Computer vision-assisted dental restoration adaptation system

Abstract

The invention relates to the technical field of computer vision, and discloses a computer vision-assisted dental restoration adaptation system, which comprises an image acquisition module, an image processing module, an adaptation evaluation module, a central control module and a feedback optimization unit, high-resolution image data, including visible light images, three-dimensional scanning data and infrared images, of the interior of the oral cavity and a restoration body are obtained through multi-source imaging equipment of the image acquisition module, the form, texture and temperature distribution characteristics of the oral cavity environment are captured, and the image contrast and the alignment precision are enhanced in combination with the illumination control unit and the positioning calibration unit. Data integrity and detail visibility are ensured, and a reliable basis is provided for subsequent image processing; the image processing module performs preprocessing, feature extraction and segmentation on the acquired image data, so that the accuracy and efficiency of adaptive state analysis are improved.

Description

Technical Field

[0001] This invention relates to the field of computer vision technology, specifically to a computer vision-assisted dental prosthesis fitting system. Background Technology

[0002] Computer vision is a science that studies how to make machines "see." More specifically, it refers to machine vision, which uses cameras and computers to replace human eyes to identify, track, and measure targets, and further performs image processing to make the computer-processed images more suitable for human observation or transmission to instruments for detection. As a scientific discipline, computer vision studies related theories and technologies in an attempt to build artificial intelligence systems that can extract "information" from images or multidimensional data.

[0003] Currently, due to the complex oral environment and multi-source data requirements involved in the fitting process of dental prostheses, traditional image acquisition methods cannot simultaneously acquire high-resolution visible light, 3D scanning and infrared image data through multimodal imaging equipment when conducting dental prosthesis fitting assessment. If the image acquisition is incomplete or there are uneven illumination or positioning errors, morphological, texture and temperature distribution features may be missed, resulting in inaccurate boundary recognition, gap analysis and fitting status judgment in image processing.

[0004] Therefore, a computer vision-assisted dental prosthesis fitting system is proposed to solve the above problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a computer vision-assisted dental prosthesis fitting system, which solves the problems mentioned in the background technology, such as incomplete image acquisition or uneven lighting and positioning errors, which may lead to the omission of morphological, texture and temperature distribution features, resulting in inaccurate boundary recognition, gap analysis and fitting status judgment in image processing.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a computer vision-assisted dental prosthesis fitting system, the system comprising an image acquisition module, an image processing module, a fitting evaluation module and a central control module, wherein the image acquisition module, the image processing module and the fitting evaluation module are respectively communicatively connected to the central control module; The image acquisition module is used to acquire high-resolution image data of the oral cavity and prostheses through multi-source imaging devices, including visible light images, three-dimensional scan data and infrared images, to capture the morphological, texture and temperature distribution characteristics of the oral environment; The image processing module is used to preprocess, extract features and segment the acquired image data, identify the boundary, gap and fit status of the restoration and tooth tissue through computer vision algorithms, and eliminate image noise and distortion. The fit assessment module is used to quantify the fit accuracy of the restoration based on the processed image data, including edge fit, placement depth and stress distribution analysis, and to generate a fit assessment report. The central control module is used to coordinate the collaborative operation of various modules, receive image data and perform analysis and processing, control the evaluation process according to preset adaptation standards, and provide a human-computer interaction interface to support parameter setting and real-time result display. The system also integrates a feedback optimization unit, which triggers adjustment suggestions when there are adaptation anomalies or the evaluation fails to meet the standards, and records experimental data for subsequent model optimization and clinical decision support.

[0007] Preferably, the image acquisition module includes a multimodal imaging unit, an illumination control unit, and a positioning calibration unit; The multimodal imaging unit integrates a visible light camera, a 3D structured light scanner, and an infrared thermal imager to achieve multi-angle, multi-spectral image acquisition; The lighting control unit uses a tunable LED light source array to simulate lighting conditions in different oral environments, including natural light, surgical light illumination, and dark scenes; The positioning and calibration unit controls the position and angle of the imaging device through a robotic arm and an optical marking system.

[0008] Preferably, the multimodal imaging unit is also equipped with a real-time dynamic capture component to record the deformation process of the restoration during simulated occlusion and functional movement; The lighting control unit includes a spectral analysis subunit, which optimizes imaging effects for different tissue types by adjusting wavelength and intensity; The positioning and calibration unit is equipped with an automatic tracking mechanism and an error compensation algorithm to correct equipment drift in real time and ensure the repeatability of the collected trajectory.

[0009] Preferably, the image processing module includes a preprocessing unit, a feature analysis unit, and a model reconstruction unit; The preprocessing unit is used to filter, denoise, and geometrically correct the original image, and uses adaptive thresholding segmentation and morphological operations to enhance key regions; The feature analysis unit extracts the features of the restoration margins, tooth contours, and gaps using a convolutional neural network algorithm, and calculates morphological parameters; The model reconstruction unit generates a three-dimensional digital model based on multi-view stereo vision technology, realizing virtual registration and overlay display of restorations and oral tissues.

[0010] Preferably, the preprocessing unit supports multi-scale filtering and non-uniform illumination correction; The feature analysis unit integrates a deep learning model to identify common adaptation defects through a training dataset. The model reconstruction unit uses point cloud registration and surface fitting algorithms, and supports real-time updates and comparison with historical data.

[0011] Preferably, the adaptation evaluation module includes a precision quantization unit, a stress simulation unit, and a report generation unit; The precision quantization unit is used to calculate the marginal gap, placement deviation and contact area between the restoration and the tooth, and outputs a fit score by comparing the measured data with the ideal standard. The stress simulation unit uses finite element analysis algorithms to predict the stress distribution of the restoration under load and assess the long-term stability risk. The report generation unit automatically generates a richly illustrated assessment summary, including defect annotations and improvement suggestions.

[0012] Preferably, the precision quantification unit has a multi-index fusion function, which performs a weighted evaluation by comprehensively considering geometric errors and biomechanical parameters; The stress simulation unit supports dynamic load simulation, including chewing cycle and temperature change scenarios; The report generation unit offers customizable templates and supports data export and cloud sharing.

[0013] Preferably, the central control module includes a main control scheduling unit, a data fusion unit, and an interactive interface unit; The main control scheduling unit is used to manage the timing logic of image acquisition, processing, and evaluation; The data fusion unit integrates multi-source image data with clinical parameters, and uses cluster analysis to remove redundancy and improve decision reliability. The interactive interface unit provides a visual operation panel that supports real-time monitoring, parameter adjustment, and result visualization.

[0014] Preferably, the main control scheduling unit adopts a distributed computing architecture, which allows for modular expansion and parallel processing; The data fusion unit integrates a multi-algorithm voting mechanism to reduce misjudgments through consistency checks; The interactive interface unit supports touch and voice input, and has auxiliary diagnostic prompts.

[0015] Preferably, the feedback optimization unit includes an anomaly detection subunit and an adaptive learning subunit; The anomaly detection subunit automatically triggers alarms based on the adaptation threshold and pushes repair suggestions; The adaptive learning subunit iteratively optimizes and evaluates the model through machine learning.

[0016] Compared with the prior art, the present invention provides a computer vision-assisted dental prosthesis fitting system, which has the following beneficial effects: 1. In this invention, high-resolution image data of the oral cavity and prosthesis are acquired through a multi-source imaging device of the image acquisition module, including visible light images, three-dimensional scan data and infrared images, capturing the morphological, texture and temperature distribution characteristics of the oral environment, and combining the illumination control unit and the positioning calibration unit to enhance image contrast and alignment accuracy, ensuring data integrity and detail visibility, and providing a reliable foundation for subsequent image processing.

[0017] 2. In this invention, the image processing module preprocesses, extracts and segments the acquired image data, uses computer vision algorithms and convolutional neural networks to identify the boundaries, gaps and fit status of the restoration and dental tissues, and generates a three-dimensional digital model through the model reconstruction unit to achieve virtual registration between the restoration and oral tissues, thereby improving the accuracy and efficiency of fit status analysis.

[0018] 3. In this invention, the fitting evaluation module quantifies the fitting accuracy of the prosthesis based on the processed image data, including edge fit, placement depth and stress distribution analysis, and automatically generates a fitting evaluation report. At the same time, the feedback optimization unit triggers adjustment suggestions when the fitting is abnormal or the evaluation is not up to standard. Through adaptive learning to optimize the evaluation model, the reliability of prosthesis fitting and clinical decision support capabilities are improved. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the computer vision-assisted dental prosthesis fitting system of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Specific embodiment: A computer vision-assisted dental prosthesis fitting system, the system includes an image acquisition module, an image processing module, a fitting evaluation module and a central control module, the image acquisition module, the image processing module and the fitting evaluation module are respectively communicatively connected to the central control module; The image acquisition module is used to acquire high-resolution image data of the oral cavity and prostheses through multi-source imaging devices, including visible light images, three-dimensional scan data and infrared images, to capture the morphological, texture and temperature distribution characteristics of the oral environment; The image processing module is used to preprocess, extract features and segment the acquired image data, identify the boundary, gap and fit status of the restoration and tooth tissue through computer vision algorithms, and eliminate image noise and distortion. The fit assessment module is used to quantify the fit accuracy of the restoration based on the processed image data, including edge fit, placement depth and stress distribution analysis, and to generate a fit assessment report. The central control module coordinates the collaborative operation of each module, receives and analyzes image data, controls the evaluation process according to preset adaptation standards, and provides a human-computer interaction interface to support parameter setting and real-time result display. The system also integrates a feedback optimization unit, which triggers adjustment suggestions when there are adaptation anomalies or the evaluation fails to meet the standards, and records experimental data for subsequent model optimization and clinical decision support.

[0022] The image acquisition module includes a multimodal imaging unit, an illumination control unit, and a positioning and calibration unit; The multimodal imaging unit integrates a visible light camera, a 3D structured light scanner, and an infrared thermal imager to achieve multi-angle and multi-spectral image acquisition, ensuring data integrity and redundant backup; The lighting control unit uses a tunable LED light source array to simulate lighting conditions in different oral environments, including natural light, surgical lighting, and dark scenes, enhancing image contrast and detail visibility; The positioning and calibration unit controls the position and angle of the imaging device through a robotic arm and an optical marking system, reducing acquisition errors and ensuring image alignment accuracy.

[0023] The multimodal imaging unit is also equipped with a real-time dynamic capture component to record the deformation process of the prosthesis during simulated occlusion and functional movement; The lighting control unit includes a spectral analysis subunit, which optimizes imaging effects for different tissue types by adjusting wavelength and intensity; First, the spectral analysis subunit integrates a tunable LED array and a spectral sensor. The light source intensity is adjusted by controlling the drive current, and the wavelength is switched using a digital micromirror device and a filter wheel. The wavelength adjustment range covers 400nm-1000nm to match the reflection characteristics of oral tissues. The optimization process is based on a pre-set database of tissue optical parameters, adjusting through iterative feedback, and using a gradient descent algorithm to minimize the image-to-noise ratio. The formula is as follows: ; in Indicates wavelength. Indicates intensity. This represents the observed signal value. Represents the ideal signal value; The positioning and calibration unit is equipped with an automatic tracking mechanism and an error compensation algorithm to correct equipment drift in real time and ensure the repeatability of the collected trajectory; The error compensation algorithm acquires device position data through the encoder and inertial measurement unit of the positioning calibration unit, and applies Kalman filtering for real-time correction. The formula is as follows: ; in Indicates time The state vector, Indicates time The state vector, Represents the state transition matrix. Represents the control matrix. Indicates time The input vector, This represents process noise. By predicting and updating steps, drift error is reduced, ensuring that the repeatability accuracy of the acquired trajectory is less than 10 micrometers.

[0024] The image processing module includes a preprocessing unit, a feature analysis unit, and a model reconstruction unit; The preprocessing unit is used to filter, denoise, and geometrically correct the original image, and uses adaptive thresholding segmentation and morphological operations to enhance key regions; The preprocessing unit first filters and denoises the original image, using a Gaussian filter kernel for smoothing, as shown in the formula: ; in Indicates the position of the Gaussian filter kernel. The value at that location, and Represents spatial coordinates in the image. Indicates standard deviation, Represents the natural constant. Represents pi; Next, geometric correction is performed, using affine transformation to correct image distortion. The formula is: ; in and Indicates the corrected image coordinates. and Represents the coordinates of the original image. Represents a transformation matrix, one The affine matrix contains rotation, scaling, and translation parameters, used to correct image distortion; The feature analysis unit extracts the features of the restoration margins, tooth contours, and gaps using a convolutional neural network algorithm, and calculates morphological parameters; The feature analysis unit uses the Convolutional Neural Network (CNN) algorithm, based on the ResNet architecture, to extract the edge features of the restoration through multiple convolutional and pooling layers, and then calculates the morphological parameters after outputting the feature map. The convolutional neural network employs an encoder-decoder structure based on the ResNet-50 architecture. The encoder part is initialized using ResNet-50 weights pre-trained on the ImageNet dataset to extract multi-level features. The decoder part consists of a series of upsampling layers and convolutional layers, progressively restoring the feature map resolution to the input image size. The training method for the CNN model is as follows: Training dataset: A dataset of visible light and 3D scan images containing at least 10,000 labeled dental restorations and natural teeth was used. The data sources were the publicly available dataset ToothFairyDataset and anonymized clinical data from partner hospitals. Data augmentation techniques included random rotation, brightness and contrast adjustment, and Gaussian noise injection to improve model robustness.

[0025] Training process: The model uses the Adam optimizer with an initial learning rate of 1e-4. The loss function is a composite loss function combining binary cross-entropy and Dice coefficient to optimize the segmentation accuracy of edge pixels. Training is performed with a batch size of 16 and a total of 100 iterations. Training is terminated early when the validation set loss no longer decreases for 5 consecutive iterations. The model reconstruction unit generates a three-dimensional digital model based on multi-view stereo vision technology, realizing virtual registration and overlay display of restorations and oral tissues; The model reconstruction unit uses multi-view stereo vision technology to generate point clouds through camera calibration and feature matching, and then uses the Poisson reconstruction algorithm to generate a 3D model to achieve virtual registration.

[0026] The preprocessing unit supports multi-scale filtering and non-uniform illumination correction to eliminate motion artifacts and specular reflection interference. Multi-scale filtering employs wavelet transform and uses the Daubechies wavelet basis to perform multi-resolution decomposition of the image, as shown in the formula: ; in Indicates the wavelet transform coefficients at scale Peaceful relocation The value at that location, Indicates the scale parameter. Indicates the translation parameter. Represents the original signal. Represents the wavelet function; Non-uniform illumination correction uses homomorphic filtering to convert the image to the frequency domain and adjust the brightness component. The feature analysis unit integrates a deep learning model to identify common adaptation defects through a training dataset. The model reconstruction unit uses point cloud registration and surface fitting algorithms to ensure that the accuracy of the 3D model is below the micrometer level, and supports real-time updates and comparison with historical data; Point cloud registration uses the Iterative Nearest Point (ICP) algorithm to minimize the distance between point pairs. The formula is: ; in Represents the rotation matrix. Represents the translation vector. Represents the first point in the source point cloud. Coordinates of a point, Represents the first point in the target point cloud The coordinates of the points and are correspond; For surface fitting, B-splines are used to fit a smooth surface through control points, ensuring model accuracy.

[0027] The adaptation evaluation module includes a precision quantization unit, a stress simulation unit, and a report generation unit; The precision quantization unit is used to calculate the marginal gap, placement deviation and contact area between the restoration and the tooth, and outputs a fit score by comparing the measured data with the ideal standard. The precision quantization unit calculates the marginal gap, placement deviation, and contact area. First, the contours of the restoration and tooth are obtained through image segmentation. The marginal gap is calculated as the average Euclidean distance between corresponding points, using the following formula: ; in This represents the average value of the edge gap. Indicates the number of point pairs. and Indicates the first on the edge of the restoration The coordinates of the points and Indicates the first tooth at the edge of the tooth. The coordinates of the points and are Point correspondence; The positioning deviation is obtained by comparing the actual position with the ideal position, and the contact area is obtained by summing the pixel count and the grid area.

[0028] The stress simulation unit uses finite element analysis algorithms to predict the stress distribution of the restoration under load and assess the long-term stability risk. The stress simulation element uses the Finite Element Analysis (FEA) algorithm to discretize the 3D model into a mesh, and applies a linear elastic model to solve for the stress distribution. The governing equations are: ; in Denotes the divergence operator, Represents the stress tensor. Represents the force vector; The report generation unit automatically generates a richly illustrated assessment summary, including defect annotations and improvement suggestions.

[0029] The precision quantification unit has a multi-index fusion function, which performs a weighted evaluation by combining geometric error and biomechanical parameters. Multi-indicator fusion first normalizes each indicator, then uses a weighted summation for fusion, as shown in the formula: ; in This indicates the overall compatibility score. Indicates the weighting coefficient. This represents the normalized index value; The weighted assessment is based on fuzzy logic, and the weights are dynamically adjusted to reflect clinical importance, ensuring that the assessment results are comprehensive. The stress simulation unit supports dynamic load simulation, including chewing cycle and temperature change scenarios; The report generation unit offers customizable templates and supports data export and cloud sharing.

[0030] The central control module includes a main control and scheduling unit, a data fusion unit, and an interactive interface unit; The main control scheduling unit is used to manage the timing logic of image acquisition, processing, and evaluation; The data fusion unit integrates multi-source image data with clinical parameters, and uses cluster analysis to remove redundancy and improve decision reliability. The interactive interface unit provides a visual operation panel that supports real-time monitoring, parameter adjustment, and result visualization.

[0031] The main control and scheduling unit adopts a distributed computing architecture, which allows for modular expansion and parallel processing; The main control and scheduling unit adopts a microservice architecture, decomposing tasks into image acquisition and processing sub-tasks, and communicating through the message queue RabbitMQ; Distributed computing uses the MapReduce framework to distribute image segmentation tasks to multiple nodes for parallel processing. The master node schedules resources and monitors the status to ensure high throughput and fault tolerance. The data fusion unit integrates a multi-algorithm voting mechanism to reduce misjudgments through consistency checks; The specific implementation steps of the multi-algorithm voting mechanism are as follows: First, algorithm integration and initialization are performed. The system loads multiple core algorithms in parallel and assigns appropriate initial weights to each algorithm, laying the foundation for subsequent collaborative decision-making. Second, parallel computation and voting decision-making are executed. Each algorithm independently analyzes the input data and outputs results. The system uses a weighted voting mechanism to synthesize the outputs of all algorithms to arrive at a final decision, avoiding potential misjudgments from a single algorithm. Subsequently, consistency checks and arbitration are initiated. The system calculates the consistency coefficient of the algorithm group's decision. If the consistency is higher than a safety threshold, the voting result is adopted; if the consistency is insufficient, the arbitration mechanism is triggered, marking the result as requiring manual review to ensure reliability. The consistency check formula is expressed as: ; in: This represents the consistency coefficient, with values ​​ranging from -1 to 1. The closer the coefficient is to 1, the higher the consistency. This represents the observational consistency ratio, which is the proportion of actual consistency in the algorithm output. It represents the expected consistency ratio, that is, the consistency ratio of random expectations; Finally, the closed-loop optimization is completed. The system compares the decision results of each vote with the actual effect and dynamically adjusts the weight of each algorithm accordingly, forming a continuously self-optimizing closed loop, so that the entire voting mechanism becomes more and more accurate as the number of times it is used increases. The interactive interface unit supports touch and voice input, and has auxiliary diagnostic prompts.

[0032] The feedback optimization unit includes an anomaly detection subunit and an adaptive learning subunit; The anomaly detection subunit triggers alarms and pushes remediation suggestions based on the adaptation threshold through the rule engine and statistical process control. The adaptive learning subunit iteratively optimizes and evaluates the model through machine learning, and uses historical data to improve prediction accuracy and system generalization ability. The adaptive learning subunit uses the machine learning algorithm: Random Forest, to iteratively train the model using historical data, with the loss function being: ; in This represents the value of the loss function. Indicates the number of samples. Indicates the predicted value. Representing the true values, gradient descent is used to optimize the parameters and improve the system accuracy; The specific implementation and optimization process of the random forest model is as follows: Feature engineering: The input feature set of the model includes all metrics output from the quantization evaluation unit and extends its derived features, including but not limited to: first-order differences, sliding window statistics, and deviations from the ideal value.

[0033] Model training and iterative optimization: Initial training: Historical successful restoration evaluation data and restoration data with known problems were used as the initial training set. The number of trees in the random forest was set to 100, the maximum depth to 20, and the Gini coefficient was used as the node splitting criterion.

[0034] Online learning: After system deployment, this unit starts online learning mode; each time the evaluation results are confirmed and corrected by the expert system, the sample and its label will be added to a first-in-first-out buffer; every 24 hours, or when the buffer accumulates more than 100 new samples, the system will use the samples in the buffer to perform incremental learning on the random forest model, dynamically adjust the model weights to adapt to changes in the distribution of clinical data, thereby achieving continuous optimization of the model.

[0035] The principle and steps of this system are as follows: First, high-resolution image data of the oral cavity and restorations are acquired through a multi-source imaging device in the image acquisition module. This includes visible light images, 3D scan data, and infrared images to capture the morphological, textural, and temperature distribution characteristics of the oral environment. The image acquisition module comprises a multimodal imaging unit, an illumination control unit, and a positioning and calibration unit. The multimodal imaging unit integrates a visible light camera, a 3D structured light scanner, and an infrared thermal imager to achieve multi-angle, multi-spectral image acquisition. The illumination control unit uses a tunable LED light source array to simulate different lighting conditions in the oral cavity. The positioning and calibration unit controls the position and angle of the imaging device through a robotic arm and an optical marking system, enhancing image contrast and alignment accuracy, ensuring data integrity and detail visibility. The multimodal imaging unit is also equipped with a real-time dynamic capture component to record the deformation process of the restoration during simulated occlusion and functional movements. The illumination control unit includes a spectral analysis subunit to optimize imaging effects for different tissue types. The positioning and calibration unit features an automatic tracking mechanism and error compensation algorithm to correct device drift in real time and ensure the repeatability of the acquisition trajectory.

[0036] Next, the image processing module preprocesses, extracts features, and segments the acquired image data. The image processing module includes a preprocessing unit, a feature analysis unit, and a model reconstruction unit. The preprocessing unit filters, denoises, and geometrically corrects the original image, employs adaptive thresholding and morphological operations to enhance key regions, and supports multi-scale filtering and non-uniform lighting correction. The feature analysis unit extracts features of the restoration's edges, tooth contours, and gaps using a convolutional neural network algorithm, calculates morphological parameters, and integrates a deep learning model to identify common fitting defects. The model reconstruction unit generates a 3D digital model based on multi-view stereo vision technology, uses point cloud registration and surface fitting algorithms to achieve virtual registration and overlay display of the restoration and oral tissues, and supports real-time updates and comparison with historical data.

[0037] Then, the fitting assessment module quantifies the fitting accuracy of the restoration based on the processed image data. The fitting assessment module includes an accuracy quantification unit, a stress simulation unit, and a report generation unit. The accuracy quantification unit calculates the marginal clearance, placement deviation, and contact area between the restoration and the tooth, outputting a fitting score by comparing measured data with ideal standards. It also features multi-index fusion capabilities, comprehensively considering geometric errors and biomechanical parameters for weighted evaluation. The stress simulation unit uses finite element analysis algorithms to predict the stress distribution of the restoration under load, assessing long-term stability risks and supporting dynamic load simulation of chewing cycles and temperature changes. The report generation unit automatically generates a richly illustrated assessment summary, including defect annotations and improvement suggestions, and provides customizable templates to support data export and cloud sharing.

[0038] The central control module coordinates the collaborative operation of the image acquisition module, image processing module, and adaptation evaluation module. It receives and analyzes image data, controls the evaluation process according to preset adaptation standards, and provides a human-computer interaction interface to support parameter setting and real-time result display. The central control module includes a main control scheduling unit, a data fusion unit, and an interactive interface unit. The main control scheduling unit manages the temporal logic of image acquisition, processing, and evaluation, employing a distributed computing architecture that allows for modular expansion and parallel processing. The data fusion unit integrates multi-source image data with clinical parameters, removes redundancy through cluster analysis to improve decision reliability, and integrates a multi-algorithm voting mechanism to reduce misjudgments through consistency checks. The interactive interface unit provides a visual operation panel, supporting real-time monitoring, parameter adjustment, and result visualization, as well as touch and voice input and auxiliary diagnostic prompts.

[0039] Finally, the system integrates a feedback optimization unit that triggers adjustment suggestions when adaptation anomalies or evaluation failures occur, and records experimental data for subsequent model optimization and clinical decision support. The feedback optimization unit includes an anomaly detection subunit and an adaptive learning subunit. The anomaly detection subunit automatically triggers alarms and pushes repair suggestions based on adaptation thresholds; the adaptive learning subunit iteratively optimizes the evaluation model through machine learning, improving the system's accuracy and generalization ability.

[0040] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0041] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A computer vision-assisted dental prosthesis fitting system, characterized in that: The system includes an image acquisition module, an image processing module, an adaptation evaluation module, and a central control module, wherein the image acquisition module, the image processing module, and the adaptation evaluation module are respectively communicatively connected to the central control module; The image acquisition module is used to acquire high-resolution image data of the oral cavity and prosthesis through multi-source imaging devices, including visible light images, three-dimensional scan data and infrared images, to capture the morphological, texture and temperature distribution characteristics of the oral environment; The image processing module is used to preprocess, extract features and segment the acquired image data, identify the boundary, gap and fit status of the restoration and the tooth tissue through computer vision algorithms, and eliminate image noise and distortion. The fitting evaluation module is used to quantify the fitting accuracy of the restoration based on the processed image data, including edge fit, placement depth and stress distribution analysis, and to generate a fitting evaluation report. The central control module is used to coordinate the collaborative operation of each module, receive image data and perform analysis and processing, control the evaluation process according to preset adaptation standards, and provide a human-computer interaction interface to support parameter setting and real-time result display. The system also integrates a feedback optimization unit, which triggers adjustment suggestions when the adaptation is abnormal or the evaluation is not up to standard, and records experimental data for subsequent model optimization and clinical decision support.

2. The computer vision-assisted dental prosthesis fitting system according to claim 1, characterized in that: The image acquisition module includes a multimodal imaging unit, an illumination control unit, and a positioning calibration unit; The multimodal imaging unit integrates a visible light camera, a three-dimensional structured light scanner, and an infrared thermal imager to achieve multi-angle, multi-spectral image acquisition. The lighting control unit uses a tunable LED light source array to simulate different lighting conditions in the oral cavity, including natural light, surgical light illumination, and dark scene. The positioning and calibration unit controls the position and angle of the imaging device through a robotic arm and an optical marking system.

3. The computer vision-assisted dental prosthesis fitting system according to claim 2, characterized in that: The multimodal imaging unit is also equipped with a real-time dynamic capture component for recording the deformation process of the prosthesis during simulated occlusion and functional movement. The lighting control unit includes a spectral analysis subunit, which optimizes the imaging effect of different tissue types by adjusting the wavelength and intensity; The positioning calibration unit is equipped with an automatic tracking mechanism and an error compensation algorithm to correct device drift in real time and ensure the repeatability of the collected trajectory.

4. The computer vision-assisted dental prosthesis fitting system according to claim 1, characterized in that: The image processing module includes a preprocessing unit, a feature analysis unit, and a model reconstruction unit; The preprocessing unit is used to filter, denoise, and geometrically correct the original image, and to enhance key regions using adaptive threshold segmentation and morphological operations. The feature analysis unit extracts the features of the restoration's edge, tooth contour, and gaps using a convolutional neural network algorithm, and calculates morphological parameters. The model reconstruction unit generates a three-dimensional digital model based on multi-view stereo vision technology, realizing virtual registration and overlay display of the restoration and oral tissue.

5. The computer vision-assisted dental prosthesis fitting system according to claim 4, characterized in that: The preprocessing unit supports multi-scale filtering and non-uniform illumination correction. The feature analysis unit integrates a deep learning model to identify common adaptation defects through a training dataset. The model reconstruction unit employs point cloud registration and surface fitting algorithms, and supports real-time updates and comparison with historical data.

6. The computer vision-assisted dental prosthesis fitting system according to claim 1, characterized in that: The adaptation evaluation module includes a precision quantization unit, a stress simulation unit, and a report generation unit. The precision quantification unit is used to calculate the marginal gap, placement deviation and contact area between the restoration and the tooth, and outputs the fit score by comparing the measured data with the ideal standard. The stress simulation unit predicts the stress distribution of the restoration under load using finite element analysis algorithms, and assesses the long-term stability risk. The report generation unit automatically generates a graphic-rich evaluation summary, including defect annotations and improvement suggestions.

7. The computer vision-assisted dental prosthesis fitting system according to claim 6, characterized in that: The precision quantification unit has a multi-index fusion function, which performs a weighted evaluation by combining geometric error and biomechanical parameters. The stress simulation unit supports dynamic load simulation, including chewing cycles and temperature change scenarios; The report generation unit provides customizable templates and supports data export and cloud sharing.

8. The computer vision-assisted dental prosthesis fitting system according to claim 1, characterized in that: The central control module includes a main control and scheduling unit, a data fusion unit, and an interactive interface unit; The main control scheduling unit is used to manage the timing logic of image acquisition, processing and evaluation; The data fusion unit integrates multi-source image data with clinical parameters, and uses cluster analysis to remove redundancy and improve decision reliability. The interactive interface unit provides a visual operation panel that supports real-time monitoring, parameter adjustment, and result visualization.

9. The computer vision-assisted dental prosthesis fitting system according to claim 8, characterized in that: The main control scheduling unit adopts a distributed computing architecture, which allows for modular expansion and parallel processing; The data fusion unit integrates a multi-algorithm voting mechanism to reduce misjudgments through consistency checks; The interactive interface unit supports touch and voice input, and has auxiliary diagnostic prompts.

10. The computer vision-assisted dental prosthesis fitting system according to claim 1, characterized in that: The feedback optimization unit includes an anomaly detection subunit and an adaptive learning subunit; The anomaly detection subunit automatically triggers an alarm based on the adaptation threshold and pushes repair suggestions; The adaptive learning subunit iteratively optimizes the evaluation model through machine learning.

Images