Jaw position registration method, device and equipment based on multi-dimensional intraoral scanning data

Abstract

The invention discloses a jaw position registration method, device and equipment based on multi-dimensional intraoral scanning data, and the method comprises the steps: reading upper and lower jaw scanning data and occlusion record data of a user, and obtaining an upper jaw dentition point cloud, a lower jaw dentition point cloud, a left buccal surface point cloud and a right buccal surface point cloud based on the upper and lower jaw scanning data and the occlusion record data; performing up-down left point cloud registration based on the maxillary dentition point cloud, the mandibular dentition point cloud and the left buccal surface point cloud, performing up-down right point cloud registration based on the maxillary dentition point cloud, the mandibular dentition point cloud and the right buccal surface point cloud, selecting an optimal registration path according to a left path and right path registration result, and obtaining a coarse registration model according to the optimal registration path; and iterating the pose of the degree of freedom of the coarse registration model to obtain a registration model, wherein the occlusion mass fraction Q of the registration model is maximum. The method solves the fundamental problem that the point clouds of the upper and lower jaws have no overlapping area and cannot be registered.

Description

Technical Field

[0001] This invention relates to the field of digital oral medicine, and in particular to a method, apparatus, and device for jaw registration based on multidimensional intraoral scanning data. Background Technology

[0002] Oral health issues are a major public health challenge faced globally. With an aging population and increasing patient demands for quality of life, the clinical need for restoring masticatory function and aesthetics through fixed prostheses (such as crowns and bridges) and implant restorations has grown significantly. The success of restorative treatment largely depends on establishing a precise occlusal relationship. A stable and harmonious occlusal relationship is not only crucial for ensuring the long-term survival of prostheses, patient comfort, and the health of the stomatognathic system, but also directly affects treatment efficiency. Therefore, how to accurately and efficiently achieve occlusal transfer and reconstruction remains a core issue in the field of prosthodontics.

[0003] Clinically, digital intraoral scanning technology, by capturing the surface morphology of oral soft and hard tissues such as teeth and gums in a high-precision, non-contact manner, has become a core driving force propelling oral medicine towards digitalization and precision. The current workflow for intraoral scanning technology is typically as follows: The dentist uses a digital intraoral scanner (IOS) to acquire the patient's three-dimensional data, including: maxillary dentition point clouds, mandibular dentition point clouds, and bilateral buccal point clouds in the occlusal position. After acquisition, the built-in software of commercial scanners (such as 3ShapeTRIOS) directly (or after processing with certain built-in algorithms) outputs the scanned three-dimensional point clouds, presenting the patient's digital jaw position.

[0004] The shortcomings of existing technology are: 1. "Black box" operation, lacking transparency and controllability: The dental reconstruction methods of commercial software are proprietary and not publicly disclosed. Users cannot know the internal workings, nor can they intervene or adjust when registration fails.

[0005] 2. Modeling often contains clinical errors: Because these "black box" algorithms lack constraints on oral biomechanics and anatomy, their automatically generated registration results frequently contain clinically unacceptable errors. The most common errors include: (1.) Model Interpenetration: This means that the tooth cusp "bites through" the tooth body or socket of the opposing tooth, which is physically impossible.

[0006] (2) Occlusal Gaps / Suspension: There is a noticeable and undesirable gap between the cusps and sockets that should be in close contact.

[0007] (3) Poor robustness: Existing algorithms suffer from a sharp decline in performance when faced with complex clinical situations. For example, when patients have malocclusion asymmetry or poor quality of buccal scan data on one side (such as incomplete scans or artifacts), these "black box" algorithms cannot intelligently process abnormal data, resulting in serious deviation or error in the final jaw position relationship.

[0008] In summary, existing technologies lack a transparent, robust, and clinically accurate automated jaw registration solution. Summary of the Invention

[0009] To address the aforementioned technical problems, this invention provides a method, apparatus, and device for jaw registration based on multidimensional intraoral scanning data, which can improve registration robustness.

[0010] A first aspect of the present invention provides a jaw registration method based on multidimensional intraoral scan data and geometric constraint optimization, comprising: Read the user’s maxillary and mandibular scan data and occlusion record data, and obtain the maxillary dentition point cloud, mandibular dentition point cloud, left buccal point cloud and right buccal point cloud based on the maxillary and mandibular scan data and occlusion record data; The optimal registration path is calculated based on the point cloud of the maxillary dentition, the point cloud of the mandibular dentition, the point cloud of the left buccal surface, and the point cloud of the right buccal surface, and a coarse registration model is obtained based on the optimal registration path. The registration model is obtained by iterating the pose of the coarse registration model, and the occlusion quality fraction Q of the registration model is maximized.

[0011] In one optional implementation, the step of registering upper and lower left point clouds based on the maxillary dentition point cloud, mandibular dentition point cloud, and left buccal point cloud, and registering upper and lower right point clouds based on the maxillary dentition point cloud, mandibular dentition point cloud, and right buccal point cloud, selecting the optimal registration path based on the registration results of the left and right paths, and obtaining a coarse registration model based on the optimal registration path; includes: ICP registration of the maxillary dentition point cloud, mandibular dentition point cloud and left buccal point cloud is calculated on the left path to obtain the first maxillary transformation matrix and the first mandibular transformation matrix. ICP registration of the maxillary dentition point cloud, mandibular dentition point cloud and right buccal point cloud is calculated on the right path to obtain the second maxillary transformation matrix and the second mandibular transformation matrix. The registration quality of the first maxillary transformation matrix and the first mandibular transformation matrix is ​​compared, and the one with higher registration quality is taken as the first initial transformation matrix. The registration quality of the second maxillary transformation matrix and the second mandibular transformation matrix is ​​compared, and the matrix with higher registration quality is used as the second initial transformation matrix.

[0012] In one optional implementation, obtaining the coarse registration model based on the optimal registration path includes: Compare the average root mean square error and / or average goodness of fit of the first initial transformation matrix and the second initial transformation matrix, and take the transformation matrix with the lowest average root mean square error and / or the highest average goodness of fit as the initial transformation matrix. The initial transformation matrix is ​​applied to the corresponding mandibular point cloud to obtain the coarse registration model after registration.

[0013] In one optional implementation, iteratively obtaining the registration model by altering the degrees of freedom pose of the coarse registration model includes: The key indicators of the coarse registration model are calculated, and the optimal optimization path for six-degree-of-freedom rigid body transformation is automatically selected from four strategies: translation priority, rotation priority, alternating optimization, and fine adjustment, based on the key indicators. The key indicators include centroid distance, contact density, normal alignment, and gap distribution pattern.

[0014] In one optional implementation, iteratively obtaining the registration model by altering the degrees of freedom pose of the coarse registration model includes: iterating the degrees of freedom pose of the coarse registration model using the following iterative strategy:

[0015] Where m1, m2, m3, m4, and m5 are adaptive adjustment parameters, and 2 ≤ m4. <m1<10,0<m2<1,m3<m5<1。

[0016] In one optional implementation, iteratively obtaining the registration model by altering the degrees of freedom pose of the coarse registration model includes: Calculate the overall quality score Q for each degree of freedom pose iteration. The registration model is obtained when the overall quality score Q is maximized, where the overall quality score is: ; Where U represents uniformity, S represents safety, Δtarget represents target deviation, and a, b, and c are weight parameters in the interval [0,1], with a+b+c=1.

[0017] N penetration It is the penetration point, S=1 indicates zero penetration; ;σ g and These are the standard deviation and mean of the gap, respectively. g target It is the target gap, g i C is the gap, C is the set of contact points, and Nc is the contact point.

[0018] In one alternative implementation, a=b=0.4, c=0.2.

[0019] A second aspect of the present invention provides a jaw registration device based on multidimensional intraoral scan data and geometric constraint optimization, comprising: The acquisition module is used to read the user's maxillary and mandibular scan data and occlusion record data, and to acquire maxillary dentition point cloud, mandibular dentition point cloud, left buccal point cloud and right buccal point cloud based on the maxillary and mandibular scan data and occlusion record data. The first registration module is used to calculate the optimal registration path based on the maxillary dentition point cloud, mandibular dentition point cloud, left buccal surface point cloud, and right buccal surface point cloud, and to obtain a coarse registration model based on the optimal registration path. The second registration module is used to iterate the pose of the coarse registration model to obtain a registration model, wherein the bite quality fraction Q of the registration model is maximized.

[0020] A third aspect of the present invention provides an electronic device comprising: At least one processor; and at least one memory communicatively connected to the processor, wherein the memory stores program instructions executable by the processor, and the processor invokes the program instructions to perform the method as described in the first aspect of the embodiments of the present invention.

[0021] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a computer, performs the method described in the first aspect of the embodiments of the present invention.

[0022] This invention calculates the optimal registration path based on the point clouds of the maxillary dentition, mandibular dentition, left buccal surface, and right buccal surface, and obtains a coarse registration model based on the optimal registration path. The pose of the coarse registration model is iterated to obtain a registration model. The occlusal quality fraction Q of the registration model maximizes, which solves the fundamental problem that the maxillary and mandibular point clouds cannot be registered due to the lack of overlapping areas.

[0023] This invention iterates through the pose of the coarse registration model to obtain a registration model, which can intelligently identify and avoid poor-quality unilateral buccal scan data, significantly improving the initial registration success rate and stability in cases of clinical asymmetric occlusion or poor scanning. Based on the comprehensive quality score Q, this invention can actively identify and eliminate common model penetration errors in existing "black box" technologies, and dynamically optimize the morphology of occlusal contact, making the cusp-fossa contact more uniform and stable, and the gap more in line with clinical requirements. Attached Figure Description

[0024] Figure 1This is a schematic diagram of the jaw registration method based on multidimensional intraoral scanning data and geometric constraint optimization in this invention.

[0025] Figure 2 This is a flowchart of the jaw registration method in one embodiment of the present invention.

[0026] Figure 3 This is a schematic diagram of a jaw registration device based on multidimensional intraoral scanning data and geometric constraint optimization in an embodiment of the present invention. Detailed Implementation

[0027] 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.

[0028] It should be noted that, in this document, relational terms such as "first" and "second" are used only 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 said element.

[0029] This invention uses any brand of intraoral scanner (e.g., 3Shape TRIOS 3) to collect patient data according to a standard procedure, the specific procedure of which is as follows: Patient oral cavity pretreatment: clean and dry the tooth surfaces; scan the maxillary and mandibular dental arches in sequence, and finally instruct the patient to gently bite down to the intercuspal position, and scan the buccal occlusal contact area of ​​the bilateral posterior teeth; each scan time is controlled within 5 minutes (900-1200 scan images) to ensure dynamic data accuracy; after the digital model of the upper and lower dentition is reconstructed in 3D, it is automatically aligned with the buccal scan and exported as a Polygon (PLY) file format.

[0030] This invention acquires four sets of point cloud data: maxillary dentition point cloud P U Mandibular dentition point cloud P L Left cheek dot cloud P B leftAnd right cheek dot cloud P B righ t.

[0031] Please see Figure 1 This invention provides a jaw registration method based on multidimensional intraoral scan data and geometric constraint optimization, comprising the following steps.

[0032] Step 100: Read the user's maxillary and mandibular scan data and occlusion record data, and obtain the maxillary dentition point cloud, mandibular dentition point cloud, left buccal point cloud and right buccal point cloud based on the maxillary and mandibular scan data and occlusion record data.

[0033] The system supports PLY point cloud file input and performs adaptive uniform sampling on STL meshes, with the number of sampling points dynamically adjusted according to mesh complexity. A file naming convention is established: {prefix}LowerJawScan, {prefix}UpperJawScan, {prefix}BiteScan, {prefix}BiteScan2; a batch processing mechanism is implemented, supporting continuous processing of multiple cases with numerical prefixes.

[0034] When performing ICP registration of the maxillary and mandibular point clouds, a fit threshold of 0.000025 was set to eliminate invalid registrations; the data source with the smallest total RMSE (root mean square error) and both maxillary and mandibular points were selected.

[0035] Step 200: Based on the maxillary dentition point cloud, mandibular dentition point cloud, and left buccal point cloud P B left And right cheek dot cloud P B right Calculate the optimal registration path and obtain the coarse registration model based on the optimal registration path.

[0036] This invention will use point clouds (P) on the left and right cheek faces. B () can be used as a connection point cloud P of the maxillary dentition U and mandibular dentition point cloud P L A rigid "bridge". Through two ICP registrations (P... B →Maxillary dentition dot cloud P U and P B →Mandibular dentition dot cloud P L By combining the first-order matrix inversion and the resultant matrix, the point cloud P of the mandibular dentition can be calculated. L →Maxillary dentition dot cloud P U Transformation matrix T (L→U) .

[0037] To improve robustness, this invention performs "cheek bridging" in parallel on the left and right cheek data: Left path: starting with P B leftAs a bridge, the left jaw position transformation matrix T is calculated. left .

[0038] Right-hand path: starting with P B right As a bridge, the right jaw position transformation matrix T is calculated. right .

[0039] Then, the registration quality of the left and right paths is compared (e.g., root mean square error (RMSE) and fit based on ICP registration). The transformation matrix (T) of the side with higher quality (e.g., lower RMSE) is automatically selected. left or T right ) as the optimal initial transformation matrix T initial T initial Applied to mandibular point cloud P L (Registration is generally based on the mandibular point cloud), resulting in the coarsely registered mandibular model P. L' .

[0040] Step 300: Iterate the pose of the coarse registration model to obtain the registration model, wherein the bite quality fraction Q of the registration model is maximized.

[0041] In order to correct P L' By minimizing the pose and eliminating penetration and distortion, this invention introduces pose adjustment and quality assessment to achieve more accurate registration.

[0042] Please see Figure 2 As shown, in stage 1, this invention uses multi-dimensional oral scan data registration based on the ICP algorithm to perform bilateral buccal registration of the left and right maxillary and mandibular oral scan clouds with the left and right buccal oral scan clouds. Then, in stage 2, pose optimization based on jaw position set constraints is performed iteratively to achieve fine registration. Details are as follows.

[0043] As described in step 200, upper and lower left point cloud registration is performed based on the maxillary dentition point cloud, mandibular dentition point cloud, and left buccal point cloud; upper and lower right point cloud registration is performed based on the maxillary dentition point cloud, mandibular dentition point cloud, and right buccal point cloud; the optimal registration path is selected based on the registration results of the left and right paths, and a coarse registration model is obtained based on the optimal registration path; including: IPC registration of the maxillary dentition point cloud, mandibular dentition point cloud and left buccal point cloud is calculated on the left path to obtain the first maxillary transformation matrix and the first mandibular transformation matrix. IPC registration of the maxillary dentition point cloud, mandibular dentition point cloud and right buccal point cloud is calculated on the right path to obtain the second maxillary transformation matrix and the second mandibular transformation matrix.

[0044] For example, execute "ICP registration:P" in the left path. B left →PU "and ICP registration: P" B left →P L Execute "ICP registration: P" in the path on the right. B right →P U "and ICP registration: P" B right →P L ".

[0045] The transformation matrix T of the left path is calculated separately for each of the two paths. left and the right-hand transformation matrix T right .

[0046] Left path: 1. Execute ICP, and transfer P B left (Source) aligned to P U (Target) Obtain the transformation matrix T U left ; 2. Execute ICP, and transfer P B left (Source) aligned to P B left →P L (Target) Obtain the transformation matrix T L left .

[0047] Path on the right: 1. Execute ICP, and transfer P B right (Source) aligned to P U (Target) Obtain the transformation matrix T U right ; 2. Execute ICP, and transfer P B right (Source) aligned to P B right →P L (Target) Obtain the transformation matrix T L right .

[0048] Then, using matrix inversion and composite transformations, the jaw position relationship between the left and right sides is calculated separately. The formula is: T (L→U) =T (B→U) *(T_ (B→L) ) (-1) ; Among them (⋅) (-1) This represents the matrix inversion. Using this formula, we obtain two candidate initial transformation matrices: Tleft and T right .

[0049] Compare the first maxillary transformation matrix T U left With the first mandibular transformation matrix T L left The registration quality is determined, and the matrix with the higher registration quality is used as the first initial transformation matrix. The second maxillary transformation matrix T is then compared. U right With the second mandibular transformation matrix T L right The registration quality is determined by selecting the matrix with the highest registration quality as the second initial transformation matrix.

[0050] Then compare T left and T right The registration quality (root mean square error, goodness of fit) is used to select the optimal T. initial The coarse registration model P is obtained by applying the transformation. L' .

[0051] In one implementation, the mean root mean square error and / or mean goodness of fit of the first initial transformation matrix and the second initial transformation matrix are compared, and the transformation matrix with the lowest mean root mean square error and / or the highest mean goodness of fit is taken as the initial transformation matrix.

[0052] Root Mean Square Error (RMSE): ;p i' It is p i The transformed points. The smaller the RMSE, the more accurate the registration.

[0053] Fit: N inlier It represents the number of point pairs whose distance after registration is less than a certain threshold (e.g., 0.2mm). The closer Fitness is to 1, the better the overlap.

[0054] Calculate the average RMSE and Fitness of the two ICPs on the left path, and then calculate the average RMSE and Fitness of the two ICPs on the right path.

[0055] Choose the transformation matrix (T) corresponding to the path with the lowest average RMSE and the highest average Fitness. left or T right ) as the final initial transformation matrix T initial .

[0056] Finally, the initial transformation matrix is ​​applied to the corresponding mandibular point cloud to obtain the coarse registration model after registration, i.e., T initial Applied to the original mandibular point cloud P LThe coarsely registered mandibular model P was obtained. L' .

[0057] It should be understood that in other embodiments of the present invention, only RMSE may be compared, or only Fitness may be compared, or a weighted sum of RMSE and Fitness may be compared.

[0058] Furthermore, in stage two, pose optimization is based on geometric constraints. Existing "black box" algorithms cannot eliminate penetration and distortion. The goal of this stage is to receive the coarse registration model P. L' It further fine-tunes its 6-DOF pose (3 translations, 3 rotations) iteratively to maximize a Q score that represents clinical occlusal quality.

[0059] In step 300 above, iterating the pose of the coarse registration model to obtain the registration model includes: The key indicators of the coarse registration model are calculated, and the optimal optimization path for six-degree-of-freedom rigid body transformation is automatically selected from four strategies: translation priority, rotation priority, alternating optimization, and fine adjustment, based on the key indicators. The key indicators include centroid distance, contact density, normal alignment, and gap distribution pattern.

[0060] At the beginning of each iteration, analyze P. U and P L' Current state: Distance between centroids (D) centroid ): c lower and c upper These are the centroids of the point clouds for the upper and lower jaws, respectively. They are used to assess macroscopic positional deviations.

[0061] Contact density (ρ) contact ): ; I(⋅) is the indicator function (1 if the condition is true, 0 otherwise). d(⋅) is the minimum distance from a point to the point cloud. τ contact This is the contact threshold (e.g., 0.5 mm). This metric assesses the proportion of potential contact points.

[0062] Normal vector alignment (A) normal ): ; C is the set of contact points (i.e., d < τ) contact point). and These are the normal vectors of the nearest points on the upper and lower jaws, respectively. |⋅| represents the absolute value. The closer this value is to 1, the better the cusp bevel fit.

[0063] Gap distribution pattern (P) g ): Analyze the gap g of the contact point set C i for its spatial distribution.

[0064] ; (relative variance, to evaluate uniformity).

[0065] ; (spatial correlation, to evaluate whether it is inclined along the X-axis).

[0066] Analyze the gap set . Calculate the relative variance and the spatial correlation ( ): ; ; where is the mean of the gaps, represents the variance of the gap set, to evaluate uniformity., is the standard deviation of the X / Y coordinate set, represents the covariance between the X / Y coordinate set of the contact points and the gap set, and uses spatial correlation to evaluate whether it is inclined along the X-axis.

[0067] Based on the above features, determine the current jaw registration mode .

[0068] ; where f1, f2, f3, f4, f5 are constants; for example: ; Use a decision tree to select the strategy for this round of iteration : ; where m1, m2, m3, m4, m5 are adaptively adjusted parameters, and 2 ≤ m4 < m1 < 10, 0 < m2 < 1, m3 < m5 < 1. For example: .

[0069] It should be understood that the above constant parameters given in the embodiments of the present invention are not used to limit the present invention and can be adjusted accordingly according to requirements.

[0070] The present invention calculates the comprehensive quality score Q of the current pose and maximizes Q through iteration. The system calculates an ideal six-degree-of-freedom (6-DOF, i.e., three-dimensional translation and three-dimensional rotation) correction transformation τ opt , ω optThe purpose of this transformation is to "push away" the penetration point, "pull closer" the dangling point, and "align" the tooth cusp bevel. An adaptive step size α is used to control the actual magnitude of the transformation applied each time, updating P. L' The pose is determined. The iteration stops when Q reaches a preset high score (e.g., Q>0.9), or when Q no longer improves significantly, and the final finely registered mandibular model P is output. L'' .

[0071] Calculate the overall quality score Q for each degree of freedom pose iteration. The registration model is obtained when the overall quality score Q is maximized, where the overall quality score is: ; Where U represents uniformity, S represents safety, Δtarget represents target deviation, and a, b, and c are weight parameters in the interval [0,1], with a+b+c=1.

[0072] like The weighting parameters can be adjusted according to clinical needs. For example, if more emphasis is placed on S, it can be set to Q = 0.6 * S + ... . It should be understood that the constant parameters given in the embodiments of this invention are not intended to limit the invention and can be adjusted accordingly as needed.

[0073] Specifically, N penetration It is the penetration point, S=1 indicates zero penetration; ;σ g and These are the standard deviation and mean of the gap, respectively. g target It is the target gap, g i C is the gap, C is the set of contact points, and Nc is the contact point.

[0074] This invention introduces the "Safety S" index to proactively identify and eliminate model penetration errors common in existing "black box" technologies. Furthermore, this invention proactively optimizes the morphology of occlusal contact through the "Uniformity U" and "Target Deviation Δtarget" indices, resulting in more uniform and stable cusp-fossa contact and intercuspal spacing that better meets clinical requirements.

[0075] Specifically, a “ideal” 6-DOF correction is calculated by fine-tuning its 6-DOF pose to maximize Q. The transformation is performed around the centroid c of the mandibular model.

[0076] Transformation matrix construction: ; ; Optimal translation vector calculation: 1. For each near contact point i (e.g., g) i Calculate the "translation requirement" vector V (<1.0mm). i : ; The direction_vector starts from the maxillary point p. j , pointing to the mandibular point p i The unit vector of g. This vector is intended to represent g. i Push or pull to g target .

[0077] 2. Define weights The smaller the gap (or the penetration), the greater the weight.

[0078] 3. Calculate the weighted average translation vector t opt : ; Optimal rotation vector calculation: 1. For each near-contact point i, calculate its "rotation requirement" vector w. i : Rotation axis (Normalization of cross product of normal vectors); Rotation angle (Dot product of normal vectors); (Rotation vector, with a maximum rotation angle limited to 0.1 radians); Using the same weight w i Calculate the weighted average rotation vector opt : .

[0079] Amplitude Limitation: To prevent iterative divergence, limit t opt and w opt The maximum amplitude.

[0080] w final Similarly.

[0081] Adaptive step size and iterative applications: Calculate the step size α for this iteration. The update of α is based on the change ΔQ in the Q value from the previous iteration. (k) : ; Where g1 and g2 are the step size adjustment weight parameters, For △Q (k) The adjustment threshold can be adjusted according to the actual situation.

[0082] For example: ; The transformation (translation t) applied in this round of practice (k) and rotation w (k) ) Applied to model P L' : ; Update model P L' Based on the pose, perform geometric feature analysis and strategy selection to begin the next iteration.

[0083] The system outputs the mandibular point cloud model P after stage two fine optimization. L'' The model P L'' With P U Together, they form a digital jaw relationship with low penetration, uniform contact, and high precision, which can be directly used for the digital design and manufacturing of subsequent restorations such as crowns and implant guides.

[0084] Compared with existing technologies (such as the "black box" algorithms built into commercial scanners), this invention has the following significant advantages: 1. Solved the problem of "black box" registration (innovation): This invention proposes and implements a transparent and controllable "cheek bridging" strategy, which solves the fundamental problem that registration is impossible due to the lack of overlapping areas in the upper and lower jaw point clouds.

[0085] 2. Strong robustness: The invention’s unique “bilateral buccal surface optimization” strategy can intelligently identify and avoid poor-quality unilateral buccal surface scanning data, which significantly improves the initial registration success rate and stability in cases of clinical asymmetrical occlusion or poor scanning.

[0086] 3. High clinical relevance (eliminating penetration): The second-stage "geometric constraint optimization" algorithm of this invention, by introducing the "safety S" index, can actively identify and eliminate the model penetration error that is common in existing "black box" technologies.

[0087] 4. High precision (optimized contact): This invention not only eliminates penetration, but also achieves high precision through "uniformity U" and "target deviation Δ". target "Indicators that proactively optimize the morphology of occlusal contact, making the contact between the cusps and fossae more uniform and stable, and the gaps more in line with clinical requirements."

[0088] Please see Figure 3 The present invention also provides a jaw registration device based on multidimensional intraoral scan data and geometric constraint optimization, comprising: The acquisition module 31 is used to read the user's upper and lower jaw scan data and occlusion record data, and to acquire the maxillary dentition point cloud, mandibular dentition point cloud, left buccal point cloud and right buccal point cloud based on the upper and lower jaw scan data and occlusion record data. The first registration module 32 is used to calculate the optimal registration path based on the maxillary dentition point cloud, mandibular dentition point cloud, left buccal surface point cloud and right buccal surface point cloud, and to obtain a coarse registration model based on the optimal registration path. The second registration module 33 is used to iterate the pose of the coarse registration model to obtain a registration model, wherein the bite quality fraction Q of the registration model is maximized.

[0089] For a description of the jaw registration device based on multidimensional intraoral scan data and geometric constraint optimization, please refer to the jaw registration method based on multidimensional intraoral scan data and geometric constraint optimization, which will not be repeated here.

[0090] The present invention also provides an electronic device, comprising: At least one processor; and at least one memory communicatively connected to the processor, wherein: the memory stores program instructions executable by the processor, and the processor can execute the above-described jaw registration method based on multidimensional intraoral scan data and geometric constraints by calling the program instructions.

[0091] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described jaw registration method based on multidimensional intraoral scan data and geometric constraint optimization.

[0092] It is understood that computer-readable storage media can include: any entity or device capable of carrying computer programs, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), and software distribution media, etc. Computer programs include computer program code. Computer program code can be in the form of source code, object code, executable files, or certain intermediate forms, etc. Computer-readable storage media can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, ROM, RAM, and software distribution media, etc.

[0093] In some embodiments of the present invention, the device may include a controller, which is a microcontroller chip integrating a processor, memory, communication module, etc. The processor may refer to the processor included in the controller. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0094] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0095] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0096] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A jaw registration method based on multidimensional intraoral scan data and geometric constraint optimization, characterized in that, include: Read the user’s maxillary and mandibular scan data and occlusion record data, and obtain the maxillary dentition point cloud, mandibular dentition point cloud, left buccal point cloud and right buccal point cloud based on the maxillary and mandibular scan data and occlusion record data; Based on the maxillary dentition point cloud, mandibular dentition point cloud, and left buccal point cloud, upper and lower left point cloud registration is performed. Based on the maxillary dentition point cloud, mandibular dentition point cloud, and right buccal point cloud, upper and lower right point cloud registration is performed. The optimal registration path is selected based on the registration results of the left and right paths, and a coarse registration model is obtained based on the optimal registration path. The registration model is obtained by iterating the pose of the coarse registration model, and the occlusion quality fraction Q of the registration model is maximized.

2. The jaw registration method based on multidimensional intraoral scan data and geometric constraint optimization according to claim 1, characterized in that, The process involves registering upper and lower left point clouds based on the maxillary dentition point clouds, mandibular dentition point clouds, and left buccal point clouds; registering upper and lower right point clouds based on the maxillary dentition point clouds, mandibular dentition point clouds, and right buccal point clouds; selecting the optimal registration path based on the registration results of the left and right paths; and obtaining a coarse registration model based on the optimal registration path. This includes: ICP registration of the maxillary dentition point cloud, mandibular dentition point cloud and left buccal point cloud is calculated on the left path to obtain the first maxillary transformation matrix and the first mandibular transformation matrix. ICP registration of the maxillary dentition point cloud, mandibular dentition point cloud and right buccal point cloud is calculated on the right path to obtain the second maxillary transformation matrix and the second mandibular transformation matrix. The registration quality of the first maxillary transformation matrix and the first mandibular transformation matrix is ​​compared, and the one with higher registration quality is taken as the first initial transformation matrix. The registration quality of the second maxillary transformation matrix and the second mandibular transformation matrix is ​​compared, and the matrix with higher registration quality is used as the second initial transformation matrix.

3. The jaw registration method based on multidimensional intraoral scan data and geometric constraint optimization according to claim 2, characterized in that, The process of obtaining the coarse registration model based on the optimal registration path includes: Compare the average root mean square error and / or average goodness of fit of the first initial transformation matrix and the second initial transformation matrix, and take the transformation matrix with the lowest average root mean square error and / or the highest average goodness of fit as the initial transformation matrix. The initial transformation matrix is ​​applied to the corresponding mandibular point cloud to obtain the coarse registration model after registration.

4. The jaw registration method based on multidimensional intraoral scan data and geometric constraint optimization according to claim 3, characterized in that, The step of iteratively obtaining the registration model by adjusting the pose of the coarse registration model includes: The key indicators of the coarse registration model are calculated, and the optimal optimization path for six-degree-of-freedom rigid body transformation is automatically selected from four strategies: translation priority, rotation priority, alternating optimization, and fine adjustment, based on the key indicators. The key indicators include centroid distance, contact density, normal alignment, and gap distribution pattern.

5. The jaw registration method based on multidimensional intraoral scan data and geometric constraint optimization according to claim 4, characterized in that, The step of iteratively obtaining the registration model by altering the degrees of freedom pose of the coarse registration model includes: iterating the degrees of freedom pose of the coarse registration model using the following iterative strategy: ； Where m1, m2, m3, m4, and m5 are adaptive adjustment parameters, and 2 ≤ m4 <m1<10,0<m2<1,m3<m5<1。 6. The jaw registration method based on multidimensional intraoral scan data and geometric constraint optimization according to claim 4, characterized in that, The step of iteratively obtaining the registration model by adjusting the pose of the coarse registration model includes: Calculate the overall quality score Q for each degree of freedom pose iteration. The registration model is obtained when the overall quality score Q is maximized, where the overall quality score is: ； Where U represents uniformity, S represents safety, Δtarget represents target deviation, and a, b, and c are weight parameters in the interval [0,1], with a+b+c=1.

7. The jaw registration method based on multidimensional intraoral scan data and geometric constraint optimization according to claim 6, characterized in that, N penetration It is the penetration point, S=1 indicates zero penetration; ;σ g and These are the standard deviation and mean of the gap, respectively. g target It is the target gap, g i C is the gap, C is the set of contact points, and Nc is the contact point.

8. A jaw registration device based on multidimensional intraoral scan data and geometric constraint optimization, characterized in that, include: The acquisition module is used to read the user's maxillary and mandibular scan data and occlusion record data, and to acquire maxillary dentition point cloud, mandibular dentition point cloud, left buccal point cloud and right buccal point cloud based on the maxillary and mandibular scan data and occlusion record data. The first registration module is used to calculate the optimal registration path based on the maxillary dentition point cloud, mandibular dentition point cloud, left buccal surface point cloud, and right buccal surface point cloud, and to obtain a coarse registration model based on the optimal registration path. The second registration module is used to iterate the pose of the coarse registration model to obtain a registration model, wherein the bite quality fraction Q of the registration model is maximized.

9. An electronic device, characterized in that, include: At least one processor; And at least one memory communicatively connected to the processor, wherein: the memory stores program instructions executable by the processor, and the processor invokes the program instructions to execute the jaw registration method based on multidimensional intraoral scan data and geometric constraint optimization as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed by a computer, performs the jaw registration method based on multidimensional intraoral scan data and geometric constraint optimization as described in any one of claims 1 to 7.

Images