A Design Method, System, Device and Medium for Woodcarving Window Grilles of Ancient Folk Houses

Abstract

This application relates to the field of computer vision, and discloses a design method, system, device and medium for ancient residential woodcarving window grilles by combining the Stable Diffusion model and intuitionistic fuzzy VIKOR. The method includes: constructing a woodcarving decoration pattern database, and using the Stable Diffusion model to learn and train in the woodcarving decoration pattern database to generate woodcarving patterns in the target style; using intuitionistic fuzzy VIKOR to sort the generated woodcarving patterns according to aesthetic priorities, and selecting multiple excellent samples; drawing sketches and evaluating the schemes for the multiple excellent samples, selecting the optimal design product sketch for modeling to obtain a positioned-designed modeling product; verifying the lighting parameters of the modeling product under different types of window grille decorations through simulation experiments to screen out the best product design. This method provides a way to quickly generate and optimize complex woodcarving window grille patterns, improve design efficiency and explore diverse innovative solutions.

Description

Technical Field

[0001] The present invention relates to the field of computer vision, and specifically to a design method, system, device and medium for woodcarved window grilles of ancient dwellings by combining the Stable Diffusion model and Intuitionistic Fuzzy VIKOR. Background Technique

[0002] As an important decorative element of ancient dwellings, woodcarved window grilles are not only key physical materials for studying the ancient architectural structure, layout and aesthetic style, but also provide important basis for exploring ancient Chinese social life and cultural beliefs. The themes and forms of expression of traditional woodcarved window grilles are relatively fixed, and there is less combination with modern life and fashion culture. This art form lacking innovation and sense of the times is difficult to attract the attention of the general public, resulting in the shrinking of the non-material cultural heritage woodcarving market. The inheritance of traditional woodcarving techniques mainly relies on the collective memory of the community and is passed down from generation to generation in the master-apprentice or family mode. However, due to the complex process, high labor intensity, long learning cycle and low economic return, many young people are reluctant to learn woodcarving techniques, resulting in a sharp decrease in the number of inheritors and the verge of loss of the techniques. Therefore, it is necessary to integrate advanced technologies and innovative designs into the non-material cultural heritage woodcarving to revive it.

[0003] Although scholars have conducted relatively in-depth research on the artistic characteristics, historical background and aesthetic value of woodcarved window grille patterns, these theoretical results provide rich perspectives for understanding and appreciating woodcarved window grilles. However, the research mostly focuses on a single discipline, lacking cross-field comprehensive exploration, and even more lacking innovative designs for woodcarved window grilles in modern life, making it difficult to meet the increasingly diverse aesthetic needs of contemporary society. This limitation makes traditional woodcarving art face many challenges in the process of integrating with modern design, and urgently needs to achieve breakthroughs and innovations with the help of new technologies and interdisciplinary research. Summary of the Invention

[0004] Based on this, the present application proposes a design method, system, device and medium for woodcarved window grilles of ancient dwellings by combining the Stable Diffusion model and Intuitionistic Fuzzy VIKOR, aiming to provide a method for quickly generating and optimizing complex woodcarved window grille patterns, improving the design efficiency and exploring diverse innovative solutions, thereby launching a fashionable lighting product design with both aesthetics and functionality, attempting to re-attract the attention of the young group to traditional cultural heritage and promoting the intelligent and digital innovation of woodcarving patterns.

[0005] The first aspect of the present application provides a design method for woodcarved window grilles of ancient dwellings, and the method includes:

[0006] Step S1: Construct a woodcarving decoration pattern database, and use the Stable Diffusion model to learn and train in the woodcarving decoration pattern database to generate woodcarving patterns of the target style;

[0007] Step S2: Use intuitionistic fuzzy VIKOR to perform aesthetic priority ranking on the generated woodcarving patterns, and select multiple excellent samples;

[0008] Step S3: Sketch and evaluate the multiple excellent samples, select the sketch of the optimal design product for modeling, and obtain the positioned-designed modeled product;

[0009] Step S4: Verify the lighting parameters of the modeled product under different types of window flower decorations through simulation experiments to screen out the best product design.

[0010] As an optional implementation manner of the first aspect, the step S1 includes: Step S11: Collect a large number of woodcarving window flower pictures to construct a woodcarving decoration pattern database; Step S12: Select a wireframe of a window flower pattern with a target style from the woodcarving decoration pattern database, and perform text description on the wireframe of the window flower pattern to generate a training data set of text-image pairs; Step S13: Use the training data set to fine-tune the model based on the pre-trained StableDiffusion model by applying the LoRA algorithm to obtain a target StableDiffusion model; Step S14: Generate woodcarving patterns with a target style according to the target StableDiffusion model.

[0011] As an optional implementation manner of the first aspect, the step S13 includes: In the process of fine-tuning the model by the LoRA technique based on the open-source pre-trained StableDiffusion model, the generation of the woodcarving pattern is realized by using the reverse diffusion process from adding noise to denoising, specifically including: gradually adding noise conforming to the normal distribution to the patterns in the training data set to generate noisy patterns; gradually removing the noise from the noisy patterns by using the Markov chain, and calculating the estimated conditional probability according to the Bayesian formula at each step.

[0012] As an alternative implementation of the first aspect, the step S2 includes: Step S21: Screening and multi-dimensional evaluation of the wood carving patterns to obtain evaluation index information and scores; Step S22: Defining an intuitionistic fuzzy set, including the satisfaction degree and dissatisfaction degree of the screened wood carving patterns, and defining the uncertainty degree according to the satisfaction degree and the dissatisfaction degree; Step S23: Obtaining the intuitionistic fuzzy entropy according to the satisfaction degree, the dissatisfaction degree and the uncertainty degree; Step S24: Using the minimum deviation optimization model to determine the index weights according to the intuitionistic fuzzy entropy; Step S25: Determining the expert weights according to the index weights of the intuitionistic fuzzy entropy; Step S26: Converting to a group decision matrix according to the expert weights to determine the positive ideal solution and negative ideal solution of each wood carving pattern under each index; Step S27: Using the intuitionistic fuzzy VIKOR method to calculate the group utility value, individual regret value and compromise evaluation value of each wood carving pattern according to the positive ideal solution and the negative ideal solution, so as to rank the wood carving patterns and select multiple excellent samples.

[0013] As an alternative implementation of the first aspect, the step S27 includes: Ranking the wood carving pattern samples according to the group utility value, the individual regret value and the compromise evaluation value; If the compromise evaluation value of a wood carving pattern is the minimum value and simultaneously satisfies the following two conditions, then the said wood carving pattern is considered the optimal sample, where: Condition 1: , where and are respectively the minimum value and the second minimum value in the ranking of the compromise evaluation value, m is the number of samples of the wood carving pattern; Condition 2: Both are the minimum values in the group utility value ranking and the individual regret value ranking; If only Condition 1 is satisfied, then all the wood carving patterns are close to the ideal solution; If only Condition 2 is satisfied, then the advantages and disadvantages are selected according to the ranking of the compromise evaluation values of all the wood carving patterns.

[0014] As an alternative implementation of the first aspect, the step S3 includes: Step S31: Clarifying the product positioning and determining the basic design framework; Step S32: Incorporating the multiple excellent samples into the product object corresponding to the clarified product positioning and the determined basic design framework, performing sketch drawing and scheme evaluation, and selecting the optimal design product sketch; Step S33: Modeling the optimal design product sketch to obtain the modeled product with positioned design.

[0015] As an alternative implementation of the first aspect, the step S4 includes: Step S41: Configuring the light source for the modeled product to obtain a ray tracing diagram and an irradiance analysis diagram to calculate the uniformity of light distribution; Step S42: Adjusting the material of the modeled product according to the uniformity of light distribution to select the best product design.

[0016] The second aspect of the present application provides a design system for ancient residential woodcarving window grilles, and the system includes:

[0017] A pattern generation module, configured to construct a woodcarving decoration pattern database, and use a stable diffusion model to learn and train in the woodcarving decoration pattern database to generate woodcarving patterns in a target style;

[0018] A pattern evaluation module, configured to perform aesthetic priority ranking on the generated woodcarving patterns by using intuitionistic fuzzy VIKOR, and select multiple excellent samples;

[0019] A product design module, configured to draw sketches and evaluate the solutions for the multiple excellent samples, select the optimal design product sketch for modeling, and obtain a modeled product with positioning design;

[0020] A product verification module, configured to verify the lighting parameters of the modeled product under different types of window grille decorations through simulation experiments to screen out the best product design.

[0021] The third aspect of the present application provides an electronic device, including: a processor; a memory for storing executable instructions that can be executed by the processor; wherein, the processor is configured to execute the executable instructions to implement the above-mentioned method for designing ancient residential woodcarving window grilles.

[0022] The fourth aspect of the present application provides a computer-readable storage medium, when the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device can execute the above-mentioned method for designing ancient residential woodcarving window grilles.

[0023] Compared with the prior art, a design method for ancient residential woodcarving window grilles provided by the present application: First, in step S1, a database containing rich and diverse woodcarving decoration patterns is constructed, and the Stable Diffusion model is used for in-depth learning and training. The application of this cross-domain data integration and machine learning technology breaks through the research limitations of traditional single disciplines, realizes the integration of different art forms and design concepts, and provides rich materials and possibilities for innovative design. Then, in step S2, the intuitionistic fuzzy VIKOR method is introduced to conduct aesthetic evaluation and priority ranking on the generated woodcarving patterns. This method combines the intuitionistic fuzzy set theory and multi-attribute decision analysis, can comprehensively consider and quantify various aesthetic factors, ensure that the selected woodcarving patterns not only have artistic value but also conform to the public aesthetic, reflecting the comprehensive exploration of interdisciplinary disciplines. Step S3 further deepens the design process. By drawing sketches and evaluating the schemes for the selected samples, designers can optimize and adjust the design scheme according to actual needs and market feedback, and finally determine the optimal design product sketch and conduct modeling. This stage emphasizes the humanization and practicality of the design, making the woodcarving window grilles not only beautiful but also adaptable to the actual needs of modern life. Finally, in step S4, the lighting performance of the modeled product is tested through simulation experiments. This step ensures the functionality and comfort of the designed woodcarving window grilles in actual applications, and also reflects the use of modern scientific and technological means, making the traditional woodcarving art more closely integrated with modern life. Therefore, through steps such as constructing a woodcarving decoration pattern database, using intuitionistic fuzzy VIKOR for aesthetic evaluation, sketch drawing and scheme evaluation, and simulation experiment verification, the present technical solution realizes the innovative design of ancient residential woodcarving window grilles. This solution solves the problems that current research mainly focuses on single disciplines, lacks cross-domain comprehensive exploration, and lacks innovative design of woodcarving window grilles in modern life, and meets the growing diverse aesthetic needs of contemporary society.

[0024] Additional aspects and advantages of the present application will be given in part in the following description, become apparent in part from the following description, or be understood through the embodiments of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a flowchart of a design method for ancient residential woodcarving window grilles proposed in the first embodiment of the present application;

[0026] Figure 2 It is a schematic structural diagram of a design system for ancient residential woodcarving window grilles proposed in the second embodiment of the present application.

[0027] The following specific embodiments will further illustrate the present application in conjunction with the above-mentioned drawings. SPECIFIC EMBODIMENTS

[0028] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0029] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here. In addition, the "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally represents that the objects associated with each other are in an "or" relationship.

[0030] In order to illustrate the technical solution described in this application, a specific embodiment is provided below for illustration.

[0031] Example 1

[0032] See also Figure 1 , which is a flow chart of a method for designing wooden window grilles in ancient dwellings proposed in the first embodiment of the present application, and the proposed method includes S1 to S4.

[0033] Step S1: construct a woodcarving decorative pattern database, and use the stable diffusion model to learn and train in the woodcarving decorative pattern database to generate woodcarving patterns of the target style.

[0034] Specifically, this step S1 includes the following steps S11 to S14.

[0035] Step S11: Collect a large number of woodcarving window grille pictures to build a woodcarving decorative pattern database.

[0036] Exemplarily, woodcarving window paper-cut images are widely collected from various online and offline platforms, including but not limited to major websites, museums, art galleries, etc. Through in-depth research, traditional woodcarving window paper-cut images collected from ancient residential houses, villages, ancestral halls, ancient towns, temples and other places were preliminarily screened out, and a total of 2476 images were collected in this embodiment.

[0037] However, not all of the collected images are suitable for building a database. Therefore, it is necessary to eliminate those photos with obvious defects such as blurriness, perspective distortion, incomplete subject, etc. In order to further improve the image quality and prepare for building a database, Photoshop software is used to perform background removal on the selected images. At the same time, in order to retain the main geometric features of the pattern, the outline of the pattern is manually outlined.

[0038] This series of processing steps ultimately forms a database of woodcarving decorative patterns. This database will provide rich materials and references for subsequent research, analysis, and potential applications. In this way, the artistic styles, technical details, historical backgrounds, etc. of woodcarving decorations can be studied and understood more systematically.

[0039] Step S12: Select a window pattern wireframe with a target style from the woodcarving decorative pattern database, and perform text description on the window pattern wireframe to generate a training dataset of text-image pairs.

[0040] Exemplarily, first, 1500 selected woodcarving window pattern wireframes are chosen from the woodcarving decorative pattern database. These images have various styles and details, but all need to conform to a specific target style.

[0041] Next, in order to construct a generative AI training dataset of "text-image", a construction method based on the AVE (encoder-decoder) structure is adopted. In this structure, the encoder is responsible for extracting the features of the image and generating a vector that can represent the content of the image. This vector contains all the important information of the image and is crucial for subsequent text generation.

[0042] Then, from these 1500 window pattern wireframes, 500 window pattern wireframe images with traditional Chinese characteristics are further selected. These images will serve as the core part of the training dataset.

[0043] To generate text descriptions corresponding to these images, a function of the visual language multimodal large model, namely the image description function, is adopted. This function can train the model to enable it to automatically generate accurate text descriptions for images. In this process, the decoder plays a key role. It takes the image content vector generated by the encoder as input and then gradually generates a text description corresponding to the image content.

[0044] Finally, these 500 window pattern wireframes and their corresponding text descriptions constitute the dataset of text-image pairs for training. This dataset can be used to train an AI model to enable it to recognize and understand the characteristics of window patterns and may further be used to generate new window pattern designs.

[0045] Step S13: Use the training dataset to fine-tune the model using the LoRA algorithm based on the pre-trained Stable Diffusion model to obtain the target Stable Diffusion model.

[0046] Step S14: Generate woodcarving patterns with the target style according to the target Stable Diffusion model.

[0047] In some embodiments, during the process of fine-tuning the model based on the open-source pre-trained Stable Diffusion model through LoRA technology, the generation of woodcarving patterns is achieved by using the reverse diffusion process from adding noise to denoising, specifically including: gradually adding noise conforming to the normal distribution to the patterns in the training dataset to generate noisy patterns; using a Markov chain to gradually remove the noise from the noisy patterns, where the estimated conditional probability is calculated according to Bayes' formula at each step.

[0048] Specifically, during the noise addition process, according to the basic principle of the diffusion model, the image at the next moment is generated by adding noise to the image at the previous moment . The noise addition process can be expressed as: ,

[0049] In the above formula, represents the image at time t, represents the image at the previous time t-1, represents the noise image with a random distribution generated at time t, represents a constant related to , usually expressed as , represents the constant specified at time t, which is used to control the intensity of the noise.

[0050] Therefore, the formula for the noise addition process can be rewritten as: ,

[0051] where ∼N(0,I), that is, it follows the standard normal distribution. In statistics and probability theory, the standard normal distribution is a special normal distribution with a mean of 0 and a standard deviation of 1. I represents the identity matrix, whose diagonal elements are all 1 and the remaining elements are all 0. The identity matrix is used to ensure that each component of is independent when adding noise, and the noise intensity of each component is the same (i.e., all follow the standard normal distribution). In the diffusion model, is used as the noise vector, which is combined with the image at the previous moment after weighting to generate the image at the next moment .

[0052] In the above description, the normal distribution formula (i.e., the Gaussian distribution formula) is a very important continuous probability distribution in probability theory and statistics. The formula for its probability density function is:

[0053] ,

[0054] ),

[0055] In the above formula, is the mean (usually set to 0), which determines the central position of the distribution, is the standard deviation, which determines the width or dispersion of the distribution, is the random variable. By adjusting value, the intensity of the noise can be controlled.

[0056] In the reverse diffusion stage of the denoising process, a series of Markov chains are used to gradually remove the predicted noise. For the estimation and removal of noise, the most commonly used is U-Net. Each step of the reverse diffusion process attempts to estimate the conditional probability distribution, and its reverse diffusion process is expressed as:

[0057] ,

[0058] It should be noted that in the reverse diffusion process, a neural network (such as U-Net) is usually used to simulate the conditional probability distribution of the reverse diffusion . This neural network will learn how to generate the data at the previous moment from the current noisy data and generate the data at the previous moment . During the training process, the model learns the parameters of the neural network by maximizing the likelihood estimation or minimizing the loss function, so that the reverse diffusion process can successfully restore the data. By repeating this process, the target stable diffusion model is trained. At the same time, the noise will be gradually removed, and a more "clean" wood carving pattern will be obtained.

[0059] Therefore, step S1 is a process of constructing and using wood carving window flower pictures to train a wood carving pattern generation model with a target style. First, a rich database of wood carving decoration patterns is established by collecting a large number of wood carving window flower pictures, which provides the basic materials for subsequent pattern recognition and learning. Then, window flower patterns that meet a specific design style are selected from this database, and these patterns are transformed into text descriptions, forming text-image pairs together with the corresponding images as the training dataset. The creation of this dataset helps the model understand the characteristics of different styles. Next, this training dataset is used to fine-tune a pre-trained stable diffusion model. Here, the LoRA algorithm is introduced, which is a lightweight parameter fine-tuning method. It can effectively adjust the performance of the large model to make it more adaptable to new task requirements, while avoiding the computational resource consumption caused by retraining the entire model. After fine-tuning with the LoRA algorithm, a stable diffusion model optimized for the target style is obtained. Finally, with the help of this fine-tuned model, wood carving patterns with the target artistic style can be generated. This series of operations not only improves the style consistency of the generated patterns, but also ensures the generation efficiency and saves computational costs. Therefore, the main technical effect of step S1 is to realize the automatic generation of wood carving patterns with personalized styles through data-driven and model optimization.

[0060] Step S2: Use intuitionistic fuzzy VIKOR to rank the generated woodcarving patterns according to aesthetic priorities and select multiple excellent samples.

[0061] It should be noted that intuitionistic fuzzy VIKOR introduces the intuitionistic fuzzy set theory into VIKOR to deal with multi-attribute decision-making problems where the attribute values are intuitionistic fuzzy numbers. The intuitionistic fuzzy set not only considers the membership degree of an element belonging to a certain set, but also considers the non-membership degree of the element not belonging to the set and the hesitation degree of the element.

[0062] Specifically, this step S2 includes the following steps S21 to S27.

[0063] Step S21: Screen the woodcarving patterns and conduct multi-dimensional evaluations to obtain evaluation index information and scores.

[0064] Exemplarily, first, 300 unique window lattice designs were created through the Stable Diffusion Model (SDM). Then, this large design library was carefully reviewed by multiple senior fashion designers, and 40 works that could meet the public's aesthetic taste were selected.

[0065] To comprehensively evaluate the artistic value of these 40 window lattices, an expert team was invited to jointly establish a series of evaluation criteria from multiple perspectives such as design innovation, production techniques, material selection, visual beauty, and cultural connotations. In this process, key adjectives that can reflect consumers' emotional responses, such as "simple", "modern", "exquisite", "curved", "rounded", etc., were selected as evaluation indicators.

[0066] To quantify these subjective evaluation indicators, a quantitative research method was adopted, and a detailed questionnaire was designed. For each window lattice, the experts scored according to the above five evaluation indicators, and the score range was from 1 to 5 points, where 1 point represents "very poor", 2 points represents "poor", 3 points represents "average", 4 points represents "good", and 5 points represents "very good".

[0067] Finally, these qualitative evaluations were further transformed into intuitive fuzzy values for subsequent data analysis and processing. The purpose of doing this is to ensure that all evaluations can accurately reflect the actual performance of the window lattices, and at the same time, it is convenient to compare and optimize different design options.

[0068] Step S22: Define the intuitionistic fuzzy set, which includes the satisfaction degree and dissatisfaction degree of the screened woodcarving patterns, and define the uncertainty degree according to the satisfaction degree and dissatisfaction degree.

[0069] Exemplarily, assume that X is a given set, then A= , where respectively represent the elements in X xMembership degree (satisfaction degree) and non-membership degree (dissatisfaction degree) belonging to A, and satisfying the condition 0 + , , hesitation degree (uncertainty degree) =1- - , A and B are any two intuitionistic fuzzy numbers on X. Qualitative indicators are often described by fuzzy language such as good, general, and poor to describe preference information. For the convenience of evaluation, this fuzzy language information needs to be converted into specific numerical values to analyze specific problems. The scores from 1 to 5 are described by the fuzzy language of very poor, poor, general, good, and very good respectively, and these qualitative indicators are converted into corresponding intuitionistic fuzzy numbers.

[0070] Step S23: Obtain the intuitionistic fuzzy entropy according to the satisfaction degree, dissatisfaction degree, and uncertainty degree.

[0071] Exemplarily, the score and language evaluation information are uniformly converted into the form of intuitionistic fuzzy numbers to obtain the intuitionistic fuzzy evaluation matrix of the expert :[[]]END]] , where = represents the intuitionistic fuzzy number.

[0072] Furthermore, the magnitude of the intuitionistic fuzzy entropy is used to reflect the fuzzy uncertainty degree of the expert's evaluation information on the index. The larger the intuitionistic fuzzy entropy, the higher the fuzzy uncertainty degree of the evaluation given by the expert, and a smaller weight should be assigned to the index. Conversely, a larger weight should be assigned. The intuitionistic fuzzy entropy of the expert 's evaluation of the index is:[[]]END]] .

[0073] Step S24: Determine the index weight using the minimum deviation optimization model according to the intuitionistic fuzzy entropy.

[0074] Calculate the fuzzy uncertainty degree of the expert's evaluation on the index, that is, the intuitionistic fuzzy entropy, and then use the minimum deviation optimization model to determine the comprehensive index weight. The index weight determined by the expert 's intuitionistic fuzzy evaluation matrix can be expressed as:[[]]END]] = , where the comprehensive index weight can be determined by establishing a minimum deviation optimization model;

[0075] Determine the reasonable interval of the weight , and the left and right endpoints of the interval respectively represent the k th expert's evaluation of the jThe minimum and maximum weights assigned to each index are used to reflect the degree of consistency of experts' opinions by the deviation between the maximum and minimum weights: = - ; In addition, check whether the left and right endpoints are singular weights to determine the final reasonable interval.

[0076] Step S25: Determine the expert weights according to the index weights of intuitionistic fuzzy entropy.

[0077] The professional scopes, cognitive levels, etc. of each expert are not the same, so it is necessary to reasonably allocate expert weights. The more accurate and reliable the evaluation information given by an expert, the more it indicates that the expert has a better understanding of the judgment object or more information, and a larger weight should be allocated. Conversely, a smaller weight should be allocated.

[0078] Specifically, first, determine the expert weights based on entropy theory and weighted intuitionistic fuzzy entropy:

[0079] = , represents the weight of the k th expert, represents the weighted intuitionistic fuzzy entropy of the k th expert, which is a quantitative representation of the degree of fuzziness. K represents the total number of experts, and n represents the number of evaluation dimensions.

[0080] In the formula = , represents the weight of the attribute j for the expert, represents the intuitionistic fuzzy entropy of the expert's evaluation information on the attribute j .

[0081] Step S26: Convert according to the expert weights to a group decision matrix to determine the positive ideal solution and negative ideal solution of each wood carving pattern under each index.

[0082] Specifically, first, use the intuitionistic fuzzy weighted average operator to aggregate each expert decision matrix into a group decision matrix, that is:

[0083] = =1- ， ,

[0084] In the above formula, represents the i in the group decision matrixThe element in the j th row and i th column represents the intuitionistic fuzzy evaluation of the j th scheme on the th attribute after synthesizing all expert opinions. k represents the intuitionistic fuzzy number of the i th expert on the j th scheme for the th attribute, which is represented by the membership degree and the non-membership degree represents the weighted average operation of intuitionistic fuzzy numbers, represents the product operation;

[0085] According to the group decision-making matrix, determine the positive ideal solution and negative ideal solution of each wood carving pattern under each index respectively:

[0086] = , = ,

[0087] In the formula, the positive ideal solution is the optimal value of the j th index among all schemes. If the j th index is a benefit-type index, then the positive ideal solution is the maximum value of this index among all schemes, = , if the j th index is a cost-type index, then the positive ideal solution is the minimum value of this index among all schemes, = ; the negative ideal solution is the worst value of the j th index among all schemes. If the j th index is a benefit-type index, then the negative ideal solution is the minimum value of this index among all schemes, = , if the j th index is a cost-type index, then the negative ideal solution is the maximum value of this index among all schemes, = , where j is a benefit-type index.

[0088] Step S27: Use the intuitionistic fuzzy VIKOR method to calculate the group utility value, individual regret value, and compromise evaluation value of each wood carving pattern based on the positive ideal solution and negative ideal solution, so as to rank the wood carving patterns and select multiple excellent samples.

[0089] In some embodiments, the wood carving pattern samples are ranked according to the group utility value, individual regret value, and compromise evaluation value; if the compromise evaluation value of a wood carving pattern is the minimum value and simultaneously satisfies the following two conditions, then the wood carving pattern is considered an optimal sample, where: Condition 1: , where and are respectively the minimum value and the second minimum value in the ranking of the compromise evaluation value, m is the number of samples of the wood carving pattern; Condition 2: The minimum values in both the group utility value ranking and the individual regret value ranking; if only Condition 1 is satisfied, then all wood carving patterns are close to the ideal solution; if only Condition 2 is satisfied, it means that and the corresponding window grilles are all compromise solutions.

[0090] Exemplarily, use VIKOR to calculate the group utility value S( ) of each sample, R( ) the individual regret value, Q( ) , that is:

[0091] S( )= ，R( )= ，Q( )= ) ,

[0092] where n represents the number of criteria, is the weight of the j th criterion, d is the Euclidean distance between any two intuitionistic fuzzy numbers, , respectively represent the best and worst performances under the j th criterion, represents the performance of the wood carving pattern under the j th criterion, and They are respectively the maximum and minimum values of the group utility values of all wood carving patterns.

[0093] Condition 1: , where and correspond to Q( ) the minimum and the second minimum values in the m value sorting, and

[0094] Condition 2: The optimal window lattice is S( ) the minimum value in both the R( ) value and

[0095] value sorting. If only Condition 1 is satisfied, it means that all samples are close to the ideal solution; if only Condition 2 is satisfied, it means that and the corresponding samples are all compromise solutions. According to the result of sorting the compromise values Q( ) from good to bad, the minimum value (such as 0) and the second value (such as 0.0374) of a certain serial number (for example, serial number 21) can be seen. According to Condition 1: , , m is 40, 1 / (m - 1) = 0.0256, 0.0374 > 0.0256, which satisfies Condition 1. It can be understood that, for example, the S( ) values of serial number 21 R ( ) are all the minimum values, that is, the 21st picture is the optimal solution. After removing the 21st picture, after calculation Q( ) the minimum and the second minimum values in the Q( ) value sorting only satisfy Condition 1, which means that all samples are close to the ideal solution. Because it is mentioned above that if the compromise evaluation value of a wood carving pattern is the minimum value and at the same time satisfies the following two conditions, then a wood carving pattern is considered an optimal sample. Therefore, the sample satisfies Q( )Sort the minimum values to select the four best window grilles. Finally, it is concluded that Nos. 21, 3, 12, 13, and 16 are the top five best solutions among all the window grilles, and then select the top three of them for subsequent research.

[0096] Step S3: Sketch and evaluate multiple excellent samples, select the best-designed product sketch for modeling, and obtain the modeled product with positioning design.

[0097] Specifically, this step S3 includes the following steps S31 to S33.

[0098] Step S31: Define the product positioning and determine the basic design framework.

[0099] Exemplarily, for product positioning, clarify the theme, style, and design goals. This design combines modern home aesthetics with traditional elements, with the theme of "classical elegance", applies the woodcarved window pattern to the atmosphere lamp to create a warm and elegant space, and brings new inspiration to modern home design. The core design concept is "inheritance and innovation", taking into account aesthetic and practicality, and the target consumers are traditional culture lovers, woodcarving art fans, and home decoration demanders.

[0100] Step S32: Incorporate multiple excellent samples into the product object corresponding to the defined product positioning and determined basic design framework, conduct sketch drawing and scheme evaluation, and select the best-designed product sketch.

[0101] Exemplarily, for the preliminary design sketch, collect user feedback and determine the scheme, select high-quality hardwood materials such as mahogany and boxwood, skillfully incorporate geometric window grille patterns into the lampshade, carefully adjust the proportion to ensure harmony and unity, draw the design sketch, invite 200 design major students, and evaluate the top three window grilles as three scheme sketches according to five emotional words (1 - 5 points) to obtain the Likert scale. The second scheme stands out with 4.19 points (the first and third schemes are 3.37 points and 3.28 points respectively).

[0102] Step S33: Model the best-designed product sketch to obtain the modeled product with positioning design.

[0103] Exemplarily, for prototype production, use 3D modeling software to construct the three-dimensional model of the scheme, and perform high-quality rendering through Keyshot software to lay a solid foundation for subsequent production.

[0104] Step S4: Verify the lighting parameters of the modeled product under different types of window grille decorations through simulation experiments to screen out the best product design.

[0105] It should be noted that TracePro software is based on the ACIS solid modeling kernel and is the first set of simulation software that combines a real solid model, powerful optical analysis capabilities, strong data conversion capabilities, and an easy-to-use interface. It combines Monte Carlo ray tracing, analysis, CAD import / export, and optimization methods with a complete and powerful macro language to solve various problems in lighting design and optical analysis.

[0106] Specifically, this step S4 includes the following steps S41 to S42.

[0107] Step S41: Configure the light source for the modeled product to obtain a ray tracing diagram and an irradiance analysis diagram to calculate the uniformity of light distribution.

[0108] Exemplarily, in the TracePro software, import the lamp model in stp format, set the light source, define parameters such as the position and luminous flux of the light source. Here, the light source is set to red light, uniformly emitting light from the central cylindrical surface, with a radiation power of 1W, the number of rays being 500,000, the material property being an opaque material, and the luminous area being the area of one window lattice pattern, 76 * 160 mm. After establishing the receiving surface, start the ray tracing simulation to obtain the ray tracing diagrams and irradiance analysis diagrams of the three window lattice patterns, and finally calculate the uniformity of light distribution in each scheme.

[0109] Step S42: Adjust the material of the modeled product according to the uniformity of light distribution to select the best product design.

[0110] Exemplarily, with other conditions unchanged, set the materials of the three window lattice patterns to a transparent material of PMMA (polymethyl methacrylate, with high transparency and a light transmittance of up to 90% - 92%) to obtain the ray tracing diagrams and illuminance analysis diagrams, and select the optimal window lattice pattern as the best product design by integrating all data.

[0111] Embodiment 2

[0112] Please refer to Figure 2 , which shows a schematic structural diagram of a traditional residential woodcarving window lattice pattern design system proposed in the second embodiment of this application. The system includes:

[0113] A pattern generation module 100, used to construct a woodcarving decoration pattern database, and use the stable diffusion model to learn and train in the woodcarving decoration pattern database to generate woodcarving patterns in a target style;

[0114] A pattern evaluation module 200, used to perform aesthetic priority ranking on the generated woodcarving patterns using intuitionistic fuzzy VIKOR to select multiple excellent samples;

[0115] The product design module 300 is used to draw sketches and evaluate the solutions for the multiple excellent samples, select the optimal design product sketch for modeling, and obtain the modeled product with positioning design.

[0116] The product verification module 400 is used to verify the lighting parameters of the modeled product under different types of window flower decorations through simulation experiments to screen out the best product design.

[0117] On the other hand, this application also proposes an electronic device, including: a processor; a memory for storing executable instructions that can be executed by the processor; wherein, the processor is configured to execute the executable instructions to implement the above-mentioned method for designing ancient residential woodcarved window flowers.

[0118] On the other hand, this application also proposes a computer-readable storage medium. When the instructions in the computer-readable storage medium are executed by the processor of an electronic device, the electronic device can execute the above-mentioned method for designing ancient residential woodcarved window flowers.

[0119] It should be noted that in this article, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without more limitations, the element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of this application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in a reverse order according to the functions involved. For example, the described methods may be executed in an order different from that described, and various steps may be added, omitted, or combined. Additionally, the features described with reference to certain examples may be combined in other examples.

[0120] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned example methods can be implemented by means of software plus a necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disc), and includes several instructions for causing a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in various embodiments of this application.

[0121] The embodiments of the present application have been described above in conjunction with the accompanying drawings. However, the present application is not limited to the above specific embodiments. The above specific embodiments are merely illustrative rather than restrictive. Under the inspiration of the present application, those of ordinary skill in the art can also make many forms without departing from the purpose of the present application and the scope protected by the claims, and all of them fall within the protection scope of the present application.

Claims

1. A method for designing wood-carved window grilles for ancient dwellings, characterized in that: The method comprises: Step S1: constructing a woodcarving decorative pattern database, and using a stable diffusion model to learn and train in the woodcarving decorative pattern database to generate a woodcarving pattern of a target style; Step S11: Collect a large number of woodcarving window grille pictures to build a woodcarving decorative pattern database; Step S12: selecting a window grille pattern wireframe image with a target style from the woodcarving decorative pattern database, and performing a text description on the window grille pattern wireframe image to generate a training data set of text-image pairs; Step S13: using the training data set, based on the pre-trained stable diffusion model, applying the LoRA algorithm to perform model fine-tuning to obtain a target stable diffusion model; Step S14: generating a wood carving pattern with a target style according to the target stable diffusion model; Step S2: using intuitive fuzzy VIKOR to sort the generated wood carving patterns by aesthetic priority and select multiple excellent samples; Step S21: screening the wood carving patterns and conducting multi-dimensional evaluation to obtain evaluation index information and scores; Step S22: defining an intuitive fuzzy set, which includes the satisfaction degree and the dissatisfaction degree of the screened wood carving pattern, and defining the uncertainty degree according to the satisfaction degree and the dissatisfaction degree; Step S23: obtaining an intuitive fuzzy entropy according to the satisfaction degree, the dissatisfaction degree and the uncertainty degree; Step S24: Determine the index weights using the minimum deviation optimization model according to the intuitive fuzzy entropy; Step S25: determining the expert weight according to the index weight of the intuitive fuzzy entropy; Step S26: converting the expert weights into a group decision matrix to determine the positive ideal solution and the negative ideal solution of each wood carving pattern under various indicators; Step S27: using the intuitive fuzzy VIKOR method to calculate the group utility value, personal regret value and compromise evaluation value of each of the wood carving patterns according to the positive ideal solution and the negative ideal solution, so as to sort the wood carving patterns and select a plurality of excellent samples; Step S3: Sketching and evaluating the schemes of the plurality of excellent samples, selecting the best design product sketch for modeling, and obtaining a modeling product with positioning design, wherein the step S3 comprises: clarifying the product positioning and determining the basic design framework; integrating the plurality of excellent samples into the product objects corresponding to the clarified product positioning and the determined basic design framework, sketching and evaluating the schemes, and selecting the best design product sketch; modeling the best design product sketch to obtain a modeling product with positioning design; Step S4: verify the lighting parameters of the modeled product under different types of window grille decorations through simulation experiments to screen out the best product design. Step S4 includes: configuring the light source for the modeled product to obtain a ray tracing diagram and an irradiance analysis diagram to calculate the uniformity of light distribution; and adjusting the material of the modeled product according to the uniformity of light distribution to select the best product design.

2. The method for designing woodcarving window grilles for ancient dwellings according to claim 1, characterized in that: The step S13 comprises: In the process of fine-tuning the model through LoRA technology based on the open source pre-trained stable diffusion model, the reverse diffusion process from noise addition to denoising is used to achieve the generation of the wood carving pattern, which specifically includes: gradually adding noise conforming to a normal distribution to the patterns in the training data set to generate a noisy pattern; The noise pattern is subjected to step-by-step noise removal using a Markov chain, wherein each step calculates an estimated conditional probability according to a Bayesian formula.

3. The method for designing woodcarving window grilles of ancient dwellings according to claim 1, characterized in that: The step S27 comprises: sorting the wood carving pattern samples according to the group utility value, the individual regret value and the compromise evaluation value; If the compromise evaluation value of a wood carving pattern is the minimum value and satisfies the following two conditions at the same time, the wood carving pattern is considered to be the optimal sample, where: Condition one: , where and are the minimum and second minimum values ​​in the compromise evaluation value sorting, respectively. m is the number of samples of wood carving patterns; Condition 2: Both the group utility value ranking and the individual regret value ranking are minimum values; If only condition 1 is met, all the wood carving patterns are close to the ideal solution; If only the second condition is met, the wood carving patterns are selected based on the compromise evaluation values ​​of all the wood carving patterns.

4. A design system for woodcarving window grilles of ancient dwellings, characterized in that: The system comprises: The pattern generation module is used to build a woodcarving decorative pattern database, and use a stable diffusion model to learn and train in the woodcarving decorative pattern database to generate a woodcarving pattern of a target style; specifically, the module includes: collecting a large number of woodcarving window grille pictures to build a woodcarving decorative pattern database; selecting a window grille pattern wireframe with a target style from the woodcarving decorative pattern database, and providing a text description for the window grille pattern wireframe to generate a training data set of text-image pairs; using the training data set, based on the pre-trained stable diffusion model, applying the LoRA algorithm to fine-tune the model to obtain a target stable diffusion model; generating a woodcarving pattern with a target style according to the target stable diffusion model; The pattern evaluation module is used to use intuitive fuzzy VIKOR to sort the generated wood carving patterns by aesthetic priority and select multiple excellent samples; specifically, it includes: screening and multi-dimensional evaluation of the wood carving patterns to obtain evaluation index information and scores; defining an intuitive fuzzy set, which includes the satisfaction and dissatisfaction of the screened wood carving patterns, and defining the uncertainty according to the satisfaction and dissatisfaction; obtaining intuitive fuzzy entropy according to the satisfaction, dissatisfaction and uncertainty; determining the index weight according to the intuitive fuzzy entropy using a minimum deviation optimization model; determining the expert weight according to the index weight of the intuitive fuzzy entropy; converting the expert weight into a group decision matrix to determine the positive ideal solution and the negative ideal solution of each wood carving pattern under various indicators; using the intuitive fuzzy VIKOR method to calculate the group utility value, personal regret value and compromise evaluation value of each wood carving pattern according to the positive ideal solution and the negative ideal solution, so as to sort the wood carving patterns and select multiple excellent samples; The product design module is used to draw sketches and evaluate the schemes of the multiple excellent samples, select the best design product sketch for modeling, and obtain a modeled product with positioning design, specifically including: clarifying the product positioning and determining the basic design framework; integrating the multiple excellent samples into the product objects corresponding to the clear product positioning and the determined basic design framework, drawing sketches and evaluating the schemes, and selecting the best design product sketch; modeling the best design product sketch to obtain a modeled product with positioning design; The product verification module is used to verify the lighting parameters of the modeled product under different types of window grille decorations through simulation experiments to screen out the best product design, specifically including: configuring the light source for the modeled product to obtain a ray tracing diagram and an irradiance analysis diagram to calculate the uniformity of light distribution; and adjusting the material of the modeled product according to the uniformity of light distribution to select the best product design.

5. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; Wherein, the processor is configured to execute the executable instructions to implement a method for designing wood-carved window grilles for ancient dwellings as described in any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that: When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the method for designing wood-carved window grilles for ancient dwellings as described in any one of claims 1 to 3.

Images