AI foot shoe tree model construction system based on image reconstruction and parameterization

Abstract

The invention discloses an AI foot shoe tree model construction system based on image reconstruction and parameterization, which comprises a multi-modal medical image fusion subsystem, a dynamic mechanical parameter adaptive subsystem, an AI feature enhancement subsystem, an intelligent parameterization driving subsystem and a real-time comfort simulation subsystem, the subsystems cooperatively work through data streams and interaction relations, and high-precision feature extraction of the extreme foot shape, individual specificity comfort simulation and rapid construction of a shoe tree model are achieved. By means of the multi-mode medical image fusion technology, low-dose CT images and optical scanning data are organically combined, the three-dimensional foot model containing skeleton and soft tissue layering is constructed, the problem that the robustness of extreme foot shape feature extraction is insufficient is solved, and the model construction precision is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of shoe last model construction, and in particular to an AI foot shoe last model construction system based on image reconstruction and parameterization. Background Art

[0002] The foot last model is a core tool in footwear design and manufacturing. It is essentially a three-dimensional model of the foot, but it is not a simple replica of the foot shape. Instead, it is a scientific design based on ergonomics, kinematics, and aesthetic requirements. Traditional methods for constructing foot last models mostly rely on contact measurement equipment or single image reconstruction technology, which has the disadvantages of low measurement accuracy and poor dynamic feature extraction. Especially when dealing with extreme foot shapes or motion blurred scenes, traditional methods have a high feature point missed detection rate and large positioning errors, resulting in distortion of the point cloud model. In addition, the physical simulation model does not fully consider individual differences, and the pressure distribution simulation error under dynamic gait is large, and adaptation deviations are prone to occur in the arch support area.

[0003] Therefore, we propose an AI foot last model construction system based on image reconstruction and parameterization. Summary of the Invention

[0004] The purpose of the present invention is to provide an AI foot shoe last model construction system based on image reconstruction and parameterization, which solves the problems raised in the background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an AI foot last model construction system based on image reconstruction and parameterization, comprising: Multimodal medical image fusion subsystem; Dynamic mechanical parameter adaptive subsystem; AI feature enhancement subsystem; Intelligent parameterized drive subsystem; Real-time comfort simulation subsystem; The multimodal medical image fusion subsystem, dynamic mechanical parameter adaptation subsystem, AI feature enhancement subsystem, intelligent parameterized drive subsystem and real-time comfort simulation subsystem work together through data flow and interactive relationships to achieve high-precision feature extraction of extreme foot shapes, individual-specific comfort simulation and rapid construction of shoe last models.

[0006] As a preferred embodiment of the present invention, the multimodal medical image fusion subsystem integrates low-dose CT images and optical scanning data, constructs a three-dimensional foot model containing bone and soft tissue layers through a medical image registration algorithm, uses a U-Net++ network to segment the CT image, obtains the soft tissue thickness distribution, and aligns the CT bone model with the optical point cloud through the ICP algorithm to generate a hybrid feature point cloud with anatomical structure.

[0007] As a preferred embodiment of the present invention, the dynamic mechanical parameter adaptive subsystem dynamically adjusts the soft tissue mechanical model based on individual parameters such as the user's BMI and age, uses the LSTM-GAN network to predict the arch deformation under different gaits, updates the finite element model parameters, and realizes individual-specific comfort simulation.

[0008] As a preferred embodiment of the present invention, the AI ​​feature enhancement subsystem achieves high-precision detection of anatomical points based on the YOLOv8 network, with a positioning accuracy of ≤0.5mm, and introduces an attention mechanism to improve detection robustness under complex lighting and background conditions.

[0009] As a preferred embodiment of the present invention, the intelligent parametric driving subsystem establishes a mapping relationship between foot shape features and shoe last parameters through the XGBoost algorithm, optimizes the mapping relationship in combination with multimodal data, and shortens the customization time for special foot shapes.

[0010] As a preferred embodiment of the present invention, the real-time comfort simulation subsystem integrates a physics engine and a CNN-LSTM model to generate a comfort score in real time, and introduces multi-gait dynamic simulation data to improve simulation accuracy.

[0011] As a preferred embodiment of the present invention, the special foot shape customization time is shortened by optimizing the XGBoost algorithm mapping relationship and combining multimodal data.

[0012] As a preferred embodiment of the present invention, the real-time comfort simulation subsystem reduces the pressure distribution simulation error by introducing multi-gait dynamic simulation data.

[0013] As a preferred embodiment of the present invention, when the U-Net++ network segments CT images, the soft tissue thickness distribution accuracy obtained is 0.2 mm.

[0014] Compared with the prior art, the present invention has the following beneficial effects: The present invention uses multimodal medical image fusion technology to organically combine low-dose CT images with optical scanning data to construct a three-dimensional foot model containing bones and soft tissue layers, which is conducive to solving the problem of insufficient robustness in extracting extreme foot shape features and improving the accuracy of model construction; the dynamic mechanical parameter adaptive subsystem dynamically adjusts the soft tissue mechanical model according to individual parameters such as the user's BMI and age, and introduces multi-gait dynamic simulation to reduce the pressure distribution simulation error, and adaptively adjusts the arch support parameters in multi-gait scenarios. The accuracy of arch fatigue prediction during running is improved, which is conducive to improving the accuracy of comfort simulation; the AI ​​feature enhancement subsystem realizes high-precision detection of anatomical points based on the YOLOv8 network, and introduces an attention mechanism to improve the detection robustness under complex light and background conditions; the intelligent parameterized driving subsystem establishes a mapping relationship between foot shape features and shoe last parameters through the XGBoost algorithm, and combines multimodal data to shorten the customization time of special foot shapes, thereby improving efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments with reference to the following drawings: Figure 1 This is an operation diagram of the system for constructing an AI foot last model based on image reconstruction and parameterization according to the present invention; Figure 2 This is a flow chart of the system for building an AI foot last model based on image reconstruction and parameterization in the present invention. DETAILED DESCRIPTION

[0016] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0017] The present invention proposes an AI foot shoe last model construction system based on image reconstruction and parameterization. Through the collaborative work of subsystems such as multimodal medical image fusion, dynamic mechanical parameter adaptation, AI feature enhancement, intelligent parameterized drive and real-time comfort simulation, it can achieve high-precision feature extraction of extreme foot shapes, individual-specific comfort simulation and rapid construction of shoe last models. Example

[0018] When implementing the AI ​​foot shoe last model construction system based on image reconstruction and parameterization, we selected a patient with severe hallux valgus as the test subject. The patient's foot features are complex, and traditional methods have certain difficulties in feature extraction and model construction. Through the system's multimodal medical image fusion, dynamic mechanical parameter adaptation, AI feature enhancement, intelligent parameterized drive and real-time comfort simulation subsystems, we aim to achieve high-precision feature extraction, individual-specific comfort simulation and rapid construction of shoe last models.

[0019] Here are the steps: 1. Implementation of multimodal medical image fusion subsystem Perform a low-dose CT scan on the patient, with the radiation dose controlled within 0.1mSv to ensure patient safety. The CT image scan obtains the bone and soft tissue structure of the foot, generating high-resolution CT image data. Use an optical scanning device to scan the patient's foot and obtain optical point cloud data of the foot surface. The U-Net++ network is used to segment CT images and obtain the soft tissue thickness distribution with an accuracy of 0.2mm. The CT bone model is registered with the optical point cloud using the ICP algorithm to generate a hybrid feature point cloud with anatomical structure. The registration process is iterated until the mean square error is reduced to within 0.2 mm to ensure the accuracy of the feature point cloud.

[0020] 2. Implementation of dynamic mechanical parameter adaptive subsystem Input the patient's BMI, age and other individual parameters to calculate the individual elastic modulus; According to the individual elastic modulus calculation formula E i =E0×(1+k×BMI+b×AGE), calculate the patient's elastic modulus.

[0021] For example, for a patient with a BMI of 28 and an age of 40, the elastic modulus is calculated as E i =700×(1+0.3×0.28+0.01×40)=864.8kPa.

[0022] Use the LSTM-GAN network to predict the arch deformation of patients in different gaits (such as standing and running); Update the finite element model parameters to adapt the model to the mechanical characteristics under different gaits.

[0023] Specifically, a 3D gait database was established, including standing, walking (4km / h), and running (8km / h), with each gait containing more than 200 dynamic arch deformation samples. The arch height changes of a specific user under different gaits are predicted using the LSTM-GAN network.

[0024] 3. Implementation of AI feature enhancement subsystem The YOLOv8 network is used to detect feature points in foot images, with positioning accuracy controlled within 0.5mm. An attention mechanism is introduced to improve detection robustness under complex lighting and background conditions.

[0025] The detected feature points are optimized to eliminate noise and redundant points and improve the quality of feature points.

[0026] 4. Implementation of intelligent parameterized drive subsystem The mapping relationship between foot shape features and shoe last parameters is established through the XGBoost algorithm. Combined with multimodal data (CT images, optical point clouds, individual parameters, etc.), the mapping relationship is optimized and the mapping accuracy is improved.

[0027] Generate a personalized shoe last model based on the mapping relationship to ensure the fit between the shoe last and the foot.

[0028] 5. Implementation of real-time comfort simulation subsystem Integrate the physics engine and CNN-LSTM model to perform real-time comfort simulation on the generated shoe last model.

[0029] Introducing multi-gait dynamic simulation data to improve simulation accuracy.

[0030] Generate a comfort score based on the simulation results to evaluate the comfort of the shoe last model; Adjust and optimize areas with lower scores until a satisfactory comfort level is achieved.

[0031] This embodiment successfully realizes the construction of an AI foot last model based on image reconstruction and parameterization through the collaborative work of subsystems such as multimodal medical image fusion, dynamic mechanical parameter adaptation, AI feature enhancement, intelligent parametric drive and real-time comfort simulation. The system has achieved significant improvements in feature extraction accuracy, dynamic simulation accuracy and customization time, providing an efficient and accurate solution for personalized shoe last customization.

[0032] In summary, the present invention uses multimodal medical image fusion technology to organically combine low-dose CT images with optical scanning data to construct a three-dimensional foot model containing bones and soft tissue layers, which is conducive to solving the problem of insufficient robustness in extracting extreme foot shape features and improving the accuracy of model construction; the dynamic mechanical parameter adaptive subsystem dynamically adjusts the soft tissue mechanical model according to individual parameters such as user BMI and age, and introduces multi-gait dynamic simulation to reduce the pressure distribution simulation error, and adaptively adjusts the arch support parameters in multi-gait scenarios. The accuracy of arch fatigue prediction during running is improved, which is conducive to improving the accuracy of comfort simulation; the AI ​​feature enhancement subsystem realizes high-precision detection of anatomical points based on the YOLOv8 network, and introduces the attention mechanism to improve the detection robustness under complex light and background conditions; the intelligent parameterized driving subsystem establishes a mapping relationship between foot shape features and shoe last parameters through the XGBoost algorithm, and combines multimodal data to shorten the customization time of special foot shapes, thereby improving efficiency.

[0033] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be regarded as illustrative and non-restrictive in all respects. The scope of the present invention is defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the claims are intended to be included therein.

[0034] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. An AI foot shoe last model construction system based on image reconstruction and parameterization, characterized in that: include: Multimodal medical image fusion subsystem; Dynamic mechanical parameter adaptive subsystem; AI feature enhancement subsystem; Intelligent parameterized drive subsystem; Real-time comfort simulation subsystem; The multimodal medical image fusion subsystem, dynamic mechanical parameter adaptation subsystem, AI feature enhancement subsystem, intelligent parameterized drive subsystem and real-time comfort simulation subsystem work together through data flow and interactive relationships to achieve high-precision feature extraction of extreme foot shapes, individual-specific comfort simulation and rapid construction of shoe last models.

2. The AI ​​foot shoe last model construction system based on image reconstruction and parameterization according to claim 1, characterized in that: The multimodal medical image fusion subsystem integrates low-dose CT images and optical scanning data, constructs a three-dimensional foot model containing bone and soft tissue layers through a medical image registration algorithm, uses a U-Net++ network to segment the CT images, obtains the soft tissue thickness distribution, and aligns the CT bone model with the optical point cloud through the ICP algorithm to generate a hybrid feature point cloud with anatomical structure.

3. The AI ​​foot shoe last model construction system based on image reconstruction and parameterization according to claim 1, characterized in that: The dynamic mechanical parameter adaptive subsystem dynamically adjusts the soft tissue mechanical model based on individual parameters such as the user's BMI and age, uses the LSTM-GAN network to predict the arch deformation under different gaits, updates the finite element model parameters, and realizes individual-specific comfort simulation.

4. The AI ​​foot shoe last model construction system based on image reconstruction and parameterization according to claim 1, characterized in that: The AI ​​feature enhancement subsystem achieves high-precision detection of anatomical points based on the YOLOv8 network, with a positioning accuracy of ≤0.5mm, and introduces an attention mechanism to improve detection robustness under complex lighting and background conditions.

5. The AI ​​foot shoe last model construction system based on image reconstruction and parameterization according to claim 1, characterized in that: The intelligent parametric driving subsystem uses the XGBoost algorithm to establish a mapping relationship between foot shape features and shoe last parameters, and optimizes the mapping relationship in combination with multimodal data, thereby shortening the customization time for special foot shapes.

6. The AI ​​foot shoe last model construction system based on image reconstruction and parameterization according to claim 1, characterized in that: The real-time comfort simulation subsystem integrates a physics engine and a CNN-LSTM model to generate a comfort score in real time and introduces multi-gait dynamic simulation data to improve simulation accuracy.

7. The AI ​​foot shoe last model construction system based on image reconstruction and parameterization according to claim 1, characterized in that: The special foot shape customization time is shortened by optimizing the XGBoost algorithm mapping relationship and combining multimodal data.

8. The AI ​​foot shoe last model construction system based on image reconstruction and parameterization according to claim 1, characterized in that: The real-time comfort simulation subsystem reduces the pressure distribution simulation error by introducing multi-gait dynamic simulation data.

9. The AI ​​foot shoe last model construction system based on image reconstruction and parameterization according to claim 1, characterized in that: When the U-Net++ network segments CT images, the obtained soft tissue thickness distribution accuracy is 0.2 mm.

Images