Dental arch arrangement state recognition method and system based on intraoral image

Abstract

The application discloses a dentition arrangement state recognition method and system based on intraoral images, and relates to the technical field of intraoral image processing. The method comprises the following steps: acquiring tooth visual structure data; recognizing visual occlusion according to the opposite boundary of adjacent teeth, restoring the potential opposite boundary affected by the occlusion, determining the potential adjacent position, signed adjacent distance and adjacent recovery confidence; constructing a dental arch reference track, determining the dental arch arc length position, dental arch normal offset, crown direction offset and normalized adjacent distance; determining the arrangement abnormal structure mechanism; and determining the dentition arrangement state according to the arrangement abnormal structure mechanism and arrangement index. The method can reduce the influence of tooth visual occlusion and image local scale difference on the dentition arrangement state recognition, and improve the distinguishing ability of different arrangement abnormal states.

Description

Technical Field

[0001] This invention relates to the field of intraoral image processing technology, specifically to a method and system for recognizing the tooth alignment status based on intraoral images. Background Technology

[0002] Dental alignment is a crucial aspect of oral examinations, orthodontic assessments, and tooth morphology analysis. It typically includes normal alignment, tooth rotation, tooth misalignment along the dental arch normal, segmental crowding, and intersegmental gaps. With advancements in intraoral imaging equipment and computer vision technology, the automatic analysis of tooth position, orientation, and adjoint relationships using intraoral images can assist in identifying dental alignment, reducing reliance on experience for manual observation.

[0003] Existing methods for identifying dentition alignment based on intraoral images typically analyze tooth crown contours, tooth center positions, tooth orientations, or image distances between adjacent teeth, and combine these with preset rules or classification models to determine the corresponding dentition alignment. However, in actual intraoral images of the occlusal surface, visual obstruction can easily occur between adjacent teeth due to tooth rotation, misalignment, or shooting angle, making it impossible to directly observe the contours of some posterior teeth. This leads to a discrepancy between the adjacent tooth alignment relationship determined based on the visible contours and the actual alignment relationship.

[0004] Meanwhile, intraoral images are affected by imaging distance, arch curvature, and different tooth spatial positions. The image scale corresponding to different tooth regions may differ. Directly using image coordinates or pixel distance for tooth alignment analysis can easily lead to a lack of a unified basis for comparison of alignment indicators between different tooth positions. In addition, tooth rotation, tooth misalignment along the normal direction of the dental arch, and abnormal spatial allocation of the dental arch tangential direction may show similar adjacent spatial changes in two-dimensional intraoral images. Classifying based solely on local appearance or a single alignment indicator makes it difficult to further distinguish the corresponding structural mechanisms of alignment abnormalities.

[0005] Therefore, how to recover reliable adjacent tooth alignment relationships and establish a unified benchmark for relative arch alignment in the presence of visual occlusion of teeth and local scale differences in images, while distinguishing different abnormal alignment structural mechanisms with similar appearances, are technical problems that need to be solved when identifying the tooth alignment status based on intraoral images. Summary of the Invention

[0006] In view of the above situation and to overcome the shortcomings of the prior art, the technical solution adopted by the present invention is as follows: The present invention provides a method for recognizing the tooth alignment status based on intraoral images, the method comprising the following steps:

[0007] Step S1: Extraction of visual structure of teeth;

[0008] Step S2: Restore the arrangement relationship of adjacent teeth that are obstructed;

[0009] Step S3: Establishing the relative alignment reference of the dental arches;

[0010] Step S4: Verification of local arrangement anomalies;

[0011] Step S5: Identify the tooth alignment status.

[0012] Further, in step S1, the extraction of the tooth visual structure is used to establish the visual structure basis corresponding to each tooth in the intraoral image. Specifically, it involves acquiring and correcting the original intraoral image of the occlusal surface corresponding to the dental arch to be identified, extracting the crown region, tooth outline and crown direction corresponding to each tooth, and determining the dentition order to obtain the tooth visual structure data.

[0013] Further, in step S2, the adjacent tooth occlusion arrangement relationship recovery is used to reduce the impact of mutual tooth occlusion on the identification of adjacent tooth arrangement relationship. Specifically, it involves determining the opposing boundaries of adjacent teeth based on the tooth visual structure data, identifying the visual occlusion between opposing boundaries and recovering the opposing boundaries affected by visual occlusion, determining the potential adjacency positions, marked adjacency distances and adjacency recovery confidence between adjacent teeth, and obtaining adjacent tooth arrangement relationship data.

[0014] The restoration of the adjacent tooth occlusion arrangement relationship includes the following steps:

[0015] Step S21: Determine the opposing boundary of adjacent teeth. This step is used to define the local boundary for the analysis of the adjacent tooth arrangement relationship from the tooth contour. Specifically, it determines the corresponding adjacency direction according to the tooth row order and the relative position between adjacent teeth, and extracts the continuous contour segment facing the opposite tooth to obtain the opposing boundary data of adjacent teeth.

[0016] Step S22: Occlusion boundary interval identification, used to determine the visual occlusion relationship between adjacent teeth and the boundary interval affected by visual occlusion. Specifically, it identifies the occlusion intersection position based on the adjacent tooth opposing boundary data, and determines the front layer teeth, back layer teeth and occlusion boundary interval to obtain adjacent tooth occlusion structure data.

[0017] Step S23: Potential opposing boundary recovery, used to recover the local boundaries of the posterior teeth that cannot be directly observed within the occlusion boundary range. Specifically, it involves establishing and screening candidate recovery boundaries based on the adjacent tooth occlusion structure data to obtain potential opposing boundary data.

[0018] Step S24: Determine potential adjacency relationship, which is used to determine the arrangement relationship between adjacent teeth based on the restored opposing boundary. Specifically, it determines the potential adjacency position, the marked adjacency distance, and the adjacency restoration confidence, and obtains the adjacent tooth arrangement relationship data.

[0019] Further, in step S3, the construction of the relative alignment reference of the dental arch is used to establish a dental arch alignment reference that reduces the influence of local scale differences in intraoral images. Specifically, it involves determining the local scale of the teeth and the adjacent position chain of the dental arch based on the visual structure data of the teeth and the adjacent tooth alignment relationship data, constructing the dental arch reference trajectory, and establishing the relative alignment representation of each tooth with respect to the dental arch reference trajectory to obtain the dental arch alignment reference data.

[0020] The acquisition of the tooth alignment reference data specifically involves determining the local scale of each tooth based on the tooth visual structure data and adjacent tooth alignment relationship data, connecting potential adjacent positions that meet preset confidence conditions according to the tooth row sequence to form a tooth row adjacent position chain, and performing scale-normalized robust curve fitting on the tooth row adjacent position chain based on the local scale of the corresponding tooth to obtain the dental arch reference trajectory. Based on the position and direction relationship of each tooth relative to the dental arch reference trajectory, a relative alignment representation of the dental arch is established to obtain the tooth alignment reference data.

[0021] The construction of the relative alignment reference of the dental arches includes the following steps:

[0022] Step S31: Determine the local scale of teeth, which is used to establish the local scale benchmark corresponding to different tooth positions in the intraoral image. Specifically, the local scale of each tooth is determined according to the recovered tooth contour to obtain the local scale data of teeth.

[0023] Step S32: Establish the dental arch adjacency position chain to form an arrangement structure that is continuously distributed along the dental arch direction and has adjacency recovery confidence constraints. Specifically, establish the dental arch adjacency position chain based on the potential adjacency positions and the corresponding adjacency recovery confidence to obtain dental arch adjacency position chain data.

[0024] Step S33: Dental arch reference trajectory construction, used to establish a dental arch reference trajectory that reduces the influence of locally misaligned teeth on the dental arch alignment reference benchmark. Specifically, the scale-normalized weighted trajectory fitting is performed on the adjacent position chain of the dental arch to obtain dental arch reference trajectory data.

[0025] Step S34: Relative alignment characterization of dental arches, used to convert the tooth positions described based on image coordinates into alignment relationships relative to the dental arch reference trajectory. Specifically, it involves determining the alignment index of each tooth relative to the dental arch reference trajectory to obtain tooth alignment baseline data.

[0026] Further, in step S4, the local alignment anomaly verification is used to distinguish different alignment anomaly structural mechanisms after detecting local alignment anomalies. Specifically, it involves determining the tooth segment to be verified based on the tooth alignment benchmark data, establishing candidate counterfactual alignment states corresponding to different alignment anomaly structural mechanisms, evaluating the recovery of local alignment anomalies by each candidate counterfactual alignment state, determining the alignment anomaly structural mechanism, and obtaining tooth alignment anomaly verification data.

[0027] The verification of local arrangement anomalies includes the following steps:

[0028] Step S41: Determine the abnormal arrangement to be verified, which is used to locate the teeth and tooth segments that need to be verified for abnormal structural mechanisms in the dental arch. Specifically, it is to determine the relationship between the teeth to be verified and the abnormal adjacent teeth according to the tooth arrangement reference data, and form the tooth segments to be verified to obtain the tooth segment data to be verified.

[0029] Step S42: Mechanism constraint reverse arrangement adjustment, used to establish comparable counterfactual states corresponding to different abnormal arrangement structural mechanisms. Specifically, reverse adjustment of tooth rotation, reverse adjustment of teeth along the dental arch normal, and reverse adjustment of dental arch tangential spatial allocation are performed on the tooth segment to be verified to obtain candidate counterfactual arrangement state data.

[0030] Step S43: Counterfactual structure recovery evaluation, used to determine the explanatory power of different permutation anomaly mechanisms for the current local permutation anomaly, specifically to evaluate the degree of local permutation anomaly recovery and normalization adjustment magnitude corresponding to each candidate counterfactual permutation state, and obtain counterfactual structure recovery evaluation data;

[0031] Step S44: Verification of the abnormal arrangement structure mechanism, used to determine the corresponding structure mechanism that can explain the current abnormal arrangement from multiple candidate abnormal arrangement structure mechanisms. Specifically, it is to determine the abnormal arrangement structure mechanism and abnormal teeth or abnormal tooth segments corresponding to the tooth segment to be verified based on the counterfactual structure recovery evaluation data, and obtain the tooth row abnormality verification data.

[0032] Further, in step S5, the tooth alignment status identification is used to determine the tooth alignment status based on the verified abnormal alignment structure mechanism and corresponding alignment index. Specifically, it determines the normal alignment status and the corresponding abnormal alignment status based on the tooth alignment benchmark data and tooth alignment abnormality verification data, and obtains the tooth alignment status identification result.

[0033] The present invention also provides a tooth alignment status recognition system based on intraoral images, including a tooth visual structure extraction module, an adjacent tooth occlusion alignment relationship recovery module, a dental arch relative alignment benchmark construction module, a local alignment anomaly verification module, and a tooth alignment status recognition module;

[0034] The tooth visual structure extraction module is used to establish the visual structure basis of each tooth in the intraoral image, obtain tooth visual structure data through tooth visual structure extraction, and send the tooth visual structure data to the adjacent tooth occlusion arrangement relationship recovery module.

[0035] The adjacent tooth occlusion arrangement relationship recovery module is used to reduce the impact of mutual tooth occlusion on the recognition of interdental arrangement relationship. Through adjacent tooth occlusion arrangement relationship recovery, adjacent tooth arrangement relationship data is obtained, and the adjacent tooth arrangement relationship data is sent to the dental arch relative arrangement benchmark construction module.

[0036] The dental arch relative alignment reference construction module is used to establish a dental arch alignment reference that reduces the influence of local scale differences in intraoral images. Through the construction of the dental arch relative alignment reference, the dental alignment reference data is obtained, and the dental alignment reference data is sent to the local alignment anomaly verification module.

[0037] The local arrangement anomaly verification module is used to distinguish the formation mechanism corresponding to the local arrangement anomaly, obtain the dental arrangement anomaly verification data through local arrangement anomaly verification, and send the dental arrangement anomaly verification data to the dental arrangement state recognition module;

[0038] The tooth alignment state recognition module is used to generate a tooth alignment state recognition result, obtain the tooth alignment state recognition result through tooth alignment state recognition, and output the tooth alignment state recognition result.

[0039] The beneficial effects achieved by the present invention using the above solution are as follows:

[0040] (1) To address the problem that the existing intraoral image tooth alignment status recognition is easily affected by mutual occlusion of teeth and changes in shooting scale, this solution adopts a continuous processing mechanism consisting of adjacent tooth alignment relationship restoration, dental arch intrinsic benchmark construction, and local alignment anomaly verification. By first restoring the potential adjacent structure affected by occlusion, then converting the tooth alignment relationship into a scale-normalized representation relative to the dental arch reference trajectory, and verifying the formation mechanism of the detected alignment anomalies, the error caused by simply judging the status based on the visible crown outline or image pixel distance is reduced. For example, when local teeth are occluded and different tooth regions have different image scales, the tooth alignment status recognition can still be completed based on the restored adjacent structure and dental arch intrinsic alignment relationship.

[0041] (2) To address the problem that the visible outline of the posterior teeth cannot directly represent their potential adjacent boundaries due to tooth occlusion, this solution adopts an occlusion-constrained adjacent tooth opposing boundary recovery mechanism. By determining the anterior and posterior teeth, the occlusion boundary interval corresponding to the posterior teeth is located, and the position and direction information of the visible boundaries at both ends of the occlusion interval is used to recover the potential opposing boundaries, so that the subsequent interdental relationship analysis is based on the recovered potential adjacent structure rather than the visible outline formed by occlusion. For example, when a rotating incisor occludes part of the adjacent boundary of an adjacent incisor, the local boundary extension trend on the occluded side can be recovered, thereby reducing the probability of directly judging visual layering as interdental space compression.

[0042] (3) To address the problem that different arrangement anomalies may produce similar adjacency anomalies in two-dimensional intraoral images and are difficult to distinguish, this scheme adopts a mechanism-constrained local arrangement anomaly verification mechanism. By restricting the arrangement degrees of freedom corresponding to tooth rotation, dental arch normal position and dental arch direction spatial allocation respectively, different candidate counterfactual arrangement states are constructed, and the degree of recovery of the original local arrangement anomaly and the required adjustment range of each candidate state are compared, so that the arrangement state recognition is transformed from simple appearance classification to structural verification of the anomaly formation mechanism. For example, when single tooth rotation and continuous tooth segment crowding both show bilateral adjacency compression, the two arrangement states can be distinguished by comparing the difference in structural recovery produced by only changing the tooth direction and only redistributing the local space along the dental arch direction. Attached Figure Description

[0043] Figure 1 A schematic flowchart illustrating the tooth alignment recognition method based on intraoral images provided by this invention;

[0044] Figure 2 A schematic diagram of the tooth alignment status recognition system based on intraoral images provided by the present invention;

[0045] Figure 3 A flowchart illustrating the process of restoring the arrangement relationship of adjacent teeth in step S2;

[0046] Figure 4 A schematic diagram of the process for constructing the relative alignment reference of the dental arches in step S3;

[0047] Figure 5 This is a flowchart illustrating the process of verifying local arrangement anomalies in step S4.

[0048] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0049] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0050] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0051] Example 1, see Figure 1 The present invention provides a method for recognizing the tooth alignment status based on intraoral images, the method comprising the following steps:

[0052] Step S1: Extraction of visual structure of teeth;

[0053] Step S2: Restore the arrangement relationship of adjacent teeth that are obstructed;

[0054] Step S3: Establishing the relative alignment reference of the dental arches;

[0055] Step S4: Verification of local arrangement anomalies;

[0056] Step S5: Identify the tooth alignment status.

[0057] By performing the above operations, this solution addresses the problem that existing intraoral image tooth alignment recognition is easily affected by mutual tooth occlusion and changes in imaging scale. It adopts a continuous processing mechanism consisting of adjacent tooth alignment relationship restoration, dental arch intrinsic benchmark construction, and local alignment anomaly verification. By first restoring the potential adjacent structures affected by occlusion, then converting the tooth alignment relationship into a scale-normalized representation relative to the dental arch reference trajectory, and verifying the formation mechanism of detected alignment anomalies, the error caused by simply judging the state based on the visible crown contour or image pixel distance is reduced. For example, when local teeth are occluded and different tooth regions have different image scales, tooth alignment status recognition can still be completed based on the restored adjacent structures and dental arch intrinsic alignment relationships.

[0058] Example 2, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S1, the extraction of tooth visual structure is used to establish the visual structure basis of each tooth in the intraoral image. Specifically, it involves acquiring and correcting the original intraoral image of the occlusal surface corresponding to the dental arch to be identified, extracting the crown region, tooth outline and crown direction of each tooth, and determining the order of the dental arch to obtain tooth visual structure data.

[0059] The acquisition of the tooth visual structure data specifically involves acquiring the original intraoral image of the occlusal surface corresponding to the dental arch to be identified, performing image correction on the original intraoral image of the occlusal surface to obtain the intraoral image of the occlusal surface, using a tooth instance segmentation method to determine the crown region and tooth contour of each tooth, determining the crown direction based on the tooth contour, determining the dentition order according to the continuous distribution relationship of teeth in the dental arch, and associating the crown region, tooth contour, crown direction, and dentition order to obtain the tooth visual structure data.

[0060] Example 3, see Figure 1 , Figure 2 and Figure 3 This embodiment is based on the above embodiment. In step S2, the adjacent tooth occlusion arrangement relationship recovery is used to reduce the impact of mutual tooth occlusion on the identification of adjacent tooth arrangement relationship. Specifically, it determines the opposing boundaries of adjacent teeth based on the tooth visual structure data, identifies the visual occlusion between opposing boundaries and recovers the opposing boundaries affected by visual occlusion, determines the potential adjacency position, the marked adjacency distance and the adjacency recovery confidence between adjacent teeth, and obtains adjacent tooth arrangement relationship data.

[0061] The acquisition of the adjacent tooth arrangement relationship data specifically involves determining instances of teeth that are sequentially adjacent based on the tooth visual structure data, identifying opposing boundaries that face each other from the tooth contours of adjacent teeth, identifying the occlusion level of visual occlusion between the opposing boundaries, restoring the continuity of opposing boundaries affected by visual occlusion, determining the potential adjacency positions and marked adjacency distances between adjacent teeth based on the restored opposing boundaries, and determining the adjacency restoration reliability based on the restoration of opposing boundaries to obtain the adjacent tooth arrangement relationship data.

[0062] The restoration of the adjacent tooth occlusion arrangement relationship includes the following steps:

[0063] Step S21: Determine the opposing boundary of adjacent teeth. This step is used to define the local boundary for the analysis of the adjacent tooth arrangement relationship from the tooth contour. Specifically, it determines the corresponding adjacency direction according to the tooth row order and the relative position between adjacent teeth, and extracts the continuous contour segment facing the opposite tooth to obtain the opposing boundary data of adjacent teeth.

[0064] The acquisition of the adjacent tooth opposing boundary data is specifically as follows: determine adjacent tooth instances according to the tooth row order, determine the corresponding adjacency direction according to the relative position between adjacent teeth, and determine the corresponding nearest contour point from the tooth contour of the two teeth respectively. Using the nearest contour point as the basis for local boundary positioning, extract continuous contour segments that face the opposite tooth and contain the corresponding nearest contour point to obtain the adjacent tooth opposing boundary data.

[0065] In specific implementation, contour tracking is performed along two extension directions of the tooth contour, with the corresponding nearest contour point as the center. The contour perimeter is determined according to the tooth contour of the corresponding tooth. The length of the continuous contour segment participating in the adjacent tooth arrangement relationship analysis is limited to 20% to 30% of the corresponding contour perimeter. Preferably, this embodiment takes 25%, that is, a continuous contour with a contour perimeter of 12.5% ​​is extracted from the corresponding nearest contour point in two extension directions of the tooth contour. When the continuous contour changes from facing the opposite tooth to facing away from the opposite tooth before reaching the above length, the position of the direction change is taken as the termination position of the corresponding continuous contour. The local contour segment formed by the continuous contours on both sides is taken as the adjacent tooth opposing boundary of the corresponding tooth, thereby reducing the participation of tooth contours far from the adjacent position in subsequent occlusion relationship recognition.

[0066] Step S22: Occlusion boundary interval identification, used to determine the visual occlusion relationship between adjacent teeth and the boundary interval affected by visual occlusion. Specifically, it identifies the occlusion intersection position based on the adjacent tooth opposing boundary data, and determines the front layer teeth, back layer teeth and occlusion boundary interval to obtain adjacent tooth occlusion structure data.

[0067] The acquisition of the adjacent tooth occlusion structure data is specifically as follows: based on the adjacent tooth opposing boundary data, the image gradient continuity and boundary tangential change corresponding to the opposing boundary in the intraoral image of the occlusal surface are analyzed, and the occlusion intersection position is determined by combining the intersection relationship between the two opposing boundaries. Based on the boundary continuity state on both sides of the occlusion intersection position, the anterior and posterior teeth are determined, and the occlusion boundary interval corresponding to the posterior teeth is determined, thus obtaining the adjacent tooth occlusion structure data.

[0068] In specific implementation, the grayscale values ​​of the intraoral image of the occlusal surface are normalized to 0-1, and the Sobel operator is used to calculate the image gradient near the opposing boundary of the adjacent teeth; for two teeth that are sequentially adjacent in the dentition, the minimum distance between the opposing boundaries of the two adjacent teeth is calculated, and the smaller value of the square root of the area of ​​the crown region of the two teeth is used as the local size reference corresponding to the current adjacent tooth; when the minimum distance between the opposing boundaries of the two adjacent teeth is not greater than 5% of the local size reference, the local image region near the corresponding position is determined as a candidate occlusion intersection region, and the boundary continuity state is judged within the candidate occlusion intersection region;

[0069] Preferably, within the candidate occlusion intersection area, local boundary segments are obtained along the opposing boundaries of adjacent teeth on both sides, located before and after the candidate occlusion intersection position, and the average gradient amplitude and average tangential direction of the corresponding local boundary segments are determined respectively. When the average gradient amplitude of the opposing boundary of one side of the adjacent teeth after the candidate occlusion intersection position is not less than 60% of the average gradient amplitude before the corresponding position, and the angle between the average tangential directions before and after the position is not greater than 30°, the opposing boundary of the adjacent teeth on that side is determined to remain continuous. When the average gradient amplitude of the opposing boundary of the other side of the adjacent teeth after the candidate occlusion intersection position is less than 40% of the average gradient amplitude before the corresponding position, or the angle between the average tangential directions before and after the position is greater than 45°, the opposing boundary of the adjacent teeth on that side is determined to have an occlusion interruption. The teeth corresponding to the adjacent opposing boundaries that remain continuous are determined to be front teeth, the teeth corresponding to the adjacent opposing boundaries that have an occlusion interruption are determined to be back teeth, and the contour missing interval between the position where the boundary of the back teeth has an occlusion interruption and the position where the boundary has resumed continuity is determined to be the occlusion boundary interval.

[0070] As a further optimization of this embodiment, when the opposing boundaries of adjacent teeth on both sides do not meet the occlusion interruption condition, the corresponding adjacent teeth are determined as unoccluded adjacent teeth; when the opposing boundary of one adjacent tooth meets the occlusion interruption condition and the opposing boundary of the other adjacent tooth meets the continuity condition, the corresponding adjacent teeth are determined as occluded adjacent teeth, and the occlusion state, front teeth, back teeth and occlusion boundary interval are associated according to the corresponding adjacent teeth to obtain the adjacent tooth occlusion structure data;

[0071] Step S23: Potential opposing boundary recovery, used to recover the local boundaries of the posterior teeth that cannot be directly observed within the occlusion boundary range. Specifically, it involves establishing and screening candidate recovery boundaries based on the adjacent tooth occlusion structure data to obtain potential opposing boundary data.

[0072] The acquisition of the potential opposing boundary data specifically involves obtaining the visible boundary positions and tangential directions at both ends of the occlusion boundary interval based on the adjacent tooth occlusion structure data, establishing candidate recovery boundaries that satisfy endpoint continuity constraints and tangential continuity constraints between the two visible boundaries, and limiting the allowable distribution range of the candidate recovery boundaries based on the crown region corresponding to the front teeth in the tooth visual structure data and the occlusion boundary interval, and determining the recovery boundaries that conform to the positional continuity and tangential change continuity of the two visible boundaries from the candidate recovery boundaries that satisfy the constraints, thereby obtaining the potential opposing boundary data.

[0073] In practice, for adjacent teeth that are determined to be obscuring adjacent teeth, the positions of the first visible boundaries are determined from both ends of the obscuring boundary interval corresponding to the posterior teeth. Second visible boundary position Continuous visible contours are obtained from the first visible boundary position and the second visible boundary position towards the outside of the occlusion boundary interval, respectively; preferably, the length of the continuous visible contour is 3% of the circumference of the corresponding back tooth contour, and the first unit tangential direction is determined by least-squares straight-line fitting of each continuous visible contour. Second unit tangential direction To reduce the impact of positional fluctuations of individual contour points on the boundary recovery direction;

[0074] A cubic Bezier curve is constructed based on the first visible boundary position P0, the second visible boundary position P1, the first unit tangential direction u0, and the second unit tangential direction u1. This cubic Bezier curve is then used as a candidate recovery boundary, and its calculation formula is as follows:

[0075] ;

[0076] In the formula, B(t) is the position of the candidate recovery boundary corresponding to the curve parameter t; t is the curve parameter of the candidate recovery boundary, with a value range of 0 to 1; The position of the first visible boundary corresponding to the first end of the occluded boundary interval; The first tangential extension length corresponding to the first visible boundary position; The first unit tangential direction corresponding to the first visible boundary position; The position of the second visible boundary corresponding to the second end of the occluded boundary interval; The second tangential extension length corresponding to the second visible boundary position; The second unit tangential direction corresponding to the second visible boundary position;

[0077] Preferably, the first visible boundary position Second visible boundary position The straight-line distance between them is denoted as the shading span D, and the first tangential extension length is varied within the range of 0.25D to 0.75D. Second tangential extension length Candidate recovery boundaries are generated with an adjustment interval of 0.10D; the first tangential extension length is limited. Second tangential extension length The range of values ​​ensures that the candidate restoration boundary continues to extend continuously within the occlusion boundary range, and reduces the situation where the restoration boundary changes sharply due to too small a tangential extension length or deviates from the original tooth contour extension trend due to too large a tangential extension length.

[0078] In the specific screening process, the crown region corresponding to the anterior teeth is obtained based on the visual structure data of the teeth. The square root of the area of ​​the crown region is used as the size reference for the anterior teeth, and the crown region of the anterior teeth is extended outward by 3% of the size reference to form an occlusion tolerance region. The curved portion of the candidate restoration boundary located within the occlusion boundary interval is restricted within the occlusion tolerance region, and the candidate restoration boundaries that meet the allowable distribution range are sampled at equal intervals according to the curve arc length. The average tangential change of the candidate restoration boundary is determined based on the tangential direction change between adjacent sampling positions. The candidate restoration boundary with the smallest average tangential change is determined as the restoration boundary of the corresponding posterior teeth, and the restoration boundary is connected to the visible adjacent tooth opposing boundary outside the occlusion boundary interval of the posterior teeth to form the potential opposing boundary of the corresponding posterior teeth. For unoccluded adjacent teeth, the candidate restoration boundary construction is not performed, and the adjacent tooth opposing boundary obtained in step S21 is directly used as the potential opposing boundary of the corresponding teeth. The potential opposing boundary data is formed by the restored posterior boundary corresponding to the occluded adjacent teeth and the original adjacent tooth opposing boundary corresponding to the unoccluded adjacent teeth.

[0079] Step S24: Determine potential adjacency relationship, which is used to determine the arrangement relationship between adjacent teeth based on the restored opposing boundaries. Specifically, it determines the potential adjacency position, the marked adjacency distance, and the adjacency restoration confidence, and obtains the adjacent tooth arrangement relationship data.

[0080] The acquisition of the adjacent tooth arrangement relationship data specifically involves forming two opposing boundaries for adjacency analysis based on the adjacent tooth opposing boundary data and potential opposing boundary data, determining the potential adjacency position and the signed adjacency distance based on the relative position between the two opposing boundaries, determining the adjacency recovery confidence based on the support of the visible boundary used for boundary recovery and the continuity deviation between the recovered boundary and the visible boundary, and correlating the potential opposing boundaries, potential adjacency positions, signed adjacency distances and adjacency recovery confidence to obtain the adjacent tooth arrangement relationship data.

[0081] In practice, for the i-th and j-th teeth that are sequentially adjacent in the dentition, a unit adjacency direction from the i-th tooth to the j-th tooth is established based on the relative position between the two teeth. ; from the potential opposing boundary corresponding to the i-th tooth Determine along the unit adjacency direction The first adjacent boundary position with the largest projection And from the potential opposing boundary corresponding to the j-th tooth Determine along the unit adjacency direction The second adjacent boundary position with the smallest projection The first adjacent boundary position Second adjacent boundary position The midpoint between them serves as the potential adjacent position of the corresponding adjacent teeth;

[0082] Based on the potential opposing boundary corresponding to the i-th tooth The potential opposing boundary corresponding to the j-th tooth And the signed adjacency distance of the corresponding adjacent teeth is calculated using the unit adjacency direction eij, and the calculation formula is as follows:

[0083] ;

[0084] In the formula, Let q be the signed adjacency distance between the i-th tooth and the j-th tooth; q is the potential opposing boundary corresponding to the j-th tooth. The boundary position on; This represents the unit adjacency direction from the i-th tooth to the j-th tooth; p represents the potential opposing boundary corresponding to the j-th tooth; p represents the potential opposing boundary corresponding to the i-th tooth. The boundary position on; Let be the potential opposing boundary corresponding to the i-th tooth; min(·) represents obtaining the potential opposing boundary of the j-th tooth along the unit adjacency direction. The minimum projection value; max(·) represents obtaining the potential opposing boundary of the i-th tooth along the unit adjacency direction. Maximum projection value; when the signed adjacency distance A value greater than 0 indicates that the potential opposing boundaries on both sides remain separated along the adjacency direction; when the signed adjacency distance is greater than 0, it indicates that the potential opposing boundaries on both sides remain separated along the adjacency direction. When the value is 0, it indicates that the potential opposing boundaries on both sides are in an adjacent contact state; when the signed adjacency distance is 0, it indicates that the potential opposing boundaries on both sides are in an adjacent contact state. When the value is less than 0, it indicates that there is a projection overlap between the potential opposing boundaries on both sides along the adjacent direction;

[0085] In practical implementation, for adjacent teeth with potential opposing boundary restoration, the reliability of adjacent restoration is determined based on the visible boundary support of the restoration boundary and the continuity deviation between the restoration boundary and the visible boundary. The ratio of the total arc length of the continuous visible contour used to determine the restoration boundary at both ends of the occluded boundary interval to the sum of the total arc length of the continuous visible contour and the arc length of the restoration boundary is determined as the visible boundary support ratio. Determine the tangential direction difference between the two ends of the restored boundary and the adjacent continuous visible contours, divide the average of the two tangential direction differences by 30° and limit it to 0~1 to obtain the continuity deviation. According to the visible boundary support ratio and continuity deviation The formula for calculating the reliability of adjacency recovery is as follows:

[0086] ;

[0087] In the formula, The reliability of the adjacency recovery between the i-th tooth and the j-th tooth is given, with a value ranging from 0 to 1. The visible boundary support ratio is the corresponding restored boundary between the i-th tooth and the j-th tooth, and its value ranges from 0 to 1. The value ranges from 0 to 1 to represent the continuity deviation between the restored boundary and the continuous visible contour; 0.6 is the weight corresponding to the visible boundary support ratio, used to improve the influence of the true visible boundary on the reliability of adjacent restoration; 0.4 is the weight corresponding to the continuity deviation, used to characterize the connection stability between the restored boundary and the original visible boundary; for unoccluded adjacent teeth, since the corresponding potential opposing boundary directly originates from the visible adjacent tooth opposing boundary and does not involve the restoration of occluded boundaries, the reliability of the corresponding adjacent restoration is... Set to 1, and associate the potential opposing boundaries, potential adjacent positions, marked adjacent distances, and adjacent recovery confidence of each adjacent tooth according to the tooth row order to obtain the adjacent tooth arrangement relationship data.

[0088] Example 4, see Figure 1 , Figure 2 and Figure 4 This embodiment is based on the above embodiment. In step S3, the construction of the relative alignment reference of the dental arch is used to establish a dental arch alignment reference that reduces the influence of local scale differences in intraoral images. Specifically, it determines the local scale of the teeth and the adjacent position chain of the dental arch based on the visual structure data of the teeth and the adjacent tooth alignment relationship data, constructs the dental arch reference trajectory, and establishes the relative alignment representation of each tooth relative to the dental arch reference trajectory to obtain the dental arch alignment reference data.

[0089] The acquisition of the tooth alignment reference data specifically involves determining the local scale of each tooth based on the tooth visual structure data and adjacent tooth alignment relationship data, connecting potential adjacent positions that meet preset confidence conditions according to the tooth row sequence to form a tooth row adjacent position chain, and performing scale-normalized robust curve fitting on the tooth row adjacent position chain based on the local scale of the corresponding tooth to obtain the dental arch reference trajectory. Based on the position and direction relationship of each tooth relative to the dental arch reference trajectory, a relative alignment representation of the dental arch is established to obtain the tooth alignment reference data.

[0090] The construction of the relative alignment reference of the dental arches includes the following steps:

[0091] Step S31: Determine the local scale of teeth, which is used to establish the local scale benchmark corresponding to different tooth positions in the intraoral image. Specifically, the local scale of each tooth is determined according to the recovered tooth contour to obtain the local scale data of teeth.

[0092] The acquisition of the local scale data of the teeth is specifically as follows: extracting potential opposing boundaries from the adjacent tooth arrangement relationship data, supplementing the tooth contour affected by visual occlusion in the tooth visual structure data according to the potential opposing boundaries, updating the crown region of the corresponding tooth according to the supplemented tooth contour, obtaining the restored crown region, and determining the local scale of the corresponding tooth according to the image area of ​​the restored crown region, thereby obtaining the local scale data of the teeth.

[0093] In practice, the image area is calculated based on the restored crown region corresponding to the i-th tooth, and the square root of the image area is used as the local tooth scale for the corresponding tooth. The calculation formula is as follows:

[0094] ;

[0095] In the formula, This represents the local scale of the i-th tooth, in pixels. denoted as the image area of ​​the recovered crown region corresponding to the i-th tooth, in square pixels; i is the tooth number determined according to the dentition sequence; by converting the two-dimensional crown region area into a tooth local scale with length dimensions, subsequent distance and positional deviations can be normalized according to the actual imaging size of different teeth in intraoral images.

[0096] Step S32: Establish the dental arch adjacency position chain to form an arrangement structure that is continuously distributed along the dental arch direction and has adjacency recovery confidence constraints. Specifically, establish the dental arch adjacency position chain based on the potential adjacency positions and the corresponding adjacency recovery confidence to obtain dental arch adjacency position chain data.

[0097] The acquisition of the dental arch adjacency position chain data specifically involves extracting potential adjacency positions and adjacency recovery confidence from the adjacent tooth arrangement relationship data, arranging consecutive potential adjacency positions according to the dental arch order, filtering out potential adjacency positions that do not meet the preset confidence conditions based on the adjacency recovery confidence, and associating and connecting the remaining potential adjacency positions and their corresponding adjacency recovery confidence according to the dental arch order to obtain the dental arch adjacency position chain data.

[0098] Preferably, the adjacency recovery confidence level obtained in step S24 is not less than 0.60, which is used as the preset confidence level condition for adjacency recovery confidence level. Potential adjacent positions are preserved, and the corresponding adjacent recovery confidence is recorded. For unobstructed adjacent teeth, since the corresponding adjacent recovery confidence is set to 1, their potential adjacent positions directly participate in the establishment of the dental arch adjacent position chain. For potential adjacent positions with an adjacent recovery confidence of less than 0.60, they do not participate in the subsequent dental arch reference trajectory fitting, so as to reduce the impact of unstable positions of the occlusion boundary recovery on the dental arch alignment reference.

[0099] Step S33: Dental arch reference trajectory construction, used to establish a dental arch reference trajectory that reduces the influence of locally misaligned teeth on the dental arch alignment reference benchmark. Specifically, the scale-normalized weighted trajectory fitting is performed on the adjacent position chain of the dental arch to obtain dental arch reference trajectory data.

[0100] The acquisition of the dental arch reference trajectory data specifically involves, based on the dental row adjacency chain data and tooth local scale data, and on the tooth local scale of the two adjacent teeth corresponding to each potential adjacency position, normalizing the positional deviation of each potential adjacency position that meets the preset confidence condition relative to the candidate dental arch trajectory, and adjusting the fitting weight of different potential adjacency positions participating in the dental arch trajectory fitting according to the adjacency recovery confidence, thereby reducing the influence of abnormal deviation positions while limiting the local curvature change of the candidate dental arch trajectory, thus obtaining the dental arch reference trajectory data.

[0101] In specific implementation, a cubic B-spline curve is used as the candidate dental arch trajectory, preferably with 6 control points, and open uniform nodes are set according to the arrangement direction from the beginning to the end of the dental arch adjacency position chain; for the k-th potential adjacency position participating in the fitting, the average local scale of the corresponding two adjacent teeth is used as the scale benchmark corresponding to the potential adjacency position, and the shortest distance from the potential adjacency position to the candidate dental arch trajectory is divided by the corresponding scale benchmark to obtain the scale-normalized position deviation; the adjacency recovery confidence obtained in step S24 is used as the fitting weight of the corresponding potential adjacency position, and Huber loss is used to reduce the influence of local obvious deviation positions on the overall trajectory fitting, and the conversion threshold of Huber loss is preferably 1.5;

[0102] Preferably, the dental arch reference trajectory is determined by the following objective function:

[0103] ;

[0104] In the formula, The fitted dental arch reference trajectory is denoted as C; the candidate dental arch trajectory to be optimized is denoted as C; k is the potential adjacent position number participating in the trajectory fitting; and M is the total number of potential adjacent positions participating in the trajectory fitting. The fitting weights corresponding to the kth potential adjacent positions are taken as the adjacent recovery confidence. Huber loss function; The shortest distance from the kth potential adjacent position to the candidate dental arch trajectory C; The average local scale of the teeth corresponding to the kth potential adjacency position is the average value of the two adjacent teeth. The trajectory smoothing constraint coefficient is preferably set to 0.10 in this embodiment; m is the control point number of the candidate dental arch trajectory; K is the total number of control points of the candidate dental arch trajectory, which is set to 6 in this embodiment; , and These are the position vectors of the (m+1)th, mth, and (m-1)th control points, respectively. The average local scale of the teeth involved in the trajectory fitting is used; the dental arch reference trajectory data is obtained by limiting the scale normalization deviation between the dental arch reference trajectory and the credible potential adjacent positions through the first term, and by limiting the local bending changes formed by adjacent control points through the second term.

[0105] Step S34: Relative alignment characterization of dental arches, used to convert the tooth positions described based on image coordinates into alignment relationships relative to the dental arch reference trajectory. Specifically, it involves determining the alignment index of each tooth relative to the dental arch reference trajectory to obtain tooth alignment baseline data.

[0106] The acquisition of the tooth alignment reference data specifically involves determining the trajectory projection position of each tooth based on the dental arch reference trajectory data, and determining the tangential and normal directions of the dental arch corresponding to the trajectory projection position. The spatial relationship between the teeth and the dental arch reference trajectory is normalized based on the tooth visual structure data and tooth local scale data. The corresponding dental arch arc length position, dental arch normal offset, and crown direction offset are determined for each tooth. The normalized adjacency distance is determined based on the signed adjacency distance in the adjacent tooth alignment relationship data and the corresponding local scale of two adjacent teeth. The dental arch reference trajectory, tooth local scale, dental arch arc length position, dental arch normal offset, crown direction offset, and normalized adjacency distance are then correlated to obtain the tooth alignment reference data.

[0107] In practice, the geometric center of the crown region is determined based on the restored crown region corresponding to the i-th tooth. and the geometric center of the crown area The nearest position on the dental arch reference trajectory is determined as the trajectory projection position. Based on the dental arch reference trajectory, at the trajectory projection position The direction of the curve at that point determines the tangential direction of the unit dental arch. And determine the tangential direction relative to the unit dental arch. The normal direction of the unit dental arch that is perpendicular to the inside of the dental arch and points from the outside of the dental arch. Use one end of the dental arch reference trajectory as the starting position of the arc length. The arch length, arch normal offset, crown offset, and normalized adjoint distance are calculated according to the unified dentition direction. The calculation method is as follows:

[0108] ;

[0109] ;

[0110] ;

[0111] In the formula, This represents the position of the dental arch arc length corresponding to the i-th tooth; Starting position along the dental arch reference trajectory from the arc length To the trajectory projection position The arc length of the trajectory; This is the starting position of the arc length of the dental arch reference trajectory; Let be the trajectory projection position corresponding to the i-th tooth; This represents the total arc length of the dental arch reference trajectory; This represents the normal offset of the dental arch corresponding to the i-th tooth; Let be the geometric center of the crown region corresponding to the i-th tooth; For trajectory projection position The corresponding unit dental arch normal direction; Let be the local scale of the tooth corresponding to the i-th tooth; The crown direction offset corresponding to the i-th tooth; Let be the crown direction vector corresponding to the i-th tooth in the visual structure data of teeth; For trajectory projection position The corresponding unit dental arch tangential direction; |·| is the absolute value operation; ||·|| is the vector magnitude; Let be the normalized adjacency distance between the i-th tooth and the j-th tooth; The signed adjacency distance between the i-th tooth and the j-th tooth obtained in step S24; Here, represents the local dimensions of the j-th tooth; j is the tooth number adjacent to the i-th tooth in the dental arch sequence; the position of the dental arch arc length... Limited to 0 to 1, the dental arch normal offset and normalized adjacency distance Both positive and negative signs are retained to represent the normal positional relationship of the teeth relative to the dental arch reference trajectory and the spatial relationship between adjacent teeth, respectively.

[0112] By performing the above operations, this scheme addresses the issue of local imaging scale differences at different tooth positions in intraoral images, and the problem that locally misaligned teeth can easily affect the overall dental arch alignment benchmark. It utilizes adjacency recovery confidence to screen potential adjacency positions and combines this with robust dental arch trajectory fitting based on local tooth scale normalization, ensuring that the dental arch reference trajectory primarily reflects the overall dental arch alignment trend. Furthermore, the position, orientation, and adjacency relationship of each tooth in the original image coordinates are converted into the dental arch arc length position, dental arch normal offset, crown direction offset, and normalized adjacency distance relative to the dental arch reference trajectory. This reduces the impact of local image scale differences on subsequent alignment anomaly judgment and provides a unified relative dental arch alignment benchmark for verifying local alignment anomalies.

[0113] Example 5, see Figure 1 , Figure 2 and Figure 5 This embodiment is based on the above embodiment. In step S4, the local arrangement abnormality verification is used to distinguish different arrangement abnormality structural mechanisms after detecting local arrangement abnormality. Specifically, it determines the tooth segment to be verified based on the tooth arrangement benchmark data, establishes candidate counterfactual arrangement states corresponding to different arrangement abnormality structural mechanisms, evaluates the recovery of local arrangement abnormality by each candidate counterfactual arrangement state, determines the arrangement abnormality structural mechanism, and obtains tooth arrangement abnormality verification data.

[0114] The acquisition of the dental alignment abnormality verification data specifically involves determining the teeth to be verified and the relationship between abnormal adjacent teeth based on the dental alignment benchmark data, forming the tooth segments to be verified based on the continuity of the dental arch, establishing corresponding candidate counterfactual alignment states according to tooth rotation mechanism, tooth misalignment mechanism along the dental arch normal direction, and dental arch tangential spatial allocation abnormality mechanism, evaluating the degree of recovery of the current local alignment abnormality by different candidate counterfactual alignment states while keeping the dental arch reference trajectory and tooth local scale unchanged, determining the structural mechanism of the alignment abnormality based on the degree of recovery and the corresponding normalization adjustment amplitude, and obtaining the dental alignment abnormality verification data.

[0115] The verification of local arrangement anomalies includes the following steps:

[0116] Step S41: Determine the abnormal arrangement to be verified, which is used to locate the teeth and tooth segments that need to be verified for abnormal structural mechanisms in the dental arch. Specifically, it is to determine the relationship between the teeth to be verified and the abnormal adjacent teeth according to the tooth arrangement reference data, and form the tooth segments to be verified to obtain the tooth segment data to be verified.

[0117] The acquisition of the tooth segment data to be verified is specifically as follows: based on the dental arch normal offset and crown direction offset in the tooth arrangement reference data, the teeth that deviate from the corresponding normal reference range are determined; based on the normalized adjacency distance, the abnormal adjacent tooth relationships that deviate from the corresponding normal reference range are determined; the corresponding teeth and the teeth involved in the abnormal adjacent tooth relationships are determined as the teeth to be verified; and the tooth segment to be verified is formed based on the continuous distribution of the teeth to be verified and the abnormal adjacent tooth relationships in the dental arch direction, thus obtaining the tooth segment data to be verified.

[0118] In specific implementation, the normal reference range is established in advance based on normal dentition samples. In this embodiment, 200 intraoral images of the occlusal surface, which have been manually confirmed to have no obvious tooth rotation, arch misalignment, or crowding or gaps in the tooth segments, are used as normal dentition samples. The arch length is used as the position index of the teeth in the dentition. The arch normal offset and crown offset are calculated according to the position interval of the arch length corresponding to the tooth. The normal adjacency distance is calculated according to the position interval of the median value of the arch length corresponding to the adjacent teeth. Preferably, the arch length in the range of 0 to 1 is divided into 10 equal-width position intervals, and the 5th percentile to 95th percentile values ​​of the normal sample distribution corresponding to each arrangement index are determined as the corresponding normal reference ranges. In subsequent processing, the corresponding normal reference range is called according to the position interval of the arch length corresponding to the tooth to be identified or the adjacent teeth, so as to avoid the difference in normal arrangement of different dentition positions affecting the abnormal screening results.

[0119] Preferably, when the dental arch normal deviation or crown direction deviation of a tooth exceeds the corresponding normal reference range, the tooth is identified as a tooth to be verified; when the normalized adjacency distance of an adjacent tooth exceeds the corresponding normal reference range, the adjacent tooth relationship is identified as an abnormal adjacent tooth relationship; for tooth ranges formed by continuous connection of the same tooth to be verified, adjacent teeth to be verified, or abnormal adjacent tooth relationships in the dentition direction, they are merged into a tooth segment to be verified, and the tooth number, abnormal arrangement index, and abnormal adjacent tooth relationship corresponding to the tooth segment to be verified are associated to obtain the tooth segment data to be verified.

[0120] Step S42: Mechanism constraint reverse arrangement adjustment, used to establish comparable counterfactual states corresponding to different abnormal arrangement structural mechanisms. Specifically, reverse adjustment of tooth rotation, reverse adjustment of teeth along the dental arch normal, and reverse adjustment of dental arch tangential spatial allocation are performed on the tooth segment to be verified to obtain candidate counterfactual arrangement state data.

[0121] The acquisition of the candidate counterfactual arrangement data specifically involves establishing reverse adjustments for tooth rotation, reverse adjustments along the dental arch normal, and reverse adjustments for dental arch tangential spatial allocation based on the data of the tooth segment to be verified. Specifically, the reverse adjustment for tooth rotation only changes the crown direction offset of the tooth to be verified relative to the dental arch reference trajectory, while maintaining the corresponding tooth's dental arch arc length position and dental arch normal offset. The reverse adjustment for tooth rotation only changes the dental arch normal offset of the tooth to be verified, while maintaining the corresponding tooth's dental arch arc length position and crown direction offset. The reverse adjustment for dental arch tangential spatial allocation only performs a local spatial redistribution of the teeth in the tooth segment to be verified along the dental arch tangential direction, while maintaining the corresponding tooth's dental arch normal offset and crown direction offset, thus obtaining the candidate counterfactual arrangement data.

[0122] In practice, for reverse adjustment of tooth rotation, the crown direction that exceeds the corresponding normal reference range is shifted towards the boundary of the normal reference range closest to the current value; for reverse adjustment of teeth along the dental arch normal, the dental arch normal that exceeds the corresponding normal reference range is shifted towards the boundary of the normal reference range closest to the current value.

[0123] For the reverse adjustment of the tangential spatial allocation of the dental arch, when the segment to be verified contains more than three teeth, the arch length positions of the teeth at both ends of the segment to be verified are kept unchanged, and the arch length between the two ends of the teeth is redistributed according to the proportion of the average local scale of adjacent teeth in the segment to be verified; when the segment to be verified contains only two teeth, the arch length positions corresponding to the nearest teeth on both sides of the segment to be verified that have not been identified as the teeth to be verified are used as fixed boundaries; when there are no corresponding teeth on one side, the endpoint of the reference trajectory of the dental arch on the corresponding side is used as a fixed boundary, and the arch length positions of the teeth to be verified are redistributed between the fixed boundaries; so that the adjusted teeth maintain the original dentition order along the dental arch direction, and the target arch length positions corresponding to each tooth are obtained;

[0124] Preferably, each reverse adjustment parameter is adjusted hierarchically from the current arrangement state to the corresponding target arrangement state, generating multiple candidate counterfactual arrangement states. The calculation method is as follows:

[0125] ;

[0126] In the formula, This is the parameter value of the h-th adjustment parameter after the r-th level reverse adjustment; This is the parameter value of the h-th adjustment parameter before reverse adjustment; This represents the adjustment ratio corresponding to the r-th level reverse adjustment; Here is the target parameter value corresponding to the h-th adjustment parameter; r is the adjustment level number, ranging from 1 to 10; h is the parameter number participating in the reverse adjustment under the current abnormal arrangement structure mechanism; when performing reverse adjustment of tooth rotation, the adjustment parameter is the crown direction offset; when performing reverse adjustment of teeth along the dental arch normal, the adjustment parameter is the dental arch normal offset; when performing reverse adjustment of dental arch tangential space allocation, the adjustment parameter is the dental arch arc length position.

[0127] The tooth arrangement relationship is re-determined based on the arrangement parameters corresponding to each adjustment level. Specifically, while keeping the dental arch reference trajectory and local tooth scale unchanged, the trajectory projection position, geometric center of the crown region, and crown direction of the corresponding teeth are re-determined based on the adjusted dental arch arc length position, dental arch normal offset, and crown direction offset. Based on this, the position translation or orientation rotation of the restored crown region and tooth contour of the corresponding teeth are performed to form candidate counterfactual tooth geometry. The opposing boundaries, signed adjacency distance, and normalized adjacency distance of adjacent teeth are re-determined based on the candidate counterfactual tooth geometry to form the candidate counterfactual arrangement state data.

[0128] Step S43: Counterfactual structure recovery evaluation, used to determine the explanatory power of different permutation anomaly mechanisms for the current local permutation anomaly, specifically to evaluate the degree of local permutation anomaly recovery and normalization adjustment magnitude corresponding to each candidate counterfactual permutation state, and obtain counterfactual structure recovery evaluation data;

[0129] The acquisition of the counterfactual structure restoration evaluation data specifically involves, while keeping the dental arch reference trajectory and local tooth scale unchanged, redetermining the corresponding normalized adjacency distance, dental arch normal offset, and crown direction offset according to each candidate counterfactual arrangement state, comparing the redetermined arrangement index with the arrangement index before reverse adjustment, determining the degree of local arrangement anomaly restoration corresponding to each candidate counterfactual arrangement state, and determining the normalized adjustment range according to the parameter adjustment amount required to form the corresponding candidate counterfactual arrangement state, thereby obtaining the counterfactual structure restoration evaluation data.

[0130] In practical implementation, for the arch normal offset, crown direction offset, and normalized adjacency distance in the tooth segment to be verified, the normalized out-of-bounds amount of the arrangement index is calculated based on the corresponding normal reference range. The calculation method is as follows:

[0131] ;

[0132] In the formula, L is the normalized out-of-bounds amount corresponding to the arrangement index x; x is the arch normal offset, crown direction offset, or normalized adjacency distance to be evaluated; L is the lower limit of the normal reference range corresponding to the arrangement index x; U is the upper limit of the normal reference range corresponding to the arrangement index x; when the arrangement index is within the normal reference range, the normalized out-of-bounds amount is 0; when the arrangement index exceeds the normal reference range, the normalized out-of-bounds amount is determined according to the degree of its exceedance relative to the width of the normal reference range.

[0133] The arithmetic mean of the normalized out-of-bounds values ​​of the arch normal offset and crown direction offset of each tooth in the same tooth segment to be verified, as well as the normalized adjacency distances corresponding to each abnormal adjacent tooth relationship, is used to obtain the degree of local arrangement abnormality corresponding to the current arrangement state; the degree of local arrangement abnormality before reverse adjustment is denoted as... Let the degree of local permutation anomaly corresponding to the m-th candidate counterfactual permutation state be denoted as And calculate the corresponding degree of recovery from local arrangement anomalies:

[0134] ;

[0135] In the formula, The degree of local permutation anomaly recovery corresponding to the m-th candidate counterfactual permutation state; To determine the degree of local alignment abnormalities in the tooth segment to be verified before reverse adjustment; The degree of local permutation anomaly corresponding to the m-th candidate counterfactual permutation state; To prevent positive numbers with a denominator of zero, this implementation uses 10^-6; m is the candidate counterfactual arrangement state number; the degree of local arrangement anomaly recovery. The larger the value, the higher the degree to which the corresponding candidate counterfactual arrangement restores the original local arrangement anomaly;

[0136] Simultaneously, the normalized adjustment magnitude is calculated based on the changes in adjustment parameters required to form the corresponding candidate counterfactual arrangement state. The calculation method is as follows:

[0137] ;

[0138] In the formula, The normalized adjustment magnitude corresponds to the m-th candidate counterfactual arrangement state; is the number of parameters involved in the reverse adjustment in the m-th candidate counterfactual arrangement; h is the parameter number involved in the reverse adjustment; The change of the h-th adjustment parameter in the m-th candidate counterfactual arrangement state relative to the state before the reverse adjustment; This represents the normalized scale corresponding to the h-th adjustment parameter; when the adjustment parameter is crown direction offset or arch normal offset... Take the width corresponding to the normal reference range; when adjusting the parameter to the position of the dental arch arc length, The average value of the arch length position interval of adjacent teeth before reverse adjustment of the tooth segment to be verified is taken; through the normalization process, the angle adjustment amount and position adjustment amount corresponding to different abnormal arrangement structural mechanisms can be compared on a unified scale to obtain the counterfactual structure restoration evaluation data;

[0139] Step S44: Verification of abnormal arrangement structure mechanism, used to determine the corresponding structure mechanism that can explain the current abnormal arrangement from multiple candidate abnormal arrangement structure mechanisms. Specifically, it is to determine the abnormal arrangement structure mechanism and abnormal tooth or abnormal tooth segment corresponding to the tooth segment to be verified based on the counterfactual structure recovery evaluation data, and obtain the tooth row abnormality verification data.

[0140] The acquisition of the dental alignment abnormality verification data specifically involves comparing the degree of local alignment abnormality recovery and the normalized adjustment range corresponding to different candidate counterfactual alignment states based on the counterfactual structure recovery evaluation data. The candidate counterfactual alignment state that meets the preset recovery requirements for local alignment abnormality recovery and has the smallest normalized adjustment range is determined as the target candidate counterfactual alignment state. The alignment abnormality structure mechanism corresponding to the target candidate counterfactual alignment state is determined as the alignment abnormality structure mechanism corresponding to the tooth segment to be verified. Abnormal teeth or abnormal tooth segments are determined based on the adjusted tooth range in the target candidate counterfactual alignment state. The corresponding tooth segment to be verified, abnormal teeth or abnormal tooth segments, alignment abnormality structure mechanism, and reverse adjustment results are recorded to obtain the dental alignment abnormality verification data.

[0141] Preferably, the degree of restoration of local arrangement anomalies As the preset recovery requirement, a normalized adjustment magnitude is selected from the candidate counterfactual arrangement states that meet the preset recovery requirement. The smallest candidate counterfactual arrangement is selected as the target candidate counterfactual arrangement. When the difference in normalized adjustment magnitude between two candidate counterfactual arrangements is no greater than 0.05, the candidate counterfactual arrangement with a greater degree of local arrangement anomaly recovery is selected. When all candidate counterfactual arrangements corresponding to the same tooth segment to be verified do not meet the preset recovery requirements, the single arrangement anomaly structure mechanism is not determined, and the corresponding tooth segment to be verified is marked as an arrangement anomaly to be verified. At the same time, its original arrangement index and the counterfactual structure recovery evaluation data corresponding to each candidate counterfactual arrangement are written into the tooth arrangement anomaly verification data.

[0142] By performing the above operations, this solution addresses the issue that tooth rotation, dental arch normal misalignment, and abnormal tangential spatial allocation of the dental arch may form similar local adjacency anomalies in two-dimensional intraoral images. It establishes candidate counterfactual arrangement states corresponding to different arrangement anomaly structural mechanisms by restricting the degrees of freedom allowed to change during different reverse adjustment processes. Furthermore, it evaluates the degree of recovery and normalized adjustment magnitude of local arrangement anomalies within a unified normal reference range. This ensures that the determination of arrangement anomaly structural mechanisms considers both whether the anomaly can be recovered by the corresponding mechanism and the amount of structural adjustment required to achieve that recovery, thereby reducing confusion caused by directly determining anomaly types based solely on a single arrangement index.

[0143] Example 6, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S5, the tooth arrangement status identification is used to determine the tooth arrangement status according to the verified abnormal arrangement structure mechanism and corresponding arrangement index. Specifically, it is to determine the normal arrangement status and the corresponding abnormal arrangement status according to the tooth arrangement benchmark data and the tooth arrangement abnormal verification data, and obtain the tooth arrangement status identification result.

[0144] The acquisition of the dentition alignment status recognition result specifically involves determining the abnormal alignment structure mechanism corresponding to each abnormal tooth or segment based on the dentition alignment baseline data and dentition alignment abnormality verification data. Abnormal segments verified by the abnormal arch tangential space allocation mechanism are divided according to the deviation direction of the normalized adjacency distance relative to the corresponding normal reference range. The range of teeth corresponding to abnormal adjacent tooth relationships with continuous and consistent deviation directions and exhibiting adjacent space compression is determined as a crowded state. The range of teeth corresponding to abnormal adjacent tooth relationships with continuous and consistent deviation directions and exhibiting adjacent space abundance is determined as a gap state. Abnormal teeth verified by the tooth rotation mechanism are determined as tooth rotation states. Abnormal teeth verified by the tooth misalignment mechanism along the dental arch normal are determined as tooth misalignment states along the dental arch normal. Teeth with dental arch normal offset, crown direction offset, and normalized adjacency distance of the corresponding adjacent tooth relationship all meet the corresponding normal reference range and are not determined as abnormal alignment states are determined as normal alignment states, thus obtaining the dentition alignment status recognition result.

[0145] Example 7, see Figure 1 and Figure 2 Based on the above embodiments, this invention also provides a tooth alignment status recognition system based on intraoral images, including a tooth visual structure extraction module, an adjacent tooth occlusion alignment relationship recovery module, a dental arch relative alignment benchmark construction module, a local alignment anomaly verification module, and a tooth alignment status recognition module.

[0146] The tooth visual structure extraction module is used to establish the visual structure basis of each tooth in the intraoral image, obtain tooth visual structure data through tooth visual structure extraction, and send the tooth visual structure data to the adjacent tooth occlusion arrangement relationship recovery module.

[0147] The adjacent tooth occlusion arrangement relationship recovery module is used to reduce the impact of mutual tooth occlusion on the recognition of interdental arrangement relationship. Through adjacent tooth occlusion arrangement relationship recovery, adjacent tooth arrangement relationship data is obtained, and the adjacent tooth arrangement relationship data is sent to the dental arch relative arrangement benchmark construction module.

[0148] The dental arch relative alignment reference construction module is used to establish a dental arch alignment reference that reduces the influence of local scale differences in intraoral images. Through the construction of the dental arch relative alignment reference, the dental alignment reference data is obtained, and the dental alignment reference data is sent to the local alignment anomaly verification module.

[0149] The local arrangement anomaly verification module is used to distinguish the formation mechanism corresponding to the local arrangement anomaly, obtain the dental arrangement anomaly verification data through local arrangement anomaly verification, and send the dental arrangement anomaly verification data to the dental arrangement state recognition module;

[0150] The tooth alignment state recognition module is used to generate a tooth alignment state recognition result, obtain the tooth alignment state recognition result through tooth alignment state recognition, and output the tooth alignment state recognition result.

[0151] 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 process, method, article, or apparatus.

[0152] 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.

[0153] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A method for recognizing the tooth alignment state based on intraoral images, characterized in that: The method includes the following steps: Step S1: Extract the visual structure of teeth to obtain visual structure data of teeth; Step S2: Restore the occlusion relationship of adjacent teeth. Based on the tooth visual structure data, determine the opposing boundaries of adjacent teeth, identify the visual occlusion between opposing boundaries and restore the opposing boundaries affected by the visual occlusion, determine the potential adjacency positions, marked adjacency distances and adjacency restoration confidence between adjacent teeth, and obtain the adjacent tooth arrangement relationship data. Step S3: Construction of relative dental arch alignment benchmark. Based on the tooth visual structure data and adjacent tooth alignment relationship data, the local scale of the teeth and the adjacent position chain of the dental arch are determined, the dental arch reference trajectory is constructed, and the relative dental arch alignment representation of each tooth relative to the dental arch reference trajectory is established to obtain the tooth alignment benchmark data. This includes the following steps: determination of local tooth scale, establishment of adjacent position chain of the dental arch, construction of dental arch reference trajectory, and representation of relative dental arch alignment. Step S4: Verification of local alignment anomalies. Based on the tooth alignment benchmark data, determine the tooth segment to be verified, establish candidate counterfactual alignment states corresponding to different alignment anomaly structural mechanisms, evaluate the recovery of local alignment anomalies by each candidate counterfactual alignment state, determine the alignment anomaly structural mechanism, and obtain tooth alignment anomaly verification data. Step S5: Identify the tooth alignment status and obtain the tooth alignment status identification result.

2. The method for recognizing tooth alignment based on intraoral images according to claim 1, characterized in that: In step S1, the tooth visual structure extraction involves acquiring and correcting the original intraoral image of the occlusal surface corresponding to the dental arch to be identified, extracting the crown region, tooth outline, and crown direction of each tooth, and determining the dentition order to obtain tooth visual structure data.

3. The method for recognizing tooth alignment based on intraoral images according to claim 2, characterized in that: In step S2, the acquisition of the adjacent tooth arrangement relationship data involves determining instances of teeth that are sequentially adjacent based on the tooth visual structure data, identifying opposing boundaries that face each other from the tooth contours of adjacent teeth, identifying the occlusion level of visual occlusions between the opposing boundaries, restoring the continuity of opposing boundaries affected by visual occlusions, determining the potential adjacency positions and marked adjacency distances between adjacent teeth based on the restored opposing boundaries, and determining the adjacency restoration reliability based on the restoration of opposing boundaries, thereby obtaining the adjacent tooth arrangement relationship data.

4. The method for recognizing tooth alignment based on intraoral images according to claim 3, characterized in that: The restoration of the adjacent tooth occlusion arrangement includes: Step S21: Determine the adjacent tooth opposing boundary. Based on the dentition sequence and the relative position between adjacent teeth, determine the corresponding adjacency direction and extract the continuous contour segment facing the opposite tooth to obtain the adjacent tooth opposing boundary data. Step S22: Identify the occlusion boundary interval. Identify the occlusion intersection position based on the adjacent tooth opposing boundary data, and determine the anterior teeth, posterior teeth and occlusion boundary interval to obtain adjacent tooth occlusion structure data.

5. The method for recognizing tooth alignment based on intraoral images according to claim 4, characterized in that: In step S2, the restoration of the adjacent tooth occlusion arrangement relationship further includes: Step S23: Potential opposing boundary recovery. Based on the adjacent tooth occlusion structure data, candidate recovery boundaries are established and screened to obtain potential opposing boundary data. Step S24: Determine potential adjacency relationships, determine potential adjacency locations, marked adjacency distances, and adjacency recovery reliability, and obtain adjacent tooth arrangement relationship data.

6. The method for recognizing tooth alignment based on intraoral images according to claim 5, characterized in that: In step S3, the local tooth scale is determined by determining the local tooth scale corresponding to each tooth based on the restored tooth contour, thereby obtaining local tooth scale data. The dental row adjacency position chain is established by establishing the dental row adjacency position chain based on the potential adjacency positions and the corresponding adjacency recovery confidence, thereby obtaining dental row adjacency position chain data; The dental arch reference trajectory is constructed by performing scale-normalized weighted trajectory fitting on the adjacent position chain of the dental arch to obtain dental arch reference trajectory data. The relative arrangement of the dental arches is characterized to determine the arrangement index of each tooth relative to the dental arch reference trajectory, thereby obtaining the tooth arrangement benchmark data.

7. The method for recognizing tooth alignment based on intraoral images according to claim 6, characterized in that: The verification of local arrangement anomalies includes: Step S41: Determine the abnormal tooth arrangement to be verified. Based on the tooth arrangement reference data, determine the relationship between the tooth to be verified and the abnormal adjacent teeth, and form the tooth segment to be verified to obtain the tooth segment data. Step S42: Mechanism constraint reverse arrangement adjustment, for the tooth segment to be verified, reverse adjustment of tooth rotation, reverse adjustment of teeth along the dental arch normal and reverse adjustment of dental arch tangential spatial allocation are performed respectively to obtain candidate counterfactual arrangement state data; Step S43: Counterfactual structure recovery evaluation, evaluate the degree of recovery of local permutation anomalies and the normalization adjustment magnitude corresponding to each candidate counterfactual permutation state, and obtain counterfactual structure recovery evaluation data; Step S44: Verification of abnormal arrangement structure mechanism. Based on the counterfactual structure recovery evaluation data, determine the abnormal arrangement structure mechanism and abnormal teeth or abnormal tooth segments corresponding to the tooth segment to be verified, and obtain the verification data of abnormal tooth arrangement.

8. A dental alignment recognition system based on intraoral images, used to implement the dental alignment recognition method based on intraoral images as described in any one of claims 1-7, characterized in that: It includes a module for extracting the visual structure of teeth, a module for restoring the arrangement relationship of adjacent teeth that are obscured, a module for constructing the relative arrangement benchmark of the dental arch, a module for verifying local arrangement anomalies, and a module for recognizing the arrangement status of the dental arch.

9. The dental alignment recognition system based on intraoral images according to claim 8, characterized in that: The tooth visual structure extraction module is used to establish the visual structure basis of each tooth in the intraoral image, obtain tooth visual structure data through tooth visual structure extraction, and send the tooth visual structure data to the adjacent tooth occlusion arrangement relationship recovery module. The adjacent tooth occlusion arrangement relationship recovery module is used to reduce the impact of mutual tooth occlusion on the recognition of interdental arrangement relationship. Through adjacent tooth occlusion arrangement relationship recovery, adjacent tooth arrangement relationship data is obtained, and the adjacent tooth arrangement relationship data is sent to the dental arch relative arrangement benchmark construction module. The dental arch relative alignment reference construction module is used to establish a dental arch alignment reference that reduces the influence of local scale differences in intraoral images. Through the construction of the dental arch relative alignment reference, the dental alignment reference data is obtained, and the dental alignment reference data is sent to the local alignment anomaly verification module. The local arrangement anomaly verification module is used to distinguish the formation mechanism corresponding to the local arrangement anomaly, obtain the dental arrangement anomaly verification data through local arrangement anomaly verification, and send the dental arrangement anomaly verification data to the dental arrangement state recognition module; The tooth alignment state recognition module is used to generate a tooth alignment state recognition result, obtain the tooth alignment state recognition result through tooth alignment state recognition, and output the tooth alignment state recognition result.

Images