. The method for generating an activity scene arrangement scheme according to claim 1, characterized in that: The step S1 specifically includes: Acquire activity scene data, and use the TF-IDF algorithm to process text data in the activity scene data to obtain keyword features of the activity scene data; According to the keyword features, principal component analysis is used to reduce the dimension of the numerical features in the activity scene data to obtain the activity theme feature vector and the participant group feature vector; According to each arrangement element in the pre-established arrangement element material library, the cosine similarity between the arrangement element and the activity theme feature vector is calculated to obtain the theme relevance score of the arrangement element, and the cosine similarity between the arrangement element and the participant group feature vector is calculated to obtain the personnel matching score of the arrangement element; The optimal weights are determined through regression analysis of historical data, and the weighted sum of the subject relevance score and the person matching score is taken to obtain the comprehensive score of the layout elements; Sort the layout elements in descending order according to the comprehensive scores, and select the layout elements whose scores exceed a preset threshold as a set of candidate layout elements; Acquire the color, shape, and material attribute features of each element in the candidate layout element set, and construct a layout element feature matrix; A hierarchical clustering algorithm is used to analyze the arrangement element feature matrix, and clustering results of a suitable level are intercepted according to a preset number of scene areas to obtain a plurality of arrangement element clusters.

3. The method for generating an activity scene arrangement scheme according to claim 2, characterized in that: In the step S1, the establishment of the layout element material library includes: Classify according to the activity theme and the characteristics of the participants, mark relevant tags, and when matching, based on keyword retrieval and feature semantic matching, determine whether the selected elements are in line with the theme and meet the preferences of the participants, and generate several candidate layout plans.

4. The method for generating an activity scene arrangement scheme according to claim 1, characterized in that: The step S2 specifically includes: Acquire point cloud data of the target site, and generate a three-dimensional model of the site based on the point cloud data, wherein the three-dimensional model includes the area size, shape and height space parameters of the site; Measure the lighting intensity at different times and locations to obtain a heat map of the lighting conditions at the site; Establishing a site constraint model according to the three-dimensional model and the lighting condition heat map, wherein the site constraint model includes a space utilization rate threshold and a lighting conflict degree threshold; The candidate layout elements are preliminarily screened by a rule-based filter, and the rule-based filter eliminates the layout elements that exceed the site carrying capacity or seriously conflict with the lighting conditions according to the site constraint model.

5. The method for generating an activity scene arrangement scheme according to claim 1, characterized in that: The step S3 specifically includes: After eliminating the layout elements that exceed the site's carrying capacity or conflict with lighting conditions, the color histogram analysis is used to obtain the color features of the target scene image, and the main color tone, color saturation and brightness parameters in the color features are extracted; According to the color features, the texture features of the scene are extracted using gray-level co-occurrence matrix analysis to obtain texture parameters including roughness, directionality and complexity. Extracting material characteristic parameters of the layout elements from a pre-established material database, wherein the material characteristic parameters include reflectivity, transparency and texture; The word vector model is used to process the description text of the layout elements, obtain style-related keywords and semantic tags, and generate the style semantic vector of the layout elements; Through principal component analysis, color features, texture parameters, material characteristic parameters and style semantic vectors are integrated through dimensionality reduction to construct a multidimensional attribute vector containing color, texture, material and style information.

6. The method for generating an activity scene arrangement scheme according to claim 1, characterized in that: The step S4 specifically includes: Using the DBSCAN clustering algorithm to perform cluster analysis on the multidimensional attribute vectors, and obtaining a plurality of arrangement element combinations by calculating the Euclidean distance between the vectors; According to the combination of layout elements, a simulated annealing algorithm is used to perform spatial layout planning, and the size, shape and placement of the layout elements are encoded as states; According to the status, the objective function is set to evaluate the space utilization, aesthetics and functionality of the layout, and the layout solution is obtained through iterative optimization; Based on the layout plan, according to user preferences or design style guidance, adjust the size and shape of layout elements, determine the proportional relationship and visual balance between elements, generate a color scheme that matches the layout plan, calculate the main color of the layout elements, and use color contrast and coordination principles to determine the overall color scheme of the scene; According to the color scheme and layout plan, the scene effects under different light source positions and intensities are simulated, and the light source parameters are iteratively adjusted to obtain a lighting effect setting with a harmonious and unified overall visual effect.

7. The method for generating an activity scene arrangement scheme according to claim 1, characterized in that: The step S5 specifically includes: Receiving particle swarm optimization algorithm input information carrying space layout planning, color scheme and lighting effect setting parameters, the particle swarm optimization algorithm input information is generated by the initialization process; Iterative optimization is performed according to the input information of the particle swarm optimization algorithm, and the iterative optimization generates multiple candidate layout solutions through speed update and position update operations; Obtaining scoring criteria for the three dimensions of integrity, creativity, and coordination, the scoring criteria being used to evaluate the candidate layout plans; A multi-layer perceptron is used to quantitatively evaluate each candidate layout scheme in terms of integrity, creativity, and coordination, and obtain scores of the candidate layout scheme in the three dimensions; After obtaining the scores of the candidate layout solutions in the three dimensions, a weighted sum method is used to calculate the total score of the candidate layout solutions, wherein the weights are dynamically adjusted according to user feedback through a reinforcement learning algorithm; The candidate arrangement scheme with the highest total score is determined as the recommended arrangement scheme; A simulated annealing algorithm is executed on the recommended arrangement scheme, the simulated annealing algorithm is used to fine-tune and optimize the recommended arrangement scheme, and the recommended arrangement scheme after fine-tune and optimization by the simulated annealing algorithm is output.

8. The method for generating an activity scene arrangement scheme according to claim 1, characterized in that: The step S6 specifically includes: Obtaining time period data in the activity information, and converting the time period data into a Unix timestamp; Assigning a time attribute tag to the layout elements in the layout plan according to the Unix timestamp, wherein the time attribute tag includes an applicable period, lighting requirements, and functional characteristics; Analyze the correlation between time periods and layout elements to obtain the time and element adaptability matrix; Genetic algorithm is used to optimize the layout scheme. The chromosome encoding of genetic algorithm is a binary string of element position and attribute. The fitness function of genetic algorithm is the time and element adaptability matrix. Dynamically adjust the position, lighting, and functional settings of elements based on the optimized layout plan. Dynamic adjustments include moving furniture positions, adjusting light brightness and color temperature, and switching device function modes.

9. The method for generating an activity scene arrangement scheme according to claim 1, characterized in that: The step S7 specifically includes: Constructing a scene basic model, wherein the scene basic model includes a spatial structure, geometric shapes and position information of layout elements; Extract the corresponding material map from the hierarchically organized material library and apply it to the surface of the scene base model to obtain a preliminary visualized 3D scene; Perform real-time lighting rendering on 3D scenes, calculate direct lighting, indirect lighting and shadows, and obtain realistic lighting effects; Set up multiple virtual cameras to capture 3D scene views from different angles and heights, and combine and process the scene views captured by the virtual cameras to achieve smooth perspective switching and roaming effects; Dynamically optimize the 3D scene, automatically adjust the rendering accuracy according to the scene complexity, and obtain a smooth interactive scene preview; receiving a real-time adjustment instruction through a user interaction interface, the real-time adjustment instruction including adjustment information of an object position, material, and lighting parameters; The 3D scene is updated according to the real-time adjustment instructions, while the frame rate and rendering time are monitored, and the one that meets the preset performance threshold is determined as the final activity scene layout plan.

10. A system for generating an activity scene arrangement scheme, using the method for generating an activity scene arrangement scheme according to claims 1 to 9, characterized in that: include: A data processing and analysis module, which is used to obtain activity scene data and analyze activity themes and participant characteristics; A layout element screening module, which is used to retrieve candidate layout elements from a material library according to themes and features; A site constraint processing module, the site constraint processing module is used to select layout elements according to site condition parameters; A layout scheme generation module, which is used to construct attribute vectors, cluster, space layout planning and generate color lighting schemes; A scheme optimization scoring module, which is used to combine and optimize layout schemes and select the best scheme through a scoring mechanism; A visualization rendering module is used to generate a three-dimensional scene model and preview the optimization layout effect.

Images