Color scheme generation method, system and product based on multi-dimensional semantics and visual perception model

Abstract

The invention belongs to the field of computer graphics and artificial intelligence, and relates to a color scheme generation method based on multi-dimensional semantics and a visual perception model, and the method comprises the steps: determining a main color based on multi-dimensional semantics and user preference; generating a candidate color matching scheme set by taking the main color as a reference on the basis of a semantic guidance harmonious template; based on the parameterized perception rule base, perception optimization is conducted on the candidate color matching scheme set, and a final scheme is output and comprises the optimal area proportion parameters of the main color, the auxiliary color and the embellishment color. According to the method, abstract design requirements and specific color parameters are subjected to accurate quantitative association, so that the generated color scheme is ensured to accord with the design intention input by a user from the source. In addition, on the basis that the general design principle is met, the personalized color matching scheme better meeting the aesthetic appreciation of the target user can be generated, and the user acceptability of the scheme is greatly improved.

Description

Technical Field

[0001] This invention belongs to the field of computer graphics and artificial intelligence, and more specifically, relates to an intelligent and personalized multi-color scheme generation method and related system that combines natural language processing, machine learning and psychophysical experimental data. Background Technology

[0002] Color design is a crucial element in industrial design, UI / UX design, and brand image building. Traditional color design processes rely heavily on the designer's personal experience and subjective judgment, resulting in inefficiency, poor reproducibility, and difficulty in meeting diverse user needs.

[0003] To solve this problem, existing technologies mainly employ the following two methods: Color matching tools based on fixed harmony rules, such as Adobe Color, generate color schemes based on classic color harmony theories (such as complementary colors, analogous colors, and triads). These tools can guarantee harmony in the geometric relationship of the color wheel, but their main drawback is: Disconnected from design intent: The generated solutions lack a direct connection with specific design goals (such as "technological feel" and "safety"), requiring designers to perform a large amount of manual screening and secondary creation.

[0004] Lack of personalization: The solution is universal and cannot be personalized based on the specific profile of the target user (such as age, cultural background, professional experience).

[0005] Color recommendation systems based on data mining: Some systems analyze the color combinations of existing design works to discover popular color schemes. The drawback of this type of method is that... "Knowing what, but not why": The recommendation results are statistical correlations, lacking underlying design theories and color psychology basis, and have poor interpretability.

[0006] Lack of innovation: It is easy to fall into the cycle of repeating popular patterns and it is difficult to generate forward-looking and innovative color schemes.

[0007] Furthermore, the aforementioned existing technologies generally overlook a crucial issue: the perceptual consistency of color. That is, a color scheme that is harmonious in terms of numerical (physical parameters) can, in practical applications, create visual illusions due to color interactions (such as the effect of simultaneous contrast) and layout factors (such as area proportions), leading to a deviation between the user's final visual perception and the designer's initial intent. For example, the theoretical primary and secondary color relationships may be reversed due to improper area allocation, and key accent colors may be "swallowed up" by the background due to insufficient contrast.

[0008] Therefore, there is an urgent need in this field for a new technical solution that can systematically integrate abstract design intentions, users' personalized preferences, and objective visual perception laws, thereby automatically generating high-quality multi-color schemes that not only meet design goals and satisfy users' aesthetics, but also maintain visual consistency in practical applications. Summary of the Invention

[0009] In view of the above-mentioned defects or improvement needs of the prior art, the present invention provides a color scheme generation method, system and product based on a multi-dimensional semantic and visual perception model, thereby solving the technical problems of color scheme generation being disconnected from design intent, lacking personalization and ignoring visual perception consistency in the prior art.

[0010] To achieve the above objectives, according to one aspect of the present invention, a color scheme generation method based on a multidimensional semantic and visual perception model is provided, comprising the following steps: Step 1: Determine the primary color based on multidimensional semantics and user preferences; Receive multi-dimensional semantic input: Receive one or more color image words selected by the user from a preset multi-dimensional semantic space, and receive the importance weights assigned by the user to each dimension; Calculate semantic centroids: Based on the semantic words selected by the user and their weights, calculate the semantic centroids in the HSB three-dimensional color space; Correction is performed using a user preference model: the calculated semantic centroid is used as the center, and a search is conducted within its neighborhood. The preference score is then given to multiple candidate colors in the neighborhood based on user preferences. Determine the final primary color: Select the color with the highest user preference score within the neighborhood as the final primary color; Step 2: Generate a set of candidate color schemes based on semantically guided harmonious templates; Based on the determined primary color, a series of candidate multicolor schemes that are theoretically harmonious and in line with the design intent are generated according to the preset semantic-harmony template mapping rules; Step 3: Based on the parameterized perception rule base, perform perception optimization and output the final solution; The candidate schemes are scored and ranked under multiple constraints: For each candidate multi-color scheme generated in step two, a comprehensive score is given from multiple dimensions such as harmony, user preference, and semantic consistency, and the top-N schemes with the highest scores are selected. Optimize using perception rules: For these Top-N schemes, based on the harmony template used in the current scheme, obtain the optimal area ratio parameters of the corresponding main, secondary, and accent colors according to the preset parameterized perception rules; Output the final solution: Finally, present one or more fully optimized color schemes to the user.

[0011] Furthermore, the multi-dimensional semantic space includes at least one of style, function, emotion, and culture dimensions.

[0012] Furthermore, in the step of calculating the semantic centroid, a pre-trained semantic-color mapping model is invoked. This semantic-color mapping model is based on an adaptive mapping algorithm containing a non-linear activation function, which is used to map semantic requirements of different intensities to specific regions of the color space.

[0013] Furthermore, in the step of applying the user preference model for correction, a pre-trained user preference prediction model is invoked to score the preference for multiple candidate colors in the neighborhood; this user preference prediction model is trained by machine learning based on user survey data and is used to predict the probability of user preference for color based on the input user profile and design context.

[0014] Furthermore, the preset semantic-harmony template mapping rule is a pre-built semantic-harmony template mapping rule library. This semantic-harmony template mapping rule library associates multidimensional semantics with color harmony templates, thereby selecting one or more color harmony templates that best express the design intent of the user input based on the received multidimensional semantic input. It also includes generating secondary and accent colors: using the primary color as a reference, the selected color harmony template is applied to calculate the theoretical HSB value range of the secondary and accent colors; sampling is performed within these ranges to generate multiple three-color candidate color schemes that include primary, secondary, and accent colors.

[0015] Furthermore, the parametric perception rule base quantifies, in a parametric form, the visual perception effect caused by factors such as area ratio and simultaneous contrast of multi-color combinations under a specific background.

[0016] Furthermore, the steps for optimizing the application-aware rules include core optimization and / or risk alerts: Core optimizations include recommending the optimal area ratio layout: querying the parameterized awareness rule base and directly obtaining the optimal area ratio parameters verified by experimental data based on the harmony template used in the current scheme; Risk warnings include analyzing perceived risks in the solution based on a parameterized perception rule base and generating natural language prompts; The final output includes presenting users with one or more fully optimized color schemes; each scheme includes HSB color values, optimal area ratio layout suggestions based on human perception experimental data, and perception risk warnings.

[0017] According to another aspect of the present invention, a computer system is provided, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement a color scheme generation method based on a multidimensional semantic and visual perception model as described in any of the preceding claims.

[0018] According to another aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of a color scheme generation method based on a multidimensional semantic and visual perception model as described in any of the preceding claims.

[0019] According to another aspect of the present invention, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the steps of a color scheme generation method based on a multidimensional semantic and visual perception model as described in any of the preceding claims.

[0020] In summary, the technical solutions conceived in this invention, compared with the prior art, can achieve the following beneficial effects: 1. This invention achieves a deep binding between design intent and color scheme: Through an innovative multi-dimensional semantic space and semantic-color mapping model, this invention precisely quantifies and associates abstract design requirements with specific color parameters, ensuring that the generated color scheme conforms to the user's input design intent from the source.

[0021] 2. This invention incorporates personalized user preferences: By introducing a user preference model trained by machine learning, this invention can generate personalized color schemes that better suit the aesthetics of the target users while meeting general design principles, thus greatly improving the user acceptance of the scheme.

[0022] 3. This invention innovatively solves the problem of "perceptual consistency": by constructing and applying a parameterized perceptual rule base based on psychophysical experimental data, it provides data-driven optimal area ratio suggestions for different harmonious schemes. This ensures that the generated schemes are not only theoretically harmonious, but also maintain visual balance and accurately convey design intent in practical applications, bridging the gap between physical color and perceived color, and significantly improving the final quality and usability of the design scheme.

[0023] 4. This invention improves design efficiency and innovation: It automates the complex design decision-making process, greatly shortening the designer's creative cycle. Simultaneously, through the algorithm's global optimization capabilities, it can explore innovative color combinations that are difficult for human designers to discover through experience, opening a new paradigm for human-computer collaborative design. Attached Figure Description

[0024] Figure 1This is an overall flowchart of a preferred embodiment of the present invention.

[0025] Figure 2 This is a schematic diagram of the multidimensional semantic space of a preferred embodiment of the present invention.

[0026] Figure 3 This is a schematic diagram illustrating the construction process of the parameterized awareness rule base according to a preferred embodiment of the present invention.

[0027] Figure 4 This is a screenshot of a system user interface according to a preferred embodiment of the present invention.

[0028] Figure 5 This is a screenshot of the simultaneous contrast effect quantification experiment of a preferred embodiment of the present invention.

[0029] Figure 6 This is a screenshot of an experiment showing the effect of the area effect on color perception in a preferred embodiment of the present invention.

[0030] Figure 7 This is a screenshot of an experimental study on the spatial perception effect of a preferred embodiment of the present invention.

[0031] Figure 8 These are screenshots from a feasibility study of microscopic perception correction according to a preferred embodiment of the present invention.

[0032] Figure 9 These are screenshots from an experiment studying the area ratio effect and multicolor balance in a preferred embodiment of the present invention.

[0033] Figure 10 This is a screenshot of an experiment exploring the optimal area ratio based on the adjustment method in a preferred embodiment of the present invention. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0035] This invention discloses a color scheme generation method based on a multidimensional semantic and visual perception model, which includes at least the following steps: Step 1: Determine the primary color based on multidimensional semantics and user preferences.

[0036] This step aims to transform the abstract design requirements input by the user into a precise primary color HSB (hue, saturation, brightness) value that is both semantically accurate and meets the preferences of the target user.

[0037] 1.1 Receiving Multidimensional Semantic Input: The system receives one or more color image words selected by the user from a preset multidimensional semantic space (including at least style, function, emotion, and culture dimensions), and receives the importance weights assigned by the user to each dimension. This semantic space is constructed through natural language processing of a large-scale text corpus, ensuring the structured and comprehensive nature of the semantics.

[0038] 1.2 Calculating the Semantic Centroid: The system invokes a pre-trained semantic-color mapping model to calculate a weighted "semantic centroid" in the HSB 3D color space based on the semantic words selected by the user and their weights. This model is based on an adaptive mapping algorithm that includes a non-linear activation function (such as the hyperbolic tangent function tanh), which can accurately map semantic requirements of different intensities to specific regions of the color space.

[0039] 1.3 Application of User Preference Model for Correction: The system searches within a neighborhood centered on the semantic centroid calculated in step 1.2. Simultaneously, a pre-trained user preference prediction model (such as a gradient boosting decision tree (GBDT) model) is invoked to score the preference for multiple candidate colors within this neighborhood. This preference model is trained using machine learning on a large amount of user survey data and can predict the probability of a user's color preference based on the input user profile (such as age, design experience) and design context (such as product category).

[0040] 1.4 Determine the final primary color: The system selects the color with the highest user preference score in the neighborhood as the final primary color (C_main).

[0041] Step 2: Generate a set of candidate color schemes based on semantically guided harmonious templates.

[0042] This step aims to generate a series of candidate multicolor schemes that are theoretically harmonious and in line with the design intent, based on the determined primary color.

[0043] 2.1 Dynamic Selection of Harmony Templates: The system queries a pre-built "semantic-harmony template mapping rule base". This rule base associates multi-dimensional semantics (such as "bright and vibrant") with classic color harmony templates (such as "split complementary colors" and "tricolor groups"). Based on the semantics input in step 1.1, the system automatically selects one or more harmony templates that best express the design intent.

[0044] 2.2 Generating Secondary and Accent Colors: Using the primary color (C_main) as a baseline, the system applies a selected harmonic template to calculate the theoretical HSB value ranges for the secondary color (C_aux) and accent color (C_acc). Sampling is then performed within these ranges to generate multiple candidate three-color schemes containing primary, secondary, and accent colors.

[0045] Step 3: Based on the parameterized perception rule base, perform perception optimization and output the final solution.

[0046] This step is the core innovation of this invention, aiming to optimize the theoretically harmonious scheme to ensure that it conforms to the laws of human visual perception and achieve "perceptual consistency".

[0047] 3.1 Construction of a Parametric Perception Rule Base: The data for this rule base was obtained through a series of psychophysical experiments. The experiments systematically quantified the visual perception effects of multi-color combinations under specific backgrounds, caused by factors such as area ratio and simultaneous contrast. Unlike the IF-THEN-style symbol rules of existing technologies, the rule base of this invention is parametric, storing core parameters such as the mean and standard deviation of the optimal area ratio for different harmonious templates.

[0048] 3.2 Multi-constraint scoring and ranking of candidate solutions: The system comprehensively scores each candidate solution generated in step two from multiple dimensions such as harmony, user preference, and semantic consistency, and selects the top-N solutions with the highest scores.

[0049] 3.3 Optimization using perception rules: For these Top-N solutions, the system performs the following perception optimization operations: Core optimization: Recommended optimal area ratio layout: The system queries the parameterized awareness rule base and directly obtains the optimal area ratio parameters (e.g., main color 51.4%, secondary color 33.0%, accent color 15.6%) based on the harmonious template used in the current scheme (e.g., "three-color group"), which are verified by experimental data and driven by data.

[0050] Risk warning: The system can also analyze the perceptual risks (such as color bias) that may exist in the scheme due to simultaneous contrast effects, based on the rule base, and generate natural language prompts.

[0051] 3.4 Final Output: The system ultimately presents the user with one or more fully optimized color schemes. Each scheme not only includes accurate HSB color values, but also an optimal area ratio layout recommendation based on human perception experimental data, as well as necessary perception risk warnings, forming a complete design solution with high practical guidance.

[0052] Example 1: Software Implementation of a Color Scheme Generation Method Based on Multidimensional Semantics and Visual Perception Model A preferred embodiment of the present invention is implemented in a desktop application called "AI Color Workshop". This application is developed using the Python programming language and the PyQt5 graphical user interface library. The following will describe in detail the specific implementation method of the color scheme generation method based on a multi-dimensional semantic and visual perception model of the present invention within the software execution flow.

[0053] 1. System initialization and data loading In this embodiment, before the method starts executing, the system first initializes by loading multiple model and rule base files required for operation.

[0054] Loading the semantic-color mapping rule library: This involves calling a module, as shown in the `load_mapping_rules` function in `app_utils.py`, to load rules from a pre-defined JSON file (e.g., "Adaptive Mapping Rule Library.json"). This JSON file uses dimensions (e.g., `styleDimensions`) as keys and a list of semantic subclasses (e.g., "Bright Vibrant") as values. Each subclass object defines its baseline HSB parameter range (H_range, S_range, B_range) and control parameters for non-linear mapping (kappa_H, alpha_S, etc.). The function parses and preprocesses this data (e.g., calculating the hue midpoint H0 and span) before storing it in memory for later use.

[0055] Load the user preference prediction model: Call the module shown in the `_load_ai_model` function in `main.py` to load a pre-trained gradient boosting decision tree model (`preference_model.pkl`), a label encoder (`label_encoder.pkl`), and a JSON file (`model_features.json`) containing the features required by the model, using the `joblib` library. Simultaneously, load a CSV file (`user_preference_vectors.csv`) containing all user abstract preference vectors for subsequent similarity calculations.

[0056] Loading the harmony and perception rule base: The `load_harmony_rules` function is called to load the semantic-harmony template mapping rule library from the JSON file ("harmony_rules.json").

[0057] The load_advanced_illusion_rules function is called to load the parameterized perception rule library from the JSON file ("advanced_perception_rules.json").

[0058] 2. Color Scheme Generation Steps Step 1: Determine the primary color based on multidimensional semantics and user preferences. In this embodiment, this step is implemented through user interaction and background calculation of the application's "AI Color Mapper" module, as follows: Receiving multi-dimensional semantic input: Users select one or more active semantic dimensions (such as "style" or "function") via checkboxes (QCheckBox) on the graphical interface. Under each active dimension, users select a specific semantic subclass (such as "Technological Simplicity") via a dropdown (QComboBox) and set the importance weight w and performance intensity C of that dimension via a slider (QSlider). The system acquires these user input values ​​in real time.

[0059] Calculating independent colors for each dimension: For each activated dimension, the system calls the `map_hsb` function in `app_utils.py` based on the user-selected semantic subclass and intensity C value. This function performs the following calculations: A normalized net weight P = 2*C-1 is calculated based on the intensity C, and its range is [-1, 1].

[0060] By applying a nonlinear mapping formula that includes a hyperbolic tangent function, such as dH=tanh(P*kappa_H)*alpha_H*(span / 2), the offsets dH, dS, and dB of hue, saturation, and brightness relative to the reference midpoint are calculated respectively.

[0061] The offset is added to the reference midpoint, and the clamp function is used to ensure that the final HSB value falls within the preset valid range, thereby generating an independent HSB color value for that dimension.

[0062] Weighted fusion generates the semantic primary color: As shown in the `_calculate_main_color` function in `main.py`, the system performs a weighted average of the independent HSB color values ​​generated from all activated dimensions according to the user-defined weight `w`. Specifically, for the circular hue channel, a vector averaging method is used for fusion calculation (`avg_h=np.rad2deg(np.angle(np.sum([w*np.exp(1j*np.deg2rad(h))...])))`) to ensure the accuracy of the hue averaging. This results in a "semantic center point" color that accurately reflects the user's multi-dimensional semantic needs, serving as the initial primary color.

[0063] User preference model correction: In another embodiment of the invention, a preference correction step can be added after step 2.3. The system generates several candidate primary colors within the neighborhood of the calculated initial primary color in its HSB space. Then, using the current user profile (obtained from the "AI Intelligent Recommender" module) as input, the loaded GBDT preference model is called to score the preference of these candidate primary colors. Finally, the candidate color with the highest score is selected as the final primary color (C_main). In the simplified process of this embodiment, the _calculate_main_color function directly uses the semantic primary color as the final primary color.

[0064] Step 2: Generate a set of candidate color schemes based on semantically guided harmonious templates. Dynamic selection of harmonic templates: As shown in the `_on_generate_scheme` function in `main.py`, the system first calls the `_get_active_semantic_class` function to obtain the most important semantic subclass (e.g., determined based on the highest weight or user-specified priority dimension, such as "Technology and Simplicity"). Then, it calls the `get_harmony_template_for_semantic` function in `app_utils.py`, using this semantic subclass as an index to query the loaded semantic-harmony template mapping rule library, and randomly or by priority selects a suitable harmonic template name (e.g., "Analogous" or "Monochromatic").

[0065] Generating secondary and accent colors: The system then calls the `apply_harmony_template` function in `app_utils.py`. This function takes the primary color HSB value determined in step one and the harmony template name selected in step 3.1 as input. Internally, the function contains a conditional structure that performs different geometric operations to calculate the HSB values ​​of the secondary and accent colors based on different template names (such as "Analogous", "Complementary", "Triadic", etc.). For example: For the "Analogous" template, the hue of the secondary color is increased by 30 degrees from the hue of the primary color, and the hue of the accent color is decreased by 30 degrees.

[0066] For the "Complementary" template, the secondary color hue is 180 degrees greater than the primary color.

[0067] For the "Triadic" template, the hues of the secondary and accent colors are increased by 120 degrees and 240 degrees respectively based on the primary color. Simultaneously, the function fine-tunes the saturation and brightness (e.g., using the `adjust_s_b` function) to enhance the color depth. Ultimately, a candidate tricolor scheme containing the primary, secondary, and accent colors is generated. In this embodiment, to simplify the process, only one optimal candidate scheme is generated.

[0068] Step 3: Based on the parameterized perception rule base, perform perception optimization and output the final solution. The system calls the perceptual rule library for analysis: As shown in the update_illusion_tips function in main.py, after generating the color scheme, the system will immediately call the check_advanced_illusion_rules function in app_utils.py.

[0069] The `check_advanced_illusion_rules` function performs the following core optimization and analysis operations to optimize and generate output: Recommending the optimal area ratio layout: This function queries the loaded parameterized perception rule library (advanced_perception_rules.json) using the harmony template name obtained in step two (e.g., "Triadic") as the key. The library's `optimal_ratio_adjustment.optimal_ratios` field stores the optimal area ratio parameters for different harmony templates, derived from psychophysical experiments. The function reads these parameters (e.g., `{"main":0.514,"aux":0.330,"acc":0.156}`) and formats them into a natural language suggestion text.

[0070] Risk warning: The function also checks features such as contrast of the scheme and generates warning text about potential perceived risks (such as color cast) based on modules such as simultaneous_contrast_model in the rule base.

[0071] Outputting the final solution: The system presents the final color scheme (containing three HSB color values) and the optimization suggestion text generated in step 4.2 to the user. In the graphical interface of this embodiment, the three HSB color values ​​are displayed in three custom ColorSchemeSlot controls, while the optimization suggestion text is displayed in the illusion_tip_label control. This constitutes a complete design solution that includes color values ​​and layout guidance.

[0072] The above is a detailed description of the specific embodiments of the present invention. Through the above steps, the present invention successfully integrates multidimensional semantics, user preferences, harmony theory, and visual perception laws, realizing an automated, intelligent, and highly practical method for generating multi-color schemes.

[0073] Example 2: Experimental Construction Method of Parametric Aware Rule Base This embodiment details the construction process of a preferred parametric perception rule base (advanced_perception_rules.json and visual perception rule base.json, which are called by the software in Embodiment 1). This process is accomplished through a series of rigorous psychophysical experiments, aiming to provide a solid, data-driven scientific basis for the color generation method of this invention.

[0074] Overall Experimental Design Objective: To bridge the gap between the "physical properties" of color and the "perceptual effects" of users, this series of experiments aims to systematically quantify the key factors affecting users' visual perception in a specific application scenario (uniform light gray background), including but not limited to the simultaneous contrast effect, area effect, spatial perception effect, and macroscopic balance under multi-color combinations.

[0075] General Settings: Participants: All experiments recruited more than 30 participants with normal or corrected visual acuity and no color blindness or color weakness.

[0076] Environment and Equipment: Experiments were conducted in a standardized lighting environment using professionally color-calibrated monitors to ensure accurate color reproduction.

[0077] Fixed background color: To enhance the applicability validity of the study, all visual stimuli were presented on a uniform light gray background (HSB value approximately 0,0,0.85), which is consistent with the default background of the AI ​​image generator in the software described in Example 1.

[0078] Experimental Sequence 1: Quantification of Basic Visual Perception Laws This series of experiments aims to construct a visual perception rule base .json, providing warnings and explanations for the perception effect of monochrome.

[0079] Experiment 1: Quantification of Simultaneous Contrast Effect Objective: To quantify the perceived brightness, saturation, and hue shift of a target color against a light gray background.

[0080] Method: An asymmetric matching task was used. A target color patch (CT) was presented on one side of the screen against a light gray background, while an adjustable color patch (CA) was presented on the other side against the same background. Participants were asked to adjust the HSB slider to make the perceived effect of the CA perfectly match that of the CT.

[0081] Conclusions and Rule Transformation: The one-sample t-test results of the experiment show that specific colors produce significant perceptual changes against a light gray background. For example, a gray patch with a physical HSB value of approximately (0,0,74) will have a significantly increased perceived brightness of 1.6 units (p=0.013). This quantification result was transformed into a rule (e.g., rule ID: LCB_001) and stored in the rule base. This rule will be used to prompt users that similar colors may be perceived as brighter when the software generates them.

[0082] Experiment 2: The Influence of Area Effect on Color Perception Objective: To investigate the effect of changes in color area on perceived brightness and saturation.

[0083] Method: A matching task was used. A target color block of a specific size (e.g., 10x10, 70x70, 300x300 pixels) was presented on one side of the screen, while a fixed, adjustable color block of medium size was presented on the other side. Participants were asked to adjust the latter to match their perceived effect with the target color blocks of different sizes.

[0084] Conclusions and Rule Transformation: Repeated measures ANOVA results showed that area has a significant impact on the perceived brightness of certain colors. For example, for a highly saturated, low-brightness color (HSB approx. 164, 90, 36), its perceived brightness increases significantly by approximately 9.2 units (p < 0.01) when its display area is reduced from a large (e.g., 300x300 pixels) to a very small (10x10 pixels). This effect was quantified and transformed into a rule (e.g., rule ID: AREA_V_001) to remind users that some dark colors may appear brighter than expected when designing small-sized elements.

[0085] Experiment 3: Spatial Perception Effect (Forward / Backward Color) Objective: To quantify the differences in perceived spatial distance between different colors against a light gray background.

[0086] Method: A forced choice task with two options is used. A pair of colors (e.g., warm color vs. cool color) are presented on the screen simultaneously, and participants are asked to judge "which color block looks closer?".

[0087] Conclusion and rule transformation: The chi-square test results showed that users' choices significantly deviated from a uniform distribution (χ²(2)=10.5, p=0.005), clearly verifying that under a light gray background, warm colors are significantly more likely to be perceived as "advancing colors" compared to cool colors. This conclusion was transformed into a rule (e.g., rule ID: ADVREC_001) to guide users on how to choose colors when they need to emphasize or weaken the sense of hierarchy of interface elements.

[0088] Experimental Sequence 2: Exploring the Macroscopic Perception Patterns under Multicolor Combinations This series of experiments aims to construct the advanced_perception_rules.json file, providing optimization parameters for the macroscopic layout of multi-color schemes.

[0089] Experiment 1: Feasibility Exploration of Microscopic Sensing Correction Objective: To attempt to construct a machine learning model that can accurately predict and compensate for the contrast effect when two colors are placed side by side.

[0090] Methods: Using the classic "center-periphery" paradigm and an asymmetric matching task, we systematically measured the mutual perceived influence between 40-50 pairs of colors with different color relationships (complementary, similar, etc.).

[0091] Conclusions and Rule Transformation: The t-test and cross-validation results of the gradient boosting regression tree model (R² values ​​ranging from -0.04 to 0.24) jointly demonstrate that constructing a universal, high-precision micro-sensing correction model is extremely difficult due to strong individual differences. This important finding has been transformed into a "meta-rule" stored in the rule base: explicitly marking the recommendation level of the micro-sensing correction model as "Low," guiding the software system in Example 1 to avoid risks and not blindly rely on simple physical models for micro-color compensation.

[0092] Experiment 2: Confirmatory Study of Classic Area Ratios Objective: To examine the harmony and balance of different harmonious templates (such as analogous colors, split complementary colors, and triads) under several classic preset area ratios (such as 7:2:1, 6:3:1, and 5:4:1).

[0093] Methods: A seven-point Likert scale was used for the rating task. Participants were presented with abstract color block compositions with different combinations of harmonious templates and area ratios, and were asked to rate the "overall harmony" and "visual balance".

[0094] Conclusions and Rule Transformation: Two-way repeated measures ANOVA results reveal that color relationships (harmony templates) are the most critical factor influencing aesthetic scores (p<.001), with an impact far exceeding that of several preset classic area ratios. Simultaneously, descriptive statistics show that the classic "6:3:1" ratio scores relatively high in high-contrast templates. These conclusions, parameterized and stored in a rule base, will be used to explain the universality of different harmony template layouts to users and provide a relatively safe classic layout reference for certain scenarios.

[0095] Experiment 3: Exploratory Study on the Optimal Area Ratio (Core Experiment) Objective: Based on previous experiments, to use a more refined adjustment method to accurately determine the "optimal" area ratio in the user's mind for different harmonious templates.

[0096] Method: An adjustment method was used. Participants were presented with a three-color combination and provided with a two-slider interactive system that allowed them to freely adjust the area ratio of the primary, secondary, and accent colors in a continuous space until they considered the current scheme to be "most harmonious".

[0097] Conclusions and Rule Transformation: This experiment successfully quantified the optimal area ratio for user preferences. Data analysis clearly reveals a systematic correlation between the optimal area ratio and the contrast intensity of the color template. Specific parameters are as follows: For high-contrast templates (such as the Triadic color scheme), the mean optimal area ratio is: primary color 51.4% (±21.9%), secondary color 33.0% (±16.5%), and accent color 15.6% (±13.6%).

[0098] For weak contrast templates (such as Analogous), the mean of the optimal area ratio is: primary color 48.3% (±19.6%), secondary color 32.5% (±14.5%), and accent color 19.2% (±16.6%).

[0099] The precise ratio parameters, including the mean and standard deviation, derived from the averaging of real user perception data, are embedded in the `optimal_ratio_adjustment` module of the `advanced_perception_rules.json` file. These constitute the direct, data-driven "golden parameters" for the "recommend optimal area ratio layout" function in the software system described in Example 1.

[0100] Through a series of progressive experiments described in this embodiment, the present invention not only reveals various color perception laws, but more importantly, it successfully transforms these laws from qualitative academic discoveries into structured and quantitative parameters and rules that can be directly invoked by computers, laying a solid technical foundation for realizing a truly scientific and intelligent color design system.

[0101] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A color scheme generation method based on a multidimensional semantic and visual perception model, characterized in that, Includes the following steps: Step 1: Determine the primary color based on multidimensional semantics and user preferences; Receive multi-dimensional semantic input: Receive one or more color image words selected by the user from a preset multi-dimensional semantic space, and receive the importance weights assigned by the user to each dimension; Calculate semantic centroids: Based on the semantic words selected by the user and their weights, calculate the semantic centroids in the HSB three-dimensional color space; Correction is performed using a user preference model: the calculated semantic centroid is used as the center, and a search is conducted within its neighborhood. The preference score is then given to multiple candidate colors in the neighborhood based on user preferences. Determine the final primary color: Select the color with the highest user preference score within the neighborhood as the final primary color; Step 2: Generate a set of candidate color schemes based on semantically guided harmonious templates; Based on the determined primary color, a series of candidate multicolor schemes that are theoretically harmonious and in line with the design intent are generated according to the preset semantic-harmony template mapping rules; Step 3: Based on the parameterized perception rule base, perform perception optimization and output the final solution; The candidate schemes are scored and ranked under multiple constraints: For each candidate multi-color scheme generated in step two, a comprehensive score is given from multiple dimensions such as harmony, user preference, and semantic consistency, and the top-N schemes with the highest scores are selected. Optimize using perception rules: For these Top-N schemes, based on the harmony template used in the current scheme, obtain the optimal area ratio parameters of the corresponding main, secondary, and accent colors according to the preset parameterized perception rules; Output the final solution: Finally, present one or more fully optimized color schemes to the user.

2. The color scheme generation method based on a multidimensional semantic and visual perception model according to claim 1, characterized in that, The multidimensional semantic space includes at least one of the following dimensions: style, function, emotion, and culture.

3. The color scheme generation method based on a multidimensional semantic and visual perception model according to claim 1, characterized in that, In the step of calculating the semantic centroid, a pre-trained semantic-color mapping model is invoked. This model is based on an adaptive mapping algorithm that includes a non-linear activation function, which is used to map semantic requirements of different intensities to specific regions of the color space.

4. The color scheme generation method based on a multidimensional semantic and visual perception model according to claim 1, characterized in that, In the step of applying the user preference model for correction, a pre-trained user preference prediction model is invoked to score the preference for multiple candidate colors in the neighborhood. This user preference prediction model is trained by machine learning based on user survey data and is used to predict the probability of a user's preference for a color based on the input user profile and design context.

5. The color scheme generation method based on a multidimensional semantic and visual perception model according to claim 1, characterized in that, The preset semantic-harmony template mapping rule is a pre-built semantic-harmony template mapping rule library. This semantic-harmony template mapping rule library associates multidimensional semantics with color harmony templates, thereby selecting one or more color harmony templates that best express the design intent of the user input based on the received multidimensional semantic input. It also includes generating secondary and accent colors: using the primary color as a reference, the selected color harmony template is applied to calculate the theoretical HSB value range of the secondary and accent colors; sampling is performed within these ranges to generate multiple three-color candidate color schemes that include primary, secondary, and accent colors.

6. The color scheme generation method based on a multidimensional semantic and visual perception model according to claim 1, characterized in that, The parametric perception rule base quantifies, in a parametric form, the visual perception effect caused by factors such as area ratio and simultaneous contrast of multi-color combinations under a specific background.

7. The color scheme generation method based on a multidimensional semantic and visual perception model according to claim 6, characterized in that, The steps for optimizing the application-aware rules include core optimization and / or risk alerts: Core optimizations include recommending the optimal area ratio layout: querying the parameterized awareness rule base and directly obtaining the optimal area ratio parameters verified by experimental data based on the harmony template used in the current scheme; Risk warnings include analyzing perceived risks in the solution based on a parameterized perception rule base and generating natural language prompts; The final output includes presenting users with one or more fully optimized color schemes; each scheme includes HSB color values, optimal area ratio layout suggestions based on human perception experimental data, and perception risk warnings.

8. A computer system comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the color scheme generation method based on a multidimensional semantic and visual perception model as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the color scheme generation method based on a multidimensional semantic and visual perception model as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the color scheme generation method based on a multidimensional semantic and visual perception model as described in any one of claims 1 to 7.

Images