A three-dimensional scanning-based personalized denture intelligent design system and method

Abstract

The application relates to the technical field of digital dental manufacturing, and discloses an intelligent design system and method for personalized dentures based on three-dimensional scanning, which comprises the following modules: a feature analysis and detection module, which is used for receiving scanning data with entity line drawing, extracting anatomic feature boundaries, and calculating the local normal thickness of grid vertices; a mechanical self-adaptive reinforcement module, which is used for calculating dynamic required thickness in combination with layout data, and performing internal thickening compensation on weak areas on the premise of maintaining the feature boundaries unchanged; a shock and vibration prevention tool path optimization module, which is used for dynamically adjusting the feed speed according to the tool point thickness, and generating numerical control codes; and a visual guidance module, which is used for calculating a polishing penetration risk coefficient and generating a three-dimensional color heat map. The application solves the problems of easy vibration and collapse of the extremely thin area under the wet cutting condition and the strength reduction of the base caused by artificial blind polishing, and significantly improves the breakage resistance load and processing yield of the finished product.

Description

Technical Field

[0001] This invention relates to the field of digital dental manufacturing technology, specifically to a personalized intelligent denture design system and method based on three-dimensional scanning. Background Technology

[0002] With the widespread application of computer-aided design and manufacturing technology in the field of oral healthcare, the production of complete resin dentures has gradually transitioned from traditional purely manual production to digital semi-automatic production. The current routine processing flow typically includes: the dentist obtains an impression of the edentulous jaw and sends it to the machining center; the technician manually marks key anatomical features such as marginal closure areas and vibration lines on a plaster model and performs a 3D scan; the design software completes the artificial tooth arrangement and base contour design based on the scan data, generating a finished digital model for cutting; subsequently, wet cutting is performed using a CNC machine tool with coolant flushing; after the machine tool cutting is completed, the technician manually adjusts and polishes the surface to remove tool marks before delivering the finished product.

[0003] Although the aforementioned process based on CNC wet cutting and entirely manual grinding is relatively mature, significant processing challenges still exist in actual production workshops. In the CNC cutting stage, existing cutting codes mostly employ a constant feed rate. When the tool rotates at high speed and cuts into extremely thin areas such as the edge-closed region or the free end of the base material as defined by the technician, due to the weak rigidity of the local material structure, high-frequency thin-wall vibrations easily occur in these extremely thin areas under the combined action of coolant flushing and the mechanical cutting force of the tool. This leads to micro-cracks or chipping at the resin edges, reducing the yield rate of the finished products.

[0004] During the later manual polishing stage, due to the complex topological undulations on the denture base surface, it is difficult for technicians to accurately estimate the actual remaining thickness of the material during physical polishing. Relying solely on the technician's subjective experience for measurement can easily lead to over-polishing in areas of concentrated chewing force or at extremely thin, free ends. Such blind adjustments can result in local cross-sectional thicknesses falling below the material's safe mechanical limits, making the final complete denture prone to stress concentration and fracture under the patient's actual chewing load, severely shortening the denture's clinical lifespan.

[0005] Therefore, this invention proposes a personalized denture intelligent design system and method based on three-dimensional scanning to address the shortcomings of existing technologies. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a personalized intelligent denture design system and method based on three-dimensional scanning. This solves the problems of existing complete resin dentures, where CNC wet cutting of extremely thin areas easily causes vibration and chipping during processing, and manual blind polishing easily leads to excessively thin base areas that can cause breakage under stress.

[0007] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of the present invention provides a personalized intelligent denture design system based on three-dimensional scanning, comprising: The feature parsing and detection module is used to receive 3D scan data of edentulous jaws with solid line textures, extract anatomical feature boundaries and construct a 3D mesh model of the basement, and calculate the local normal thickness of the mesh vertices. The mechanical adaptive strengthening module is used to extract chewing force nodes by combining design layout data, calculate dynamic required thickness based on the local normal thickness, perform internal adaptive thickening compensation on the weak areas of the base three-dimensional mesh model, and generate a standard processing model. The anti-vibration toolpath optimization module is used to plan the CNC cutting toolpath according to the standard machining model. During the path generation process, the feed rate of the extremely thin region is dynamically reduced according to the local normal thickness of the corresponding tool position point to generate CNC machining code. The visualization guidance module is used to calculate the grinding penetration risk coefficient of the surface mesh vertices based on the local normal thickness after the cutting of the standard machining model, and map it to the color space to generate a three-dimensional color heat map.

[0008] Preferably, the feature parsing and detection module extracts the anatomical feature boundaries and calculates the local normal thickness of the mesh vertices, specifically including: The RGB color values ​​of the nodes in the edentulous 3D scan data are converted to the HSV color space. The target node set is extracted using the set hue and saturation threshold range. The spatial curve fitting operation is performed on the target node set to generate the anatomical feature boundary. Calculate the outward unit normal vector of the mesh vertex, emit a computational ray in the opposite direction of the outward unit normal vector and detect the first intersection point, and calculate the straight-line Euclidean distance between the mesh vertex and the first intersection point as the local normal thickness.

[0009] Preferably, the outward unit normal vector of the computational mesh vertex specifically includes: Extract the set of first-order adjacent triangular faces that share mesh edges with the mesh vertices. Use the geometric area of ​​each first-order adjacent triangular face as a weight distribution factor to calculate the weighted normal and perform normalization to obtain the outward unit normal vector.

[0010] Preferably, the mechanical adaptive strengthening module extracts chewing force nodes based on design layout data and calculates the dynamic required thickness based on the local normal thickness, specifically including: Analyze the design layout data and extract the functional cusp apex of the artificial posterior tooth area as the chewing force node; Based on the spatial Euclidean distance from the grid vertex to each chewing force node, an exponential decay model is constructed using the set basic safety section thickness, stress sensitivity coefficient, and stress influence attenuation radius to calculate the dynamic demand thickness corresponding to the grid vertex. When the local normal thickness is less than the dynamic required thickness, the corresponding area is determined to be the weak area.

[0011] Preferably, after the dynamic required thickness corresponding to the vertex of the computational mesh, the method further includes: A maximum allowable thickness threshold is set. When the calculated dynamic requirement thickness is greater than the maximum allowable thickness threshold, the dynamic requirement thickness is forcibly truncated to the maximum allowable thickness threshold.

[0012] Preferably, the mechanical adaptive strengthening module performs internal adaptive thickening compensation on the weak areas of the base 3D mesh model, specifically including: Calculate the thickness difference between the dynamic required thickness and the local normal thickness, and generate an initial compensation displacement vector in the opposite direction of the outward unit normal vector corresponding to the mesh vertex; The initial compensation displacement vector is subjected to iterative smoothing operation using a spatial Gaussian smoothing filter algorithm. During the filtering operation, the displacement weight of the mesh vertices associated with the anatomical feature boundary is forced to be zero. After the iterative smoothing operation converges, the smoothed compensation displacement vector is superimposed onto the spatial coordinates of the corresponding mesh vertex.

[0013] Preferably, the anti-vibration toolpath optimization module dynamically reduces the feed rate of the extremely thin region based on the local normal thickness at the corresponding tool position point, specifically including: Set a safety thickness threshold and a rigidity feedback index; When the local normal thickness at the corresponding tool position is less than the safe thickness threshold, the dynamic feed rate is calculated based on the ratio of the local normal thickness to the safe thickness threshold, combined with the rigid feedback index. The cutting feed rate at the corresponding tool position point is updated to the dynamic feed rate.

[0014] Preferably, after calculating the dynamic feed rate, the method further includes: The lower limit safety threshold is set based on the product of the minimum allowable feed per tooth of the selected cutting tool, the number of tool teeth, and the spindle speed. When the calculated dynamic feed rate is lower than the lower safety threshold, the dynamic feed rate is forcibly truncated to the lower safety threshold.

[0015] Preferably, the visualization guidance module calculates the grinding penetration risk coefficient of the surface mesh vertices and maps it to a color space to generate a three-dimensional color heatmap, specifically including: Set the minimum and safe thickness benchmarks for grinding; Calculate the difference between the safe thickness reference and the local normal thickness, and the difference between the safe thickness reference and the grinding limit thickness. Use the ratio of the two and the upper and lower limit cutoff functions to calculate the normalized grinding penetration risk coefficient. A linear mapping relationship is established between the grinding penetration risk coefficient and the RGB color channel values. The converted color data is then bound to the corresponding mesh vertices of the standard processing model to generate the three-dimensional color heatmap.

[0016] A personalized denture intelligent design method based on 3D scanning includes the following steps: Receive 3D scan data of edentulous jaws with solid line textures, extract anatomical feature boundaries and construct a 3D mesh model of the denture base, and calculate the local normal thickness of the mesh vertices; By combining the design layout data to extract the chewing force nodes, calculating the dynamic required thickness based on the local normal thickness, performing internal adaptive thickening compensation on the weak areas of the base three-dimensional mesh model, and generating a standard processing model; The CNC cutting tool path is planned according to the standard machining model. During the path generation process, the feed rate of the extremely thin region is dynamically reduced according to the local normal thickness of the corresponding tool position point to generate CNC machining code. Based on the local normal thickness after the cutting of the standard machining model, the grinding penetration risk coefficient of the surface mesh vertices is calculated, and it is mapped to the color space to generate a three-dimensional color heat map.

[0017] This invention provides a personalized intelligent denture design system and method based on 3D scanning. It has the following beneficial effects: 1. This invention calculates the dynamic required thickness of the denture base by extracting the chewing force nodes and constructing an exponential decay model. It performs adaptive thickening compensation for weak internal areas and forces the displacement weight of anatomical feature boundaries to be zero in the filtering operation. This design increases the cross-sectional thickness of the stress concentration area to improve the fracture resistance of the denture base while maintaining the spatial coordinates of key feature boundaries such as the edge closure area, ensuring the clinical fit accuracy between the complete denture and the patient's oral mucosa.

[0018] 2. This invention introduces a nonlinear dynamic control mechanism during the CNC toolpath planning stage. It dynamically adjusts the feed rate based on the local normal thickness corresponding to the tool position point and configures a lower limit safety cutoff threshold based on tool geometry parameters. This underlying intervention strategy can dynamically regulate the cutting force based on the local stiffness of the material, preventing thin-wall chatter when machining the extremely thin free end of the base, reducing edge chipping, and maintaining normal chip breaking and removal physical conditions, thus improving the machining yield of CNC wet cutting processes.

[0019] 3. This invention utilizes the local normal thickness of the processing model and a set grinding limit benchmark to calculate the grinding penetration risk coefficient of the surface mesh vertices, and maps it into a three-dimensional color heatmap for output to the display terminal. This mapping mechanism transforms the solid thickness parameter of the resin material into an intuitive color code, providing a quantitative visual reference boundary for technicians to manually adjust the process, avoiding over-grinding of local structures and product scrapping due to human experience bias. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the system framework of the present invention; Figure 2 This is a schematic diagram of the method flow of the present invention; Figure 3 This is a schematic diagram of the data processing principle of the feature analysis and detection module of the present invention; Figure 4 This is a schematic diagram of the data processing principle of the mechanical adaptive reinforcement module of the present invention; Figure 5 This is a schematic diagram of the data processing principle of the anti-vibration toolpath optimization module of the present invention; Figure 6 This is a data processing principle diagram of the visualization guidance module of the present invention; Figure 7 This is a graph showing the dynamic response relationship between local thickness and feed rate under wet cutting conditions according to the present invention.

[0021] Among them, 10 is the feature analysis and detection module; 20 is the mechanical adaptive reinforcement module; 30 is the anti-vibration tool rail optimization module; and 40 is the visualization guidance module. Detailed Implementation

[0022] The technical solutions in 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.

[0023] See attached document Figure 1This invention provides a personalized denture intelligent collaborative manufacturing system based on feature line recognition, comprising: The feature analysis and detection module 10 is used to receive edentulous jaw 3D scan data with texture of physical lines drawn by technicians, identify and extract anatomical feature boundaries, construct a 3D mesh model of the base, and calculate the local normal thickness of the mesh vertices. The mechanical adaptive strengthening module 20 is used to extract stress nodes by combining the design layout data of integrated cutting or split cutting, calculate the dynamic required thickness based on the local normal thickness, perform internal adaptive thickening compensation on weak areas in the base three-dimensional mesh model that have the risk of fracture, and generate a standard machining model. The anti-vibration toolpath optimization module 30 is used to plan the CNC cutting toolpath according to the standard machining model. During the path generation process, the feed rate of the extremely thin area is dynamically reduced according to the local normal thickness of the corresponding tool position point to generate anti-vibration CNC machining code. The visualization guidance module 40 is used to calculate the grinding penetration risk coefficient of the surface mesh vertices based on the local normal thickness after the standard machining model is cut, and to map it to the color space to generate a three-dimensional color heat map.

[0024] The personalized dental prosthesis intelligent collaborative manufacturing system is deployed within a computer terminal equipped with graphics processing and data computing capabilities. The front end establishes data communication with 3D scanning equipment and layout design software, while the back end establishes command transmission connections with CNC machine tool controllers and workshop display equipment. During operation, these system modules execute data processing and workflow actions sequentially over time.

[0025] See attached document Figure 2 This invention provides a personalized denture intelligent collaborative manufacturing method based on feature line recognition, comprising the following steps: S100 receives 3D scan data of edentulous jaws with textured lines drawn by technicians, identifies and extracts anatomical feature boundaries, constructs a 3D mesh model of the denture base, and calculates the local normal thickness of the mesh vertices. S200 extracts stress nodes by combining the design layout data of integrated cutting or split cutting, calculates the dynamic required thickness based on the local normal thickness, performs internal adaptive thickening compensation on weak areas in the base three-dimensional mesh model that have the risk of fracture, and generates a standard machining model. S300 plans the CNC cutting tool path according to the standard machining model. During the path generation process, the feed rate of the extremely thin area is dynamically reduced according to the local normal thickness of the corresponding tool position point to generate anti-chatter CNC machining code. S400 calculates the grinding penetration risk coefficient of the surface mesh vertices based on the local normal thickness after the standard machining model is cut, and maps it to the color space to generate a three-dimensional color heat map.

[0026] To further clarify the implementation of each technical aspect of the present invention, the following will provide a detailed description of the implementation of each functional module involved above and its internal processing flow.

[0027] See attached document Figure 3 , Figure 3 This is a schematic diagram of the data processing principle of a feature parsing and detection module according to an embodiment of the present invention.

[0028] In this embodiment, the personalized denture intelligent collaborative manufacturing system based on feature line recognition provided by the present invention converts the original scanned image containing traces of technician experience into a geometric thickness reference that can be processed by a computer through the feature analysis and detection module 10. The data processing in this stage specifically includes the following steps: S101, acquire the digital anatomical feature boundary of the physical line drawing features. The feature analysis and detection module 10 receives 3D scan data of the edentulous jaw with the texture of the physical line drawing by the technician. This scan data contains the 3D spatial coordinate system nodes of the model surface and the corresponding color channel information. In conventional dental processing, when technicians mark structures such as edge closure areas and vibration lines, they usually use marker pens with high color contrast, which makes the target features significantly different from the plaster base in color data. In order to accurately extract these artificial experience lines, the feature analysis and detection module 10 converts the RGB color values ​​of the extracted 3D scan data nodes to the HSV color space and filters the mesh nodes using a set hue and saturation threshold range. As a specific implementation, this color threshold range can be pre-calibrated according to the standard color spectrum of the marker pen used in actual work.

[0029] The system removes mesh nodes belonging to the plaster base color, retaining target nodes within the marked color threshold range. The feature parsing and detection module 10 performs spatial curve fitting operations on the extracted target node set, generating continuous anatomical feature boundary lines in three-dimensional space. These anatomical feature boundary lines provide a clear reference for mesh physical segmentation and constraints for subsequent calculations. For the processing of separated discrete noise points and broken line connections, those skilled in the art can use conventional neighborhood interpolation smoothing algorithms for repair calculations. The specific repair logic is well-known in the field and will not be elaborated here.

[0030] S102, calculate the outward unit normal vector of the vertices of the 3D mesh model. After establishing the anatomical feature boundary, it is necessary to clarify the geometric orientation of each node on the model surface. Considering that the 3D mesh formed by topological reconstruction of the original 3D scanning data is prone to local micro-jagged edges or rough surfaces, directly using the normal of a single triangular facet as the vertex normal can easily cause vector mutations. To eliminate the calculation error caused by mesh noise, the feature analysis and detection module 10 uses an area-weighted average algorithm to calculate the normal orientation of discrete mesh vertices. The feature analysis and detection module 10 traverses each mesh vertex in the model and extracts the set of first-order adjacent triangular facets that share a mesh edge with the current mesh vertex. The system uses the physical area of ​​each adjacent triangular facet as a weight distribution factor to calculate its weighted normal and perform normalization processing. The specific calculation formula is as follows: ; In the formula, For grid vertices The outward unit normal vector; To be related to the grid vertex The total number of adjacent first-order triangular faces; The index value of the first-order adjacent triangle; For the first The geometric area of ​​a first-order adjacent triangular facet; For the first The unit normal vector of each first-order adjacent triangular facet; Mathematical operators for modulo operations on vectors S103, the local normal thickness of the mesh vertices is calculated using a ray-mapping mechanism. After obtaining the outward unit normal vector of the mesh vertices, the feature analysis detection module 10 quantifies the physical thickness of the base structure. The feature analysis detection module 10 uses the mesh vertices... Taking the three-dimensional spatial coordinates as the geometric starting point, and the outward unit normal vector... The opposite direction is taken as the emission direction of the ray. From a physical structure perspective, vector − The feature analysis and detection module 10 emits a spatial computation ray within the model coordinate system, traverses the triangular facets of the inner wall of the 3D mesh model, and detects the first intersection point where the computation ray and the inner wall facet collide in space.

[0031] The system records the 3D coordinates of the intersection point and calculates the mesh vertices. The Euclidean distance between the geometric starting point coordinates and the first intersection point coordinates is calculated. The feature analysis and detection module 10 assigns this Euclidean distance value to the corresponding mesh vertex. The local normal thickness is denoted as To address potential topological incompleteness or voids in the scanned model due to optical path obstruction, the feature analysis and detection module 10 configures a maximum detection distance threshold for the spatial computational ray. This value is typically defined in the range of 15mm to 20mm, taking into account the maximum physical thickness of complete dentures. When the computational ray extends beyond the maximum detection distance threshold without capturing a collision intersection, the system performs truncation and assigns the maximum detection distance threshold to the corresponding mesh vertex as a replacement thickness, thus preventing the ray detection from falling into an algorithmic dead zone of infinite tracking.

[0032] The feature analysis and detection module 10 traverses the discrete mesh vertices on the outer surface of the base 3D mesh model, repeatedly performing ray intersection detection and Euclidean distance calculation. After the above data detection process, the system integrates the spatial 3D coordinate data of all surface mesh vertices, associated anatomical feature boundary labels, and local normal thickness data to construct a basic structural feature matrix. This matrix is ​​transmitted to the subsequent mechanical optimization calculation layer via the data bus. For anomalies caused by multiple penetrations of computational rays in some complex concave regions, those skilled in the art can use a reverse normal projection comparison algorithm to screen for true inner wall intersections. The detection and correction logic is a well-known technique in this field and will not be elaborated here.

[0033] See attached document Figure 4 , Figure 4 This is a data processing principle diagram of a mechanical adaptive reinforcement module according to an embodiment of the present invention.

[0034] In this embodiment, the personalized denture intelligent collaborative manufacturing system based on feature line recognition provided by the present invention intelligently strengthens the fracture-prone area of ​​the denture base internally through the mechanical adaptive strengthening module 20. Under the processing boundary where the global polishing allowance setting is cancelled, the data processing in this stage mainly solves the physical balance problem between the internal structural strength of the denture base and the external edge fit, specifically including the following steps: S201, analyzes the layout mesh data to locate the nodes of the chewing load distribution. Before the actual cutting and machining of the complete denture, the mechanical adaptive strengthening module 20 receives the integrated or split cutting layout mesh data output by the front-end design software. Since the occlusal force distribution is not uniform when the complete denture performs its chewing function, mechanical stress is usually concentrated in the posterior tooth area, so this area needs to be accurately located. The mechanical adaptive strengthening module 20 analyzes the geometric features of the artificial tooth arrangement in the layout mesh data and extracts the spatial coordinates of the functional cusps of the artificial posterior teeth.

[0035] The system defines the extracted set of functional cusp vertices as the nodes with the maximum chewing force, using them as the load boundary reference for subsequent mechanical compensation calculations. For the automated search and extraction of cusp feature vertices in the mesh model, those skilled in the art can employ conventional curvature extremum detection algorithms; the specific extraction logic is well-known in the field and will not be elaborated here. To prevent algorithm failure due to abnormal front-end tooth arrangement data or missing teeth, the system establishes a fault-tolerant mechanism: if no functional cusp vertices meeting the feature threshold are detected in the model, the entire alveolar ridge crest line is extracted as a substitute set of force-bearing node reference points, avoiding subsequent calculations from falling into dead zones.

[0036] S202, an exponential decay model is constructed to calculate the dynamic required thickness of the baseplate mesh vertices. After obtaining the node with the maximum chewing force, the mechanical adaptive strengthening module 20 quantitatively evaluates the baseplate structural thickness required to resist mechanical bending moments. From the perspective of the solid mechanical properties of polymer resin materials, the transmission of chewing force inside the baseplate usually exhibits a nonlinear decay state; the closer the baseplate area is to the center of force, the greater the bending moment and shear force it bears, and correspondingly, a larger cross-sectional thickness is required to provide flexural strength. Based on the above physical mechanism, the mechanical adaptive strengthening module 20 calculates the spatial Euclidean distance from each mesh vertex of the baseplate to the extracted node with the maximum chewing force, and uses this distance to construct a parameterized nonlinear decay equation. The specific calculation formula is as follows: ; In the formula, For grid vertices Required local dynamic thickness; The basic safety section thickness of the resin material is determined in advance based on the tensile fracture strength of the selected cutting resin block material, and the conventional setting range is 1.5mm to 2.0mm. This represents the total number of nodes with the greatest chewing force extracted. This is the index value of the node experiencing the greatest chewing force. It is a dimensionless stress sensitivity coefficient used to adjust the gain of thickness compensation. As a preferred method, its value range is set to 0.5 to 1.0. For grid vertices To the The three-dimensional Euclidean distance between the nodes with the greatest chewing force; The set stress influence attenuation radius determines the range of load diffusion to the surrounding base, and is usually set to 10mm to 15mm based on the average anatomical span of the edentulous model.

[0037] To prevent the stress attenuation values ​​of multiple nodes from abnormally accumulating and causing the calculated local dynamic required thickness to be too large and encroach on the effective space of the patient's oral cavity, the system introduces a maximum allowable thickness threshold into the calculation results of the above equations. Perform a safe truncation. The specific truncation judgment logic is as follows: Considering the limitations of adult oral physiological structure, The value range is generally set to 3.5mm to 4.5mm. The mechanical adaptive strengthening module 20 traverses all mesh vertices on the base 3D mesh model, calculates and outputs the dynamic required thickness value corresponding to the vertex based on the exponential decay model.

[0038] S203, execute adaptive internal thickening compensation to maintain zero displacement of anatomical feature boundaries. The mechanical adaptive strengthening module 20 retrieves the previously generated thickness feature matrix and compares the local normal thickness of each mesh vertex with its corresponding dynamic required thickness. When the local normal thickness of a mesh vertex is detected to be less than its dynamic required thickness, the system determines that the denture base area is at risk of fracture under stress. The mechanical adaptive strengthening module 20 calculates the thickness difference between the two and uses this thickness difference as the displacement compensation amount for that vertex. To avoid changing the external shape of the denture through outward thickening, which could lead to clinical tooth arrangement interference, the system generates an initial compensation displacement vector in the opposite direction of the outward unit normal vector (i.e., the internal direction of the solid). The mathematical expression for this vector is: ,in For grid vertices Local normal thickness, Let it be its corresponding outward unit normal vector. It is its corresponding outward unit normal vector.

[0039] Directly applying discrete displacement vectors to mesh vertices can induce topological steps on the internal surface of the model, increasing the foreign body sensation experienced by the patient. The mechanical adaptive reinforcement module 20 employs a spatial Gaussian smoothing filter algorithm to perform iterative smoothing operations on the initial compensation displacement vectors. During the filtering process, any coordinate offset at the edge positions will compromise the clinical fit accuracy between the denture tissue surface and the patient's oral mucosa. To ensure edge fit, the mechanical adaptive reinforcement module 20 establishes a rigid body constraint mechanism, forcing the displacement weights of mesh vertices associated with the previously extracted anatomical feature boundaries to be zero. This constraint ensures that key anatomical boundaries such as edge closure areas and vibration lines remain stationary before and after the thickening operation. After the iterative smoothing operation reaches convergence, the mechanical adaptive reinforcement module 20 superimposes the smoothed compensation displacement vectors onto the original spatial coordinates of the corresponding mesh vertices, completing the local geometric thickening of the weak areas of the denture base inwards, and outputting a standard fabrication model with mechanical flexural strength that ensures a good fit.

[0040] See attached document Figure 5 , Figure 5 This is a data processing principle diagram of an anti-vibration toolpath optimization module according to an embodiment of the present invention.

[0041] In this embodiment, the personalized denture intelligent collaborative manufacturing system based on feature line recognition provided by the present invention performs low-level code intervention for wet cutting conditions through the anti-vibration toolpath optimization module 30. The data processing at this stage mainly aims to reduce the risk of chipping in extremely thin areas under the combined action of water flow impact and tool cutting force, and specifically includes the following steps.

[0042] S301, spatial mapping of local normal thickness data at the tool position point is performed. The anti-chatter toolpath optimization module 30 plans the CNC cutting toolpath based on the geometric features of the outer surface of the standard machining model. For conventional tool interference checks and basic cutting trajectory planning in computer-aided manufacturing, those skilled in the art can use existing spatial trajectory generation algorithms to implement them. The path generation logic is a well-known technology in this field and will not be elaborated here. Based on the generation of the basic cutting trajectory, the anti-chatter toolpath optimization module 30 discretizes the continuous toolpath to generate a spatial discrete tool position point sequence containing three-dimensional coordinate information. To obtain the deep structural stress information of the material, the system uses a spatial nearest point search algorithm (such as the KD tree algorithm) to find the grid vertex on the outer surface of the standard machining model that is closest to the current tool position point and extracts the local normal thickness of the grid vertex. Through the above spatial position correspondence calculation, the system reads and binds the local normal thickness parameters corresponding to the tool position point sequence one by one.

[0043] As a preferred approach, to prevent matching failures due to discretization accuracy errors in the mesh model, the system sets a maximum search radius tolerance. When no valid mesh vertex can be matched within the defined radius around the tool position, the system assigns a preset safety thickness value to that tool position as a fault tolerance, avoiding dead zones in the underlying calculations caused by missing data. This preset safety thickness value is typically set with reference to the safety cross-sectional thickness of the resin material to be processed, and its conventional range is 1.5mm to 2.5mm.

[0044] S302, constructs a nonlinear dynamic deceleration control equation to calculate the dynamic feed rate. After obtaining the local normal thickness corresponding to the discrete tool position, the anti-vibration toolpath optimization module 30 dynamically adjusts the cutting parameters. In the CNC wet cutting environment, the physical scouring of the high-pressure coolant sprayed by the machine tool and the mechanical shearing force generated by the high-speed rotation of the tool are superimposed, causing extremely thin areas such as the closed area of ​​the base edge to easily experience high-frequency vibration due to insufficient structural rigidity, which in turn leads to material fracture and breakage.

[0045] Cutting physics principles indicate that the local thickness of a material directly determines its bending stiffness against cutting forces; the thinner the region, the smaller the allowable cutting force it can withstand. To address this machining hazard, the anti-chatter toolpath optimization module 30 extracts the local normal thickness corresponding to the current discrete tool position and constructs a nonlinear dynamic control equation using a set safety thickness threshold and a rigidity feedback index. When the local normal thickness is greater than or equal to the safety thickness threshold, the system determines that the structure in that region is stable, and the feed rate at the corresponding tool position is maintained at the set machine tool reference cutting feed rate. The system dynamically reduces the feed rate based on the proportion of local normal thickness below the safety thickness threshold. The specific mathematical expression of the dynamic control equation is as follows: when At that time, the system calculates the dynamic feed rate at the current tool position point according to the following formula: ; In the formula, This is the feed rate output after dynamic adjustment of the current discrete tool position; The machine tool reference cutting feed rate set for the equipment; The local normal thickness is extracted from the spatial mapping of the current discrete tool position. The safety thickness threshold is determined by combining the critical yield strength and shear resistance of the resin material being processed, and its conventional value range is set between 1.5 mm and 2.5 mm. It is a dimensionless rigid feedback exponent used to control the nonlinear curvature of the deceleration process.

[0046] To ensure that the feed rate decreases at a steeper curve when the local thickness is reduced, The value of is typically set as a constant parameter greater than 1, with a preferred range of 1.5 to 2.0. Using this nonlinear dynamic control equation, the system can achieve targeted deceleration intervention based on physical rigidity feedback when the tool approaches the extremely thin free end of the base.

[0047] S303 executes a lower limit cutoff constraint on the feed rate to maintain chip breaking and chip removal. A mismatched feed rate alters the original cutting mechanism of the CNC tool. When the local feed rate is too low, the cutting edge of the tool will experience frictional compression with the resin surface instead of normal physical shearing, leading to microscopic tool deflection, affecting the surface finish of the cut surface and preventing normal chip removal. To maintain basic physical chip breaking, the anti-chatter toolpath optimization module 30 introduces a lower limit safety cutoff constraint mechanism during the feed rate calculation process.

[0048] As a specific implementation, the anti-chatter toolpath optimization module 30 sets the product of the minimum allowable feed per tooth, the number of tool teeth, and the spindle speed as a lower safety threshold based on the geometric parameters of the selected cutting tool. During path planning calculations, when the dynamic feed rate calculated according to the aforementioned control equations is lower than this lower safety threshold, the anti-chatter toolpath optimization module 30 forcibly truncates the feed rate to the lower safety threshold. The logical expression of this forced truncation constraint is as follows: ; In the formula, This is the set lower safety threshold. The dynamic feed rate value, after being verified and truncated by this lower threshold, is jointly encapsulated with its corresponding tool position spatial coordinate data to generate CNC machining code containing specific physical protection mechanisms. This CNC machining code is sent to the CNC wet cutting machine controller through a data communication interface to drive the milling of the base entity in a controlled working state.

[0049] See attached document Figure 6 , Figure 6 This is a data processing principle diagram of a visualization guidance module according to an embodiment of the present invention.

[0050] In this embodiment, the personalized denture intelligent collaborative manufacturing system based on feature line recognition provided by the present invention performs visual mapping and transformation of parameters for the entirely manual grinding process through the visualization guidance module 40. The data processing in this stage mainly aims to eliminate the risk of localized excessive thinning and subsequent breakage due to blind grinding by technicians, and specifically includes the following steps: S401 calculates the grinding penetration risk coefficient of surface mesh vertices. After the CNC machine tool completes solid cutting, the resin blank usually needs to be manually adjusted and ground to remove surface cutting marks. Due to the topological undulations on the base surface, it is difficult for technicians to estimate the actual remaining thickness of the material inside during physical operations. To assess the impact of the grinding operation on the structural strength, the visualization guidance module 40 retrieves the local normal thickness of each surface mesh vertex of the standard machining model.

[0051] The system sets the maximum grinding thickness based on the physical fracture limit of the resin material and establishes a safety thickness benchmark. As a preferred approach, the maximum grinding thickness, combined with the minimum structural tolerance of complete dentures, is set to 0.5mm to 0.8mm, while the safety thickness benchmark, combined with the material's standard safe cross-sectional thickness, is set to 1.5mm to 2.0mm. The visualization guidance module 40 uses mathematical interpolation to calculate a dimensionless grinding penetration risk coefficient for each surface mesh vertex. The specific calculation model is expressed as follows: ; In the formula, For surface mesh vertices The risk factor of grinding through; For surface mesh vertices The corresponding local normal thickness; As a safety thickness benchmark; The set grinding limit thickness. From the perspective of algorithm data processing principles, through the normalization function with upper and lower limit truncation mechanisms mentioned above, the thickness variable is mapped to a continuous numerical range between zero and one. When the local normal thickness is greater than or equal to the safe thickness benchmark, the risk coefficient is calculated to be zero, indicating that the area has sufficient grinding space; when the local normal thickness decreases and approaches the grinding limit thickness, the risk coefficient approaches or equals one, indicating that there is a risk of mechanical penetration in the area.

[0052] S402, map and convert the 3D color heatmap and execute the terminal output. After obtaining the risk quantification index of discrete grid vertices, the visualization guidance module 40 establishes a corresponding conversion function between the risk coefficient and the RGB three-channel color values. The calculated grinding penetration risk coefficient is converted into a numerical color code, providing operators with an intuitive division of area status. As a specific implementation method, the system's color space mapping equation is as follows: ; ; ; In the formula, , and Representing the vertices of the surface mesh The color component values ​​in the red, green, and blue channels. In human visual perception, the red-green color scheme is often used to represent warning levels of danger and safety. Based on the aforementioned linear continuous mapping function, when the risk coefficient... When the value is high, the vertex is assigned a high red component; when When the value is low, the vertex is assigned a high green component. And... When the values ​​are in the middle range, the red and green channel values ​​are mixed, resulting in a gradual transition zone of yellow or orange on the surface of the 3D model.

[0053] The visualization guidance module 40 binds the converted color data to the corresponding mesh vertices of the standard processing model, generating a 3D color heatmap. The system transmits this 3D color heatmap to the display terminal for rendering output. In human-machine collaborative manufacturing scenarios, technicians can perform differentiated operations based on the distribution of the 3D color heatmap on the screen. For example, normal de-massaging operations are performed on the green rendering area, while operations are avoided on the red rendering area or only a soft polishing wheel is used for surface finishing. This heatmap provides a quantitative reference for the physical thickness of the material during manual grinding, which helps reduce over-adjustment caused by human experience bias.

[0054] To help those skilled in the art better understand the technical solution of the present invention and its actual engineering effects, a complete specific application embodiment is provided below, and the solution of the present invention is compared and analyzed with traditional solutions in conjunction with experimental verification data.

[0055] Specific application examples: Reference Appendix Figure 7 In this embodiment, taking the processing of a complete resin denture for a 72-year-old maxillary edentulous patient as an example, a mainstream highly cross-linked polymethyl methacrylate (PMMA) resin disc is selected as the cutting blank, and a five-axis CNC wet cutting machine with a water-cooled spray system is used for processing.

[0056] Factory technicians used a high-contrast blue marker to draw the posterior maxillary fremitus and the closed area on the labial and buccal margins of the plaster model. The system received the 3D scan data with blue texture, converted the RGB color space to HSV space, and successfully extracted continuous anatomical feature boundary lines by setting a blue threshold range. Subsequently, the system calculated the mesh normal vector and emitted internal rays, measuring the extremely thin local normal thickness of the model at the labial free end of the base, with the thinnest point being only 0.75 mm.

[0057] The system analyzes the integrated cutting and layout data input from the front end, extracting the central fossa and functional cusps of the first and second molars on both sides as the nodes of maximum masticatory force. The basic safety thickness of PMMA is then set. =1.8mm attenuation radius =12mm. Calculated using the exponential decay model, the dynamic required thickness for the midline region of the maxillary palatal flap reaches 2.3mm. While maintaining the displacement weight of the blue-lined area (edge-closed region) at zero, the system performs smooth positive compensation thickening inwards on the insufficiently thick palatal flap region, generating a standard fabrication model that balances flexural strength and edge fit.

[0058] The CNC programming software generates the basic toolpath and sets the machine tool reference feed rate. =2500mm / min. The system reads the thickness data at the tool position and sets the safe thickness threshold. =1.8mm, rigidity feedback index =1.5. When the CNC tool travels to the extremely thin edge closed area (thickness 0.75mm) on the lip side, the dynamic control equation automatically and non-linearly adjusts the feed rate of this micro-segment to approximately 673mm / min. At the same time, the system verifies that this value is higher than the set minimum chip-breaking feed limit for the tool (500mm / min), generates anti-vibration CNC machining code, and sends it to the wet cutting machine tool, achieving smooth cutting of the extremely thin area under the flushing of water.

[0059] After cutting, manual grinding begins. The system sets the maximum grinding thickness. =0.6mm, calculate the grinding penetration risk coefficient of all model mesh vertices, and map the result into an RGB 3D color heatmap for output to the technician's workstation screen. The technician observed on the screen that the labial edge area was dark red (high risk), while the palatal flap and alveolar ridge main area were green (safe zone). Based on this, the technician used a carbide grinding head for routine deburring of the green area, and only used a soft wool wheel for slight polishing of the red area, effectively avoiding over-grinding.

[0060] Experimental verification and effect comparison: Experimental conditions and group assignments: Traditional control group (5 pieces): Commercially available dental CAM software was used without adaptive thickness compensation; wet cutting was performed at a constant feed rate of 2500 mm / min throughout the process; after cutting, the technician performed conventional manual grinding and polishing based on subjective experience.

[0061] Experimental group of the present invention (5 pieces): The system of the present invention generates a standard machining model and anti-vibration CNC machining code (dynamic feed speed). After cutting, the same technician completes the grinding operation under the visual guidance of a three-dimensional color heat map.

[0062] Evaluation metrics and testing methods Edge micro-chipping rate: The percentage of pieces with micro-cracks or chipping is recorded by observing the extremely thin closed edge area after cutting under a stereomicroscope.

[0063] Tissue surface RMS fit error: The finished product was scanned using a high-precision blue light scanner, and the three-dimensional deviation analysis software was used to perform best-fit alignment with the original oral mucosa mesh to calculate the root mean square (RMS) error.

[0064] Ultimate fracture load: Apply a static compressive load to the molar area on both sides of the finished product on a universal testing machine and record the ultimate force value when the base breaks.

[0065] Grinding penetration failure rate: The final thickness of the extremely thin area after grinding is measured using vernier calipers. A thickness of less than 0.5mm is considered a penetration failure.

[0066] The experimental data are shown in Table 1: Table 1: Comparison of Processing Quality Evaluation Indicators between Traditional Processing Schemes and Embodiments of the Invention

[0067] in conclusion: Experimental data objectively demonstrate that the proposed solution is highly compatible with existing wet cutting and fully manual polishing production lines, effectively addressing the pain points inherent in traditional processing. By introducing a nonlinear dynamic deceleration control equation into the underlying code, this solution completely eliminates the edge micro-chipping phenomenon caused by water flow impact and mechanical cutting forces. Combined with an internal adaptive thickening mechanism that maintains zero boundary displacement and visual guidance for manual polishing based on a three-dimensional color heatmap, this solution significantly increases the ultimate fracture load of the finished product from the traditional 398N to 586N while completely preventing the ultrathin areas from being polished through. Furthermore, the RMS error of the tissue surface of the finished product stably converges to an extremely low level of 32.6μm, verifying that the proposed solution can significantly improve the mechanical strength of complete resin dentures while still strictly ensuring a high-precision clinical fit with the oral mucosa. This system possesses outstanding engineering practical value and yield improvement effects.

[0068] 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 personalized dental prosthesis intelligent design system based on 3D scanning, characterized in that, include: The feature parsing and detection module is used to receive 3D scan data of edentulous jaws with solid line textures, extract anatomical feature boundaries and construct a 3D mesh model of the basement, and calculate the local normal thickness of the mesh vertices. The mechanical adaptive strengthening module is used to extract chewing force nodes by combining design layout data, calculate dynamic required thickness based on the local normal thickness, perform internal adaptive thickening compensation on the weak areas of the base three-dimensional mesh model, and generate a standard processing model. The anti-vibration toolpath optimization module is used to plan the CNC cutting toolpath according to the standard machining model. During the path generation process, the feed rate of the extremely thin region is dynamically reduced according to the local normal thickness of the corresponding tool position point to generate CNC machining code. The visualization guidance module is used to calculate the grinding penetration risk coefficient of the surface mesh vertices based on the local normal thickness after the cutting of the standard machining model, and map it to the color space to generate a three-dimensional color heat map.

2. The personalized denture intelligent design system based on three-dimensional scanning according to claim 1, characterized in that, The feature parsing and detection module extracts anatomical feature boundaries and calculates the local normal thickness of mesh vertices, specifically including: The RGB color values ​​of the nodes in the edentulous 3D scan data are converted to the HSV color space. The target node set is extracted using the set hue and saturation threshold range. The spatial curve fitting operation is performed on the target node set to generate the anatomical feature boundary. Calculate the outward unit normal vector of the mesh vertex, emit a computational ray in the opposite direction of the outward unit normal vector and detect the first intersection point, and calculate the straight-line Euclidean distance between the mesh vertex and the first intersection point as the local normal thickness.

3. The personalized denture intelligent design system based on three-dimensional scanning according to claim 2, characterized in that, The outward unit normal vectors of the computational grid vertices specifically include: Extract the set of first-order adjacent triangular faces that share mesh edges with the mesh vertices. Use the geometric area of ​​each first-order adjacent triangular face as a weight distribution factor to calculate the weighted normal and perform normalization to obtain the outward unit normal vector.

4. The personalized denture intelligent design system based on three-dimensional scanning according to claim 1, characterized in that, The mechanical adaptive strengthening module extracts chewing force nodes based on design and layout data, and calculates the dynamic required thickness based on the local normal thickness, specifically including: Analyze the design layout data and extract the functional cusp apex of the artificial posterior tooth area as the chewing force node; Based on the spatial Euclidean distance from the grid vertex to each chewing force node, an exponential decay model is constructed using the set basic safety section thickness, stress sensitivity coefficient, and stress influence attenuation radius to calculate the dynamic demand thickness corresponding to the grid vertex. When the local normal thickness is less than the dynamic required thickness, the corresponding area is determined to be the weak area.

5. The personalized denture intelligent design system based on three-dimensional scanning according to claim 4, characterized in that, Following the dynamic required thickness corresponding to the vertex of the computational grid, the following is also included: A maximum allowable thickness threshold is set. When the calculated dynamic requirement thickness is greater than the maximum allowable thickness threshold, the dynamic requirement thickness is forcibly truncated to the maximum allowable thickness threshold.

6. The personalized denture intelligent design system based on three-dimensional scanning according to claim 1, characterized in that, The mechanical adaptive strengthening module performs internal adaptive thickening compensation on the weak areas of the 3D mesh model of the base, specifically including: Calculate the thickness difference between the dynamic required thickness and the local normal thickness, and generate an initial compensation displacement vector in the opposite direction of the outward unit normal vector corresponding to the mesh vertex; The initial compensation displacement vector is subjected to iterative smoothing operation using a spatial Gaussian smoothing filter algorithm. During the filtering operation, the displacement weight of the mesh vertices associated with the anatomical feature boundary is forced to be zero. After the iterative smoothing operation converges, the smoothed compensation displacement vector is superimposed onto the spatial coordinates of the corresponding mesh vertex.

7. The personalized denture intelligent design system based on three-dimensional scanning according to claim 1, characterized in that, The anti-vibration toolpath optimization module dynamically reduces the feed rate of the extremely thin region based on the local normal thickness at the corresponding tool position point, specifically including: Set the safety thickness threshold and rigidity feedback index; When the local normal thickness at the corresponding tool position is less than the safe thickness threshold, the dynamic feed rate is calculated based on the ratio of the local normal thickness to the safe thickness threshold, combined with the rigid feedback index. The cutting feed rate at the corresponding tool position point is updated to the dynamic feed rate.

8. The personalized denture intelligent design system based on three-dimensional scanning according to claim 1, characterized in that, After calculating the dynamic feed rate, the method further includes: The lower limit safety threshold is set based on the product of the minimum allowable feed per tooth of the selected cutting tool, the number of tool teeth, and the spindle speed. When the calculated dynamic feed rate is lower than the lower safety threshold, the dynamic feed rate is forcibly truncated to the lower safety threshold.

9. The personalized denture intelligent design system based on three-dimensional scanning according to claim 1, characterized in that, The visualization guidance module calculates the grinding penetration risk coefficient of the surface mesh vertices and maps it to a color space to generate a three-dimensional color heatmap, specifically including: Set the minimum and safe thickness benchmarks for grinding; Calculate the difference between the safe thickness reference and the local normal thickness, and the difference between the safe thickness reference and the grinding limit thickness. Use the ratio of the two and the upper and lower limit cutoff functions to calculate the normalized grinding penetration risk coefficient. A linear mapping relationship is established between the grinding penetration risk coefficient and the RGB color channel values. The converted color data is then bound to the corresponding mesh vertices of the standard processing model to generate the three-dimensional color heatmap.

10. A personalized denture intelligent design method based on three-dimensional scanning, applied to a personalized denture intelligent design system based on three-dimensional scanning as described in any one of claims 1-9, characterized in that, Includes the following steps: Receive 3D scan data of edentulous jaws with solid line textures, extract anatomical feature boundaries and construct a 3D mesh model of the denture base, and calculate the local normal thickness of the mesh vertices; By combining the design layout data to extract the chewing force nodes, calculating the dynamic required thickness based on the local normal thickness, performing internal adaptive thickening compensation on the weak areas of the base three-dimensional mesh model, and generating a standard processing model; The CNC cutting tool path is planned according to the standard machining model. During the path generation process, the feed rate of the extremely thin region is dynamically reduced according to the local normal thickness of the corresponding tool position point to generate CNC machining code. Based on the local normal thickness after the cutting of the standard machining model, the grinding penetration risk coefficient of the surface mesh vertices is calculated, and it is mapped to the color space to generate a three-dimensional color heat map.

Images