Automatic patchwork pattern generation method based on intelligent algorithm

Abstract

The invention relates to the technical field of data processing, and discloses a method for automatically generating a patchwork pattern based on an intelligent algorithm, and the method comprises the following steps: S1, user demand analysis: collecting and analyzing design demands, including color preference, style requirements, pattern complexity and other parameters, input by a user; s2, preparing a patchwork element data constructing the patchwork element database, wherein the patchwork element database comprises geometrical shapes, textures, color matching schemes and historical design cases; s3, generating a patchwork pattern: adopting a GAN deep learning model to generate a candidate patchwork pattern; s4, optimizing the pattern of the patchwork: optimizing and generating the pattern in combination with reinforcement learning; s5, performing image segmentation and optimization algorithm: through the image segmentation and optimization algorithm, ensuring seamless connection when the cloth splicing blocks are spliced, and adapting to actual cloth processing requirements; s6, editing a patchwork design scheme: generating a patchwork design scheme in an editable format for a user to preview and modify; and S7, outputting the patchwork pattern: outputting the final patchwork pattern.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to an automatic generation method of patchwork patterns based on intelligent algorithms. Background Art

[0002] Patchwork art is a technique that combines multiple fabric pieces for splicing and combination, and is widely used in fields such as clothing, home textiles, and art decoration. Traditional patchwork designs mainly rely on manual work, including pattern conception, color matching, fabric block cutting, and splicing techniques. Designers often need to have rich experience and strong aesthetic qualities. However, manual design has the following problems:

[0003] Long design cycle: The design of patchwork patterns requires multiple adjustments and optimizations, and the manual design process takes a long time, making it difficult to meet the requirements of rapid customization.

[0004] Limited innovation: Manual design mainly relies on designers' experience and inspiration, and it is difficult to automatically generate diverse patchwork patterns, resulting in similar design styles.

[0005] Difficult to optimize: Manual design is difficult to balance the coordination of patchwork block shapes, color matching, and fabric cutting, easily causing fabric waste or splicing errors.

[0006] Difficult to customize individually: It is difficult to quickly analyze and convert users' individual needs into feasible patchwork design schemes.

[0007] In recent years, with the development of artificial intelligence (AI) and computer vision technologies, intelligent design methods have gradually been applied to the field of pattern generation. For example, deep learning models such as generative adversarial networks (GANs) and style transfer have achieved remarkable results in fields such as art creation and image processing. At the same time, reinforcement learning can be used for pattern optimization to meet specific design requirements. In addition, image segmentation and optimization algorithms can improve the boundary processing effect of patchwork patterns and ensure the accuracy of fabric cutting.

[0008] The present invention proposes an automatic generation method of patchwork patterns based on intelligent algorithms. By combining the GAN deep learning model, reinforcement learning optimization strategy, and image segmentation algorithm, it realizes the intelligent generation, optimization, and output of patchwork patterns. This method can not only improve design efficiency, but also meet individual needs and optimize the patchwork layout scheme, thereby enhancing the overall quality and feasibility of the patchwork technique. Summary of the Invention

[0009] The purpose of the present invention is to solve the drawbacks existing in the prior art and provide an automatic generation method of patchwork patterns based on intelligent algorithms.

[0010] To achieve the above object, the present invention provides the following technical solutions:

[0011] An automatic quilting pattern generation method based on intelligent algorithms, comprising the following steps:

[0012] S1. User requirement analysis: Collect and analyze the design requirements input by the user, including parameters such as color preference, style requirement, pattern complexity, etc.;

[0013] S2. Quilting element database: Construct a quilting element database, including geometric shapes, textures, color schemes, and historical design cases;

[0014] S3. Quilting pattern generation: Use a GAN deep learning model to generate candidate quilting patterns;

[0015] S4. Quilting pattern optimization: Combine reinforcement learning to optimize the generated pattern, adjust the model parameters based on a scoring system, and make the output result meet aesthetic standards;

[0016] S5. Image segmentation and optimization algorithm: Through image segmentation and optimization algorithms, ensure seamless connection when quilting blocks are spliced and adapt to actual fabric processing requirements;

[0017] S6. Editable quilting design scheme: Generate an editable format of the quilting design scheme for the user to preview and modify;

[0018] S7. Quilting pattern output: Output the final quilting pattern and convert it into a cutting template for fabric cutting and sewing.

[0019] The user requirement analysis S1 includes the following sub-steps:

[0020] S11. Collect the user's input quilting design preferences through the GUI interface or voice interaction method;

[0021] S12. Use natural language processing (NLP) technology to parse the text description and extract style, theme, and color preferences;

[0022] S13. Use a color extraction algorithm (such as K-means clustering) to analyze the reference image uploaded by the user and determine the main color and color scheme;

[0023] S14. Adopt user portrait analysis, combine historical data to predict the quilting styles that the user may like, and improve the recommendation effect.

[0024] As a preferred technical solution of the present utility model, the quilting element database S2 includes the following data types:

[0025] S21. Common geometric shapes, such as triangles, rectangles, hexagons, etc.;

[0026] S22. Traditional patchwork patterns, such as nine-square grid, windmill, star, concentric circles, etc.;

[0027] S23. Modern style elements, such as abstract patterns, irregular splicing, etc.;

[0028] S24. Color matching templates, including color matching schemes such as complementary colors, analogous colors, and contrast colors;

[0029] S25. Texture database, containing simulated texture information of different fabric materials (such as cotton, silk, wool);

[0030] S26. Classic patchwork works cases, used for style transfer or similarity matching recommendations.

[0031] The patchwork pattern generation S3 uses a generative adversarial network (GAN) to generate patchwork patterns, specifically including the following steps:

[0032] S31. Train the generation model so that it can learn patchwork design patterns of different styles based on the patchwork element database;

[0033] S32. Use a generative adversarial network (GAN) or an attention mechanism to improve the adaptability of the model to user input conditions; Assume that the generator is G and the discriminator is D. The goal of the generator is to maximize the probability of misjudgment by the discriminator, while the goal of the discriminator is to correctly distinguish real images from generated images as much as possible. The loss function is: L GAN =E x～pdata(x) [logD(x)] + E z～pz(z) [log(1 - D(G(z)))]

[0034] Among them, x is the real data, z is the input noise, G(z) is the generated image, and D(x) is the probability output by the discriminator.

[0035] S33. Adjust the generation result through style transfer technology to make it conform to a specific artistic style;

[0036] S34. Combine an autoencoder for feature extraction to achieve efficient pattern synthesis.

[0037] As a preferred technical solution of the present utility model, the patchwork pattern optimization S4 uses a reinforcement learning algorithm to optimize the patchwork pattern generation process, specifically including the following steps:

[0038] S41. Set the objective function, including optimization objectives such as minimizing the number of patchwork blocks and maximizing color coordination;

[0039] S42. Train an agent model using Deep Q - learning so that it can iteratively optimize the patchwork pattern; Q - learning algorithm formula: The update formula for the state - action value function Q(s, a) is:

[0040] where s is the state, a is the action, r is the reward, γ is the discount factor, and α is the learning rate.

[0041] S43. Combine Monte Carlo Search Tree (MCTS) to evaluate the scores of different patchwork schemes and screen the optimal scheme;

[0042] S44. Adjust the reward function through user feedback to improve the aesthetics and sewability of the generated pattern.

[0043] The image segmentation and optimization algorithm S5 includes the following:

[0044] S51. Use image segmentation algorithms such as GrabCut, U - Net, or Mask R - CNN to extract the outlines of patchwork blocks;

[0045] S52. Calculate the edge matching degree of patchwork blocks to avoid obvious seams during splicing;

[0046] S53. Combine morphological operations (such as erosion, dilation) to optimize the pattern boundary and improve the processability of patchwork blocks;

[0047] S54. Adjust the transition between patchwork blocks through Gaussian blur or edge sharpening to make the overall design more natural.

[0048] As a preferred technical solution of the present utility model, the patchwork pattern can be converted into an editable format and supports the following output methods:

[0049] (1) Vector formats (SVG, DXF), which are convenient for subsequent processing and scaling;

[0050] (2) Bitmap formats (PNG, JPEG), for users to directly print and preview;

[0051] (3) CAD formats (DWG, DXF), for direct reading by a numerical control cutting machine;

[0052] (4) 3D simulations (OBJ, STL), for three - dimensional display of fabrics and virtual trial assembly.

[0053] The patchwork pattern can be converted into an actual sewing scheme through a cutting template. The specific steps include:

[0054] (1) Automatically calculate the sizes of patchwork blocks and sewing boundaries to generate a cut - out template;

[0055] (2) Combine intelligent layout algorithms (such as stripe optimization algorithms) to reduce fabric waste;

[0056] (3) Adopt a marking algorithm to generate numbers and sewing sequences for each patchwork;

[0057] (4) Generate step-by-step sewing guides to instruct users to piece together the fabric blocks in the best order.

[0058] Compared with the prior art, an automatic patchwork pattern generation method based on intelligent algorithms provided by the present invention has the following beneficial effects:

[0059] (1) Improve design efficiency

[0060] Adopt a generative adversarial network (GAN) to automatically generate patchwork patterns, significantly reducing the manual design time. At the same time, combined with user demand analysis, personalized design recommendations are realized, enhancing the design efficiency;

[0061] (2) Enhance creativity and diversity

[0062] Through the GAN deep learning model and style transfer technology (Style Transfer), patchwork patterns of different styles can be generated, covering various artistic styles such as traditional, modern, and abstract, breaking through the limitations of manual design, and improving the innovation and diversity of the patterns;

[0063] (3) Optimize the quality of patchwork

[0064] Combine reinforcement learning to optimize patchwork patterns, and use deep Q-learning (Deep Q-learning) and Monte Carlo search tree (MCTS) to adjust pattern design, making the shapes and color combinations of patchwork blocks more coordinated, enhancing the overall aesthetics and sewability.

[0065] (4) Reduce fabric waste

[0066] Through intelligent layout algorithms (such as stripe optimization algorithms), automatically calculate the best layout of patchwork blocks, minimize fabric loss to the greatest extent, improve material utilization rate, and reduce production costs;

[0067] (5) Improve the accuracy of patchwork

[0068] Adopt image segmentation and optimization algorithms (such as GrabCut, U-Net, Mask R-CNN) to accurately extract the outlines of patchwork blocks, optimize boundary processing, ensure seamless splicing between patchwork blocks, and enhance the accuracy and operability of actual sewing;

[0069] (6) Support multiple format outputs

[0070] The generated patchwork patterns can be output in vector formats (SVG, DXF), bitmap formats (PNG, JPEG), CAD formats (DWG, DXF), and 3D simulation formats (OBJ, STL), which are suitable for different processing methods, such as CNC cutting, hand sewing, virtual trial patching, etc., to meet diverse needs. Description of the Drawings

[0071] Figure 1 It is a schematic flowchart of the process structure of an automatic patchwork pattern generation method based on an intelligent algorithm according to the present invention. Detailed Embodiments

[0072] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0073] In order to enable those skilled in the art to better understand the technical solutions of the present invention, the following will further introduce the present invention in detail in conjunction with the drawings.

[0074] As Figure 1 shown, an automatic patchwork pattern generation method based on an intelligent algorithm provided by an embodiment of the present invention includes the following steps:

[0075] S1. User requirement analysis: Collect and analyze the design requirements input by the user, including parameters such as color preference, style requirement, pattern complexity, etc.;

[0076] S2. Patchwork element database: Build a patchwork element database, including geometric shapes, textures, color matching schemes, and historical design cases;

[0077] S3. Patchwork pattern generation: Use a GAN deep learning model to generate candidate patchwork patterns;

[0078] S4. Patchwork pattern optimization: Combine reinforcement learning to optimize the generated pattern, adjust the model parameters based on a scoring system, and make the output result meet the aesthetic standards;

[0079] S5. Image segmentation and optimization algorithm: Through the image segmentation and optimization algorithm, ensure seamless connection when patching blocks and adapt to the actual fabric processing requirements;

[0080] S6. Editable patchwork design scheme: Generate an editable patchwork design scheme for the user to preview and modify;

[0081] S7. Patchwork pattern output: Output the final patchwork pattern and convert it into a cutting template for fabric cutting and sewing.

[0082] The user requirement analysis S1 includes the following sub-steps:

[0083] S11. Collect the quilt design preferences input by the user through the GUI interface or voice interaction;

[0084] S12. Use natural language processing (NLP) technology to parse the text description and extract style, theme, and color preferences;

[0085] S13. Use a color extraction algorithm (such as K-means clustering) to analyze the reference images uploaded by the user and determine the main color tone and color matching scheme;

[0086] S14. Adopt user portrait analysis and combine historical data to predict the quilt styles that the user may like to improve the recommendation effect.

[0087] An embodiment of the present invention is that the quilt element database S2 includes the following data types:

[0088] S21. Common geometric shapes, such as triangles, rectangles, hexagons, etc.;

[0089] S22. Traditional quilt patterns, such as nine-square grids, windmills, stars, concentric circles, etc.;

[0090] S23. Modern style elements, such as abstract patterns, irregular splicing, etc.;

[0091] S24. Color matching templates, including color matching schemes such as complementary colors, analogous colors, and contrast colors;

[0092] S25. Texture database, containing simulation texture information of different fabric materials (such as cotton, silk, wool);

[0093] S26. Classic quilt work cases for style transfer or similarity matching recommendations.

[0094] The quilt pattern generation S3 uses a generative adversarial network (GAN) to generate quilt patterns, specifically including the following steps:

[0095] S31. Train the generation model so that it can learn different style quilt design patterns based on the quilt element database;

[0096] S32. Use a generative adversarial network (GAN) or an attention mechanism to improve the adaptability of the model to the user input conditions; assuming that the generator is G and the discriminator is D, the goal of the generator is to maximize the probability of misjudgment by the discriminator, while the goal of the discriminator is to correctly distinguish between real images and generated images as much as possible. The loss function is: L GAN =E x～pdata(x) [logD(x)]+Ez～pz(z) [log(1 - D( G (z)))]

[0097] where x is the real data, z is the input noise, G (z) is the generated image, D (x) is the probability output by the discriminator.

[0098] S33. Adjust the generation result through the Style Transfer technique to make it conform to a specific artistic style;

[0099] S34. Combine the Autoencoder for feature extraction to achieve efficient pattern synthesis.

[0100] Another embodiment of the present invention is that the patchwork pattern optimization S4 uses a reinforcement learning algorithm to optimize the patchwork pattern generation process, specifically including the following steps:

[0101] S41. Set the objective function, including optimization objectives such as minimizing the number of patchwork pieces and maximizing the color coordination;

[0102] S42. Use Deep Q - learning to train the agent model so that it can iteratively optimize the patchwork pattern; The formula of the Q - learning algorithm: The update formula of the state - action value function Q(s,a) is:

[0103] where s is the state, a is the action, r is the reward, γ is the discount factor, and α is the learning rate.

[0104] S43. Combine the Monte Carlo Search Tree (MCTS) to evaluate the scores of different patchwork schemes and select the optimal scheme;

[0105] S44. Adjust the reward function through user feedback to improve the aesthetics and sewability of the generated pattern.

[0106] The image segmentation and optimization algorithm S5 includes the following:

[0107] S51. Use image segmentation algorithms such as GrabCut, U - Net or Mask R - CNN to extract the outlines of patchwork pieces;

[0108] S52. Calculate the edge matching degree of patchwork pieces to avoid obvious seams during splicing;

[0109] S53. Combine morphological operations (such as erosion, dilation) to optimize the pattern boundary and improve the processability of patchwork pieces;

[0110] S54. Adjust the transition between the quilt blocks through Gaussian blur or edge sharpening to make the overall design more natural.

[0111] The quilt pattern can be converted into an editable format and supports the following output methods:

[0112] (1) Vector formats (SVG, DXF), which are convenient for subsequent processing and resizing.

[0113] (2) Bitmap formats (PNG, JPEG), for users to directly print and preview.

[0114] (3) CAD formats (DWG, DXF), for direct reading by CNC cutting machines.

[0115] (4) 3D simulations (OBJ, STL), for three-dimensional display of fabrics and virtual trial assembly.

[0116] The quilt pattern can be converted into an actual sewing plan through a cutting template. The specific steps include:

[0117] (1) Automatically calculate the sizes of the quilt blocks and the sewing boundaries to generate a cutting template.

[0118] (2) Combine intelligent layout algorithms (such as the strip optimization algorithm) to reduce fabric waste.

[0119] (3) Adopt a marking algorithm to generate numbers and sewing sequences for each quilt block.

[0120] (4) Generate a step-by-step sewing guide to instruct users to piece the blocks in the best order.

[0121] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An automatic generation method of patchwork patterns based on intelligent algorithms, characterized in that It includes the following steps: S1. User requirement analysis: Collect and analyze the design requirements input by the user, including parameters such as color preference, style requirement, pattern complexity, etc.; S2. Quilting element database: Build a quilting element database, including geometric shapes, textures, color schemes, and historical design cases; S3. Quilting pattern generation: Use a GAN deep learning model to generate candidate quilting patterns; S4. Quilting pattern optimization: Combine reinforcement learning to optimize the generated pattern, adjust the model parameters based on a scoring system to make the output result meet the aesthetic standards; S5. Image segmentation and optimization algorithm: Through image segmentation and optimization algorithm, ensure seamless connection when quilting blocks are joined and adapt to the actual fabric processing requirements; S6. Editable quilting design scheme: Generate an editable format of the quilting design scheme for the user to preview and modify; S7. Quilting pattern output: Output the final quilting pattern and convert it into a cutting template for fabric cutting and sewing.

2. The automatic generation method of the patchwork pattern based on the intelligent algorithm according to claim 1, wherein, The user requirement analysis S1 includes the following sub-steps: S11. Collect the user's quilting design preferences input through the GUI interface or voice interaction; S12. Use natural language processing (NLP) technology to parse the text description and extract style, theme, and color preferences; S13. Use a color extraction algorithm (such as K-means clustering) to analyze the reference image uploaded by the user to determine the main color and color scheme; S14. Adopt user portrait analysis, combine historical data to predict the quilting styles that the user may like, and improve the recommendation effect.

3. The automatic generation method of patchwork patterns based on intelligent algorithms according to claim 1, characterized in that, The quilting element database S2 includes the following data types: S21. Common geometric shapes, such as triangles, rectangles, hexagons, etc.; S22. Traditional quilting patterns, such as nine-square grids, windmills, stars, concentric circles, etc.; S23. Modern style elements, such as abstract patterns, irregular splicing, etc.; S24. Color matching templates, including complementary colors, analogous colors, contrast colors, etc. color schemes; S25. Texture database, including simulation texture information of different fabric materials (such as cotton, silk, wool); S26. Classic quilting work cases for style transfer or similarity matching recommendations.

4. A method for automatically generating a patchwork pattern based on an intelligent algorithm according to claim 1, characterized in that, The quilting pattern generation S3 uses a generative adversarial network (GAN) to generate quilting patterns, specifically including the following steps: S31. Train the generative model so that it can learn different styles of quilting design patterns based on the quilting element database; S32. Use a generative adversarial network (GAN) or an attention mechanism to improve the adaptability of the model to user input conditions; Assume that the generator is G and the discriminator is D. The goal of the generator is to maximize the probability of misjudgment by the discriminator, while the goal of the discriminator is to correctly distinguish between real images and generated images as much as possible. The loss function is: Among them, x is the real data, z is the input noise, G(z) is the generated image, and D(x) is the probability output by the discriminator. S33. Adjust the generated result through style transfer technology to make it conform to a specific artistic style; S34. Combine an autoencoder for feature extraction to achieve efficient pattern synthesis.

5. The automatic generation method of patchwork patterns based on intelligent algorithms according to claim 1, characterized in that The quilting pattern optimization S4 uses a reinforcement learning algorithm to optimize the quilting pattern generation process, specifically including the following steps: S41. Set the objective function, including optimization objectives such as minimizing the number of quilting blocks and maximizing the color coordination degree; S42. Train an agent model using Deep Q-learning so that it can iteratively optimize the patchwork pattern; Q-learning algorithm formula: The update formula for the state-action value function Q(s,a) is: Among them, s is the state, a is the action, r is the reward, γ is the discount factor, and α is the learning rate. S43. Evaluate the scores of different quilting patterns by combining Monte Carlo Tree Search (MCTS) and screen out the optimal pattern. S44. Adjust the reward function based on user feedback to improve the aesthetics and sewability of the generated patterns.

6. The automatic generation method of the patchwork pattern based on the intelligent algorithm according to claim 1, characterized in that The image segmentation and optimization algorithm S5 includes the following: S51. Use image segmentation algorithms such as GrabCut, U-Net or Mask R-CNN to extract the outlines of the quilting blocks. S52. Calculate the edge matching degree of the quilting blocks to avoid obvious seams during splicing. S53. Combine morphological operations (such as erosion and dilation) to optimize the pattern boundaries and improve the processability of the quilting blocks. S54. Adjust the transition between the quilting blocks through Gaussian blur or edge sharpening to make the overall design more natural.

7. A method for automatically generating a patchwork pattern based on an intelligent algorithm according to claim 1, characterized in that, The quilting pattern can be converted into an editable format and supports the following output methods: (1) Vector formats (SVG, DXF), which are convenient for subsequent processing and scaling. (2) Bitmap formats (PNG, JPEG), for users to directly print and preview. (3) CAD formats (DWG, DXF), for direct reading by CNC cutting machines. (4) 3D simulations (OBJ, STL), for three-dimensional display of fabrics and virtual trial quilting.

8. The automatic generation method of patchwork patterns based on intelligent algorithms according to claim 1, characterized in that The quilting pattern can be converted into an actual sewing plan through a cutting template. The specific steps include: (1) Automatically calculate the sizes of the quilting blocks and the sewing boundaries to generate a cuttable template. (2) Combine intelligent layout algorithms (such as strip optimization algorithms) to reduce fabric waste. (3) Use a marking algorithm to generate numbers and sewing orders for each quilting block. (4) Generate a step-by-step sewing guide to instruct users to splice the fabric blocks in the best order.

Images