Method and system for manufacturing AI art refrigerator sticker

Abstract

The invention relates to an intelligent AR sticker manufacturing system and method, and belongs to the technical field of augmented reality. The system realizes user registration and auditing and PNG background image configuration of a sticker template through the management module; the AR function module sets a white edge parameter 15 pixel based on the object identification unit to realize accurate positioning; the PC beautifying module adopts a bilateral filtering algorithm to dynamically adjust and beautify in real time according to a spatial domain standard deviation sigma d = 15 pixels and a color domain standard deviation sigma r = 0.3 * sv, and the frame rate under CUDA acceleration is greater than or equal to 30fps; the printing unit controls the instruction format of the die-cutting machine to be instruction head + length + data + CRC32 check code through a CRC32 check protocol, and the cutting error is smaller than or equal to 1 mm; and the memory module supports the two-dimensional code to share the electronic sticker. The system achieves full-process automation, the sticker registration error is smaller than or equal to 0.1 pixel, and the dynamic tracking displacement error is smaller than or equal to 1 cm. According to the invention, integrated control of AR sticker creation, facial beautification parameter optimization and printing calibration is realized through the centralized management module, the problems of AR special effect and printing output color distortion are effectively solved, and efficient operation of multi-device cooperation in a commercial scene is supported.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of smart home decoration, and particularly relates to a refrigerator sticker manufacturing method and system integrating AI artistic generation, personalized material configuration and magnetic manufacturing. BACKGROUND

[0002] The current refrigerator sticker manufacturing method mainly relies on manual design and traditional materials, and has problems of low design efficiency, single material and poor user interaction. Therefore, an automatic, personalized and interactive refrigerator sticker production system is needed to meet the diversified needs of modern users. SUMMARY

[0003] In view of the deficiencies in the prior art, the application provides a refrigerator sticker manufacturing method and system based on artificial intelligence artistic generation and multi-material adaptation. The system integrates user management, AI image processing, magnetic manufacturing and social communication modules, and can realize automatic control of the whole process from design, manufacturing to delivery, thereby improving the production efficiency and user interaction experience of the system.

[0004] The application is implemented as follows: a method and system for manufacturing intelligent AR stickers, the AR-based sticker processing system comprising:

[0005] A management module: including a user management unit, a material library unit and a manufacturing management unit; used as the core of the system, to uniformly supervise user account review, material library configuration and manufacturing process control, and to ensure the collaborative operation of the whole process from design to delivery;

[0006] An AI artistic generation module: including a style transfer unit, a text special effect unit and a preview editing unit; used to instantly convert user-uploaded pictures or text into various artistic style effects (such as oil painting and ink painting), to realize automatic design of personalized refrigerator stickers;

[0007] A magnetic manufacturing module: including a 3D printing unit and a magnetic pressing unit; used to 3D print a base material and hot-press a neodymium iron boron magnetic sheet, to manufacture refrigerator sticker finished products with strong adsorption force;

[0008] A social interaction module: including a work recommendation unit and a UGC community unit; used to provide a work sharing and community interaction platform, to stimulate user creativity through UGC content sedimentation and intelligent recommendation.

[0009] Further, the management module specifically includes:

[0010] A user management unit: including a mobile phone number verification interface, an account state machine and a unique ID generator, to realize a registration / login process such as a front-end form + back-end review, to activate an account and assign a "USER_DATE_RANDOM_CODE" format ID through an administrator review panel, to ensure controllable system permissions;

[0011] Material Library Unit: Integrates a magnetic material database, surface process configurator, and user collection module. Administrators can add magnetic substrate parameters such as soft magnet thickness of 0.5mm and surface treatment process relief accuracy of ±0.1mm in the backend. The user end provides material template preview and collection functions to support personalized material selection.

[0012] Manufacturing Management Unit: Composed of a 3D printing console for adjusting layer thickness / temperature / fill rate and a magnetic pressing protocol engine, it drives the printer to perform 0.1mm high-precision layer manufacturing via G-code, and links the hot pressing equipment to complete the magnetic sheet embedding at preset parameters of 180℃ / 30s, realizing fully automated production monitoring.

[0013] Furthermore, the AI ​​art generation module specifically includes:

[0014] Style transfer unit: The AdaIN algorithm is used to perform real-time artistic processing on user-uploaded images, such as oil painting / ink painting conversion, and style intensity control is achieved by adjusting feature distribution;

[0015] Text effects unit: Based on the CycleGAN model, the input text (ancient poems / custom content) is transformed into calligraphy fonts, and the brushstroke texture and layout optimization are integrated;

[0016] The preview and editing unit provides a 3D interactive interface, supports dragging and dropping to adjust the position and size of design elements, and generates 3D rendered images with materials and lighting in real time. These three units work together to provide end-to-end design support from content input to visual customization.

[0017] Furthermore, the magnetic manufacturing module specifically includes:

[0018] 3D printing unit: Using FDM fused deposition modeling technology, PLA or ABS materials are used to precisely print the refrigerator magnet substrate layer with a thickness accuracy of ±0.05mm, thus constructing the solid structure of the patch;

[0019] Magnetic pressing unit: Using a hot pressing process at 180±5℃, neodymium iron boron magnetic sheets with a magnetic field strength ≥1200 Gauss are embedded into the back of the substrate, realizing the physical integration of magnetic function and substrate, and finally forming a finished refrigerator magnet with artistic appearance and strong adsorption.

[0020] Furthermore, the social interaction module specifically includes:

[0021] The recommended design section analyzes users' historical design preferences, such as frequently used art styles and materials, and uses a collaborative filtering algorithm to match popular design templates and push them to the user interface.

[0022] UGC Community Unit: Construct an "Inspiration Wall" interactive platform that allows users to publish finished refrigerator magnets with 3D renderings and production parameters, receive real-time likes / comments, and trigger in-site message notifications, forming a closed-loop ecosystem of creation, sharing, and interaction.

[0023] In summary, the advantages and positive effects of this invention are as follows:

[0024] This invention provides an intelligent refrigerator magnet manufacturing system covering the entire process from idea generation to finished product output. The system integrates an AI art style transfer module based on the AdaIN algorithm, capable of image style processing within one second. Simultaneously, combined with optimized manufacturing processes using magnetic sheet distribution, it achieves an adsorption strength of ≥1200 Gauss. The system supports personalized designs in over ten art styles, achieving a manufacturing yield rate of 99.2%. Furthermore, through a user-generated content (UGC) community module, it facilitates the sharing and reuse of design content, with a user-original work reuse rate of up to 35% and a conversion rate for popular designs increasing by approximately 50%. This invention significantly improves personalized design efficiency, manufacturing precision, and user interaction.

[0025] This invention constructs a complete intelligent refrigerator magnet creation ecosystem through deep collaboration of management, AI art generation, magnetic manufacturing, and social interaction modules. Its core breakthrough lies in integrating the discrete stages of traditional refrigerator magnet production—design, production, output, and sharing—into a fully automated process, solving three major industry pain points: high creative threshold, weak material compatibility, and poor user interaction. The system adopts a modular design to achieve a closed-loop technology: after user registration and administrator approval to generate a unique ID account, users can configure material templates on the management side—selecting silicone / soft magnet substrates through the material library unit, setting the embossing layer depth accuracy to ±0.05mm, and UV coating process; after users upload images / text, the AI ​​art generation module, based on the AdaIN algorithm, renders artistic effects in real time with a processing speed of <1s, dynamically optimizing the output using a style intensity slider within the 0-1 range; on the manufacturing side, a magnetic sheet distribution algorithm (adsorption force formula) is used... The neodymium iron boron magnetic sheets are automatically arranged and precisely formed by layered temperature-controlled hot pressing at 180℃±5℃. The final product generates an electronic work card through the UGC community unit. Users can scan the code to share it to the "Inspiration Wall" and receive interactive feedback, forming a complete value chain from creativity to social interaction.

[0026] This invention relies on a precise mapping mechanism between AI generation and physical manufacturing to construct a seamless transformation chain from digital design to physical products. Its core technology lies in establishing a geometrical fit between virtual art creation and physical substrates. The specific implementation path and technical support are as follows:

[0027] User interaction and AI design phase 1. 3D modeling technology Implementation path: Real-time 3D reconstruction using Kinect Fusion, fusing depth data via TSDF (Truncate Symbolic Distance Function): Where v: three-dimensional space voxel coordinates (unit: mm); D i (v): Distance value measured by the depth sensor in the i-th frame (unit: mm); d i : The true distance from voxel v to the sensor plane (unit: mm); ò: Truncation distance threshold (empirical value 50 mm, used to filter noise); TSDF(v): Truncation sign distance function value, characterizing the surface internal and external state (dimensionless). Point cloud registration is performed using the ICP algorithm. The output is an STL mesh model with an error ≤0.1mm. 2. Artistic processing techniques Core support: Style transfer AdaIN algorithm Where c: feature tensor of the content image (dimensions C×H×W); s: feature tensor of the style image (dimensions C×H×W); μ(·): feature mean (calculated along the spatial dimension); σ(·): feature standard deviation (calculated along the spatial dimension); α: style intensity adjustment coefficient (preserves the original image, fully stylizes it). Interactive control: Style intensity coefficient α∈[0,1] Default value α=0.5, supports dynamic adjustment of output in 0.01 step size.

[0028] Manufacturing instruction compilation stage 1. Substrate slicing optimization Technical basis: Adaptive layering: adopts the Cura engine open source algorithm, and the layer thickness is dynamically adjusted according to the curvature (0.05-0.3mm); Support structure generation: based on topology optimization. 2. Slab Layout Algorithm Mathematical Principles: Halton Low Dissimilarity Sequences The planar coordinates (unit: mm) of the k-th magnetic piece; d i : Integer digits represented by radix d (d x =2,d y =3); m: number of terms in the sequence expansion (controls precision, default m=10). Magnetic field homogeneity verification: passed Solve Maxwell's equations.

[0029] Physical production stage (1) Precision 3D printing Control technology: PID temperature control Where, u(t): control signal output (heater power percentage %); e(t): deviation between temperature setpoint and actual value (unit: °C); K p ,K i ,K d Proportional / Integral / Derivative Gain (based on) (E5CC manual settings). Path planning: A* algorithm to avoid magnetic strip areas (2) Magnetic sheet hot pressing process Key parameters: The thermodynamic model uses the Fourier heat conduction equation; the pressure control is a constant pressure of 20N (error ±0.5N). Accuracy assurance: visual positioning is achieved through SIFT feature matching; robotic arm compensation ensures repeatability accuracy of ±0.01mm.

[0030] Quality verification phase (1) Geometric accuracy inspection Testing standard: Laser triangulation method. Tolerance formula: Δh: Average relief depth deviation (unit: mm); h i h: Measured height of the i-th sampling point (unit: mm); design Design height value (unit: mm); n: total number of sampling points (according to ISO 25178 standard). (2) Magnetic performance test Based on the standard: IEC 60404-5:2015 Magnetic Materials Standard; Weber distribution analysis of failure probability.

[0031] Social transmission stage Recommendation algorithms and collaborative filtering models: Where sim(u,v): the similarity of preferences between users u and v (range [-1,1]); r ui User u's rating of item i (normalized [0,1]); User u's average rating; I uv A collection of items rated by both users (u and v). UGC Community: Real-time interaction enables likes / comments via the WebSocket protocol (RFC 6455).

[0032] This invention achieves groundbreaking innovation through interdisciplinary technological restructuring: For the first time, Halton low-discrepancy sequences from computational geometry are introduced into magnetic sheet layout optimization. Empirical research published in the Journal of Magnetism and Magnetic Materials in 2021 verified that this technology reduces adsorption force fluctuations by 76%. Simultaneously, it integrates the Kinect Fusion algorithm from 3D reconstruction with LSCM texture mapping technology from computer graphics, breaking through the limitations of traditional planar design and achieving artistic adaptive mapping of curved substrates. This results in significant performance improvements—verified by actual testing, the system can shorten the production cycle of a single refrigerator magnet from approximately 45 minutes using traditional processes to 12 minutes, increasing efficiency by approximately 73%. The standard deviation of product adsorption force decreases from 38N to 9N, and the yield rate increases to 99.2%. The UGC community platform built by the system enables a 35% reuse rate of user-generated content, effectively improving user engagement and content dissemination efficiency. Attached Figure Description

[0033] Fig. 1 It is a diagram of the entire system process and core technology innovation.

[0034] Fig. 2 It is a 3D reconstruction and manufacturing execution detail diagram. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0036] The application principle of the present invention will be further described below with reference to the accompanying drawings.

[0037] like Figs. 1-2 As shown in the figure, the method and system for making AI-powered art refrigerator magnets provided by this embodiment of the invention include: a management module, an AI art generation module, a magnetic manufacturing module, and a social interaction module;

[0038] The following example, using the production of refrigerator magnets with embossed ancient Chinese poems, illustrates the implementation process of this system in detail:

[0039] Step 1: User Registration and Design Preparation (1) Account Creation Users register by entering their mobile phone number; the system automatically checks for duplicates and generates an account to be activated (status: pending); after administrator approval, a unique ID is assigned: USER_20240810_8F3C (2) Material configuration Select a silicone substrate thickness of 2.0mm and a UV coating process of 0.1mm ± 0.02mm; set the embossing process depth to 0.3mm ± 0.03mm; specify the NdFeB N52 magnetic sheet specifications as 8 pieces with a diameter of 5mm.

[0040] Step 2: AI Art Generation (1) Content input: The user uploaded the text of Wang Wei's "Autumn Evening in the Mountain Dwelling" and selected "Ink Painting Style" (Intensity coefficient α = 0.8). (2) Real-time processing Text effects unit: Generates vertical calligraphy fonts using the CycleGAN model, simulating the brushstroke characteristics of Yan Zhenqing. Style transfer unit: Blends landscape texture backgrounds using the AdaIN algorithm. Processing time: 0.8 seconds (actual measurement) (3) 3D editing Users can drag and adjust the following in the preview interface: the text position is moved down by 10%; the background transparency is reduced to 40%; the system renders a 3D effect image with lighting and shadows in real time.

[0041] Step 3: Precision Manufacturing (1) 3D printing stage The material used is PLA environmentally friendly plastic, with key parameters set at nozzle temperature of 210℃ (PID constant temperature control ±1℃); layer thickness of 0.1mm (dynamically adjustable curvature). Simultaneously, the avoidance design utilizes an A* algorithm to bypass eight pre-embedded magnetic sheet areas. (2) Magnetic sheet pressing stage The vision system uses SIFT feature matching coordinates (accuracy ±0.05mm) for positioning; a hot pressing process is performed in a constant temperature environment of 180℃ (temperature fluctuation ≤ ±5℃), constant pressure of 20N (error ±0.5N), and a duration of 30 seconds. The coordinates are calculated according to the Halton sequence and the magnetic sheet layout is performed according to formula (3) (base 2 / 3 distribution).

[0042] Step 4: Quality Verification (1) Geometric Detection Laser scanning of 100 sampling points to determine relief depth tolerance calculate. (2) Magnetic test The strength of each magnetic sheet measured by the gaussmeter is ≥1200 Gauss; the overall adsorption force fluctuation is ≤9N (76% lower than the traditional process).

[0043] Step 5: Social Dissemination (1) Generation of electronic portfolio cards Includes 3D renderings, production parameters, and a unique traceability code; users can scan the code to obtain the electronic version. (2) Community Posting After being uploaded to the "Inspiration Wall" UGC platform, the system automatically pushes the content to 50 users tagged "Poetry Lovers" and calculates the popularity score in real time (popularity = 0.7 × number of likes + 0.3 × number of comments). (3) Ecological closed loop When the popularity score is greater than 100, the design is included in the homepage recommendation pool, and other users can directly reuse the design (the system records a reuse rate of 38.7%).

[0044] Technical effect verification (1) Timeliness The entire process takes 14 minutes and 20 seconds (45 minutes with traditional methods), while AI processing takes a constant 0.8 seconds per image (unaffected by content complexity). (2) Quality stability After 10 cycles of hot and cold cycling (-20℃ to 70℃), the adsorption force decreased by ≤3%, and the yield rate was 99.2% (tested on 100 samples). (3) User value Design to delivery takes 15 minutes (meeting immediate customization needs), and social interaction increases the conversion rate of popular works by 52%.

[0045] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A system for creating AI-generated artistic refrigerator magnets, characterized in that, include: The system comprises a management module, an AI art generation module, a magnetic manufacturing module, and a social interaction module. The management module includes a user management unit, a material library unit, and a manufacturing management unit. The AI ​​art generation module includes a style transfer unit, a text effects unit, and a preview and editing unit. The magnetic manufacturing module includes a 3D printing unit and a magnetic bonding unit. The social interaction module includes an artwork recommendation unit and a UGC community unit.

2. The system according to claim 1, characterized in that: The user management unit is used to implement the registration / login process and assign a unique ID in the format of "USER date random code"; the material library unit is used to configure the magnetic substrate parameters and surface treatment process. The manufacturing management unit drives the 3D printer to perform high-precision layered manufacturing with a precision of 0.1mm via G-code.

3. The system according to claim 1, characterized in that: The style transfer unit uses the AdaIN algorithm for real-time artistic processing; the text effects unit uses the CycleGAN model to convert text into calligraphic fonts; and the preview and editing unit provides a 3D interactive interface that supports dragging and adjusting design elements.

4. The system according to claim 1, characterized in that: The 3D printing unit uses FDM fused deposition modeling technology with a layer thickness accuracy of ±0.05mm; the magnetic pressing unit uses a hot pressing process at 180±5℃ to embed neodymium iron boron magnetic sheets into the substrate.

5. The system according to claim 1, characterized in that: The work recommendation unit matches popular design templates based on a collaborative filtering algorithm; the UGC community unit builds an "inspiration wall" interactive platform to support work publishing.

6. A method for creating AI-generated artistic refrigerator magnets, characterized in that, include: After registering, users select material parameters; upload content and generate art designs using the AdaIN algorithm and CycleGAN model; and then manufacture the finished product using 3D printing and hot pressing processes. Share the finished product on social media platforms.

7. The method according to claim 6, characterized in that: The style intensity coefficient α of the AdaIN algorithm is [0,1]; the hot pressing process parameters are 180±5℃ and the duration is 30 seconds.

8. The method according to claim 6, characterized in that: Finished products pass geometric accuracy testing: The finished product passed the magnetic performance test: magnetic sheet strength ≥ 1200 Gauss.

9. The method according to claim 6, characterized in that: Social sharing includes: generating digital portfolio cards containing 3D renderings and production parameters; and pushing works to target users through collaborative filtering algorithms.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, it implements the steps of the method according to any one of claims 6-9.

Images