Based on the corona data extraction method of digital modelTechnical field
The invention belongs to tooth model technical field of image processing, particularly a kind of method that corona data in digital detal cast are extracted automatically.
Technical background
In recent years, along with computer graphics and the continuous of virtual reality emulation technology are improved and develop, computer science and infotech are day by day mutually integrated with modern medicine and correlation technique thereof and interpenetrate.Simultaneously, along with computer hardware technique, image acquisition and treatment technology, three-dimensional digital imaging technique etc. are in the ground develop rapidly of dentistry field, computer technology has been made to become an important component part in Education in Stomatology, clinical, scientific research and capital construction.And Orthodontics is as a branch in dentistry field, main research tooth jaw, face, the growing and deformity that a variety of causes causes of cranium three-dimensional structure, nowadays a digitized revolution is faced with to rescuing of dentognathic deformity, especially diagnose at mouth cavity orthodontic with three-dimensional digital imaging and measuring technique, design, treat and apply more and more widely in outcome prediction, make Orthodontics more and more towards computerized future development, and have development prospect well.
Orthodontics is a stomatologic subdiscipline.Orthodontic treatment mainly adjusts the harmony between the nerve of facial skeleton, tooth and Maxillary region and muscle by various apparatus for correcting, namely between the upper mandibular of adjustment, between upper lower teeth, between tooth and jawbone and the abnormal relation contacted between their nerve and muscle, it finally rescues target is the balance, stable and attractive in appearance reaching stomatognathic system.The rescuing of malocclusion mainly relies on tooth and wears with appliance, applies suitable Biological Strength, make it produce physiological and move, thus rescue malocclusion tooth, alveolar bone and jawbone.
Dental appliance is divided into fixed appliance and invisible orthotic device two kinds, and fixed appliance is fixed on corona by orthodontic bracket bonding, and arch wire, through orthodontic bracket, regulates the Biological Strength of rectification by arch wire.The shortcoming of fixed appliance is: orthodontic bracket is fixed on corona, can damage corona glaze paint, and not easily cleaning teeth.If orthodontic bracket is fixed on the buccal aspect of corona, cause appearance unsightly.If orthodontic bracket is fixed on the tongue side of corona, tongue side mucous membrane easily damages again.Invisible orthotic device then has specious, and tooth cleans well, can not the advantage of lesion ports transmucosal, and therefore invisible orthotic device is employed more and more widely.
When making invisible orthotic device, first carry out three-dimensional data typing by computed tomography scanning technology acquisition oral cavity former or gypsum formpiston, carry out the three-dimensional reconstruction of model; Digital model after reconstruction directly can carry out the rotation in each orientation, observation, Scalable on computers, automatically can measure measure the items such as tooth, dental arch, base bones; On the basis of this research, in conjunction with visualization of 3 d image procossing and three-dimensional laser rapid shaping technique, move mode and the step of orthodontic design and tooth can be simulated, carry out rescuing of visualization of 3 d dentognathic deformity, and each three-dimensional tooth model rescuing the stage is carried out three-dimensional fast laser shaping, then make the transparent press mold invisible orthotic device in each stage on the master mold be shaped.By the stealthy orthodontic system that corona and gum in computer-implemented simulation orthodontic process are split, the tooth model of patient is read in system, the shape of tooth, parameters are recorded and preserved.Further, simulate the whole corrective therapy process of this patient on computers, record the change in location of each stage tooth, according to the invisible orthotic device of these each different phases of design data.As can be seen here, corona and gum segmentation are important and key part.Single corona extracts by its fundamental purpose exactly from gum, and realize being separated between corona with corona, the final time carries out the object of flexible operating to single tooth, for postorder moves tooth, to arrange and the work such as collision detection lays the foundation.
Now conventional separation corona and the method for gum data are the discrete curvature by calculating tooth jaw three-dimensional model, according to curvature distribution, digital detal cast are divided into different groupings, thus segmentation gum and corona.The shortcoming utilizing the method that the method calculating discrete curvature splits gum and corona to exist is: 1, mathematically, the curvature of three-dimensional model is defined on continuous curve surface, and digital model generally adopts triangular mesh or polygonal mesh to represent, namely the curvature of digital model is discrete, replaces mathematical continuous curvature and unreasonable by the discrete curvature of estimation.2, in the discrete process of the curvature digitizing by three-dimensional model, inevitably introduce noise, the curvature of estimation originally can be made more inaccurate.
Summary of the invention
In order to the accuracy overcoming existing corona and the existence of gum separation method is not high, the shortcoming that success ratio is low, the invention provides one and accurately can split corona data and gum data, is divided into the corona data extraction method based on digital model that power is high.
Based on the corona data extraction method of digital model, comprise the following steps:
1) the oral cavity data of patient, are obtained, utilize the three-dimensional dentognathic model of the oral cavity data reconstruction patient of patient, three-dimensional dentognathic model is converted to digital detal cast, digital detal cast D Triangulation M=(G, R) represent, wherein G=(V, E, F) is by vertex set V={vi}, limit collection E={ (vi, vj) }, gore collection F={ (vi, vj, vk) } connected graph that forms, and grid model meets two-dimensional manifold character;
2) tooth to be split in digital detal cast, by operator is clicked successively, using the point of operator's click as the center of current clicked tooth;
3), PCA(principal component analysis (PCA) is utilized) algorithm, calculate the least square plane of whole digital detal cast, the normal direction of this least square plane is designated as N;
4), successively each tooth is proceeded as follows:
(4.1) center of current dental and the normal of least square plane create a reference plane, with the normal of reference plane for datum line excessively; Datum line is pressed clockwise step-by-step movement to rotate, datum line often walks and all creates a stepping plane with the center of current dental further; Current stepping plane and digital detal cast ask the intersection point handed over and obtain current stepping plane and digital tooth cast, and these intersection points form the intersection of current stepping plane and digital detal cast;
(4.2), using the concave point on each intersection as unique point, adopt bilateral filtering algorithm, detect the unique point on each intersection;
(4.3), the minimum intersection of unique point is found as seed intersection, using the unique point on this seed intersection as Seed Points;
(4.4), arrange in adjacent stepping plane between unique point weights, rule is as follows:
A, respectively using the distance of any unique point A of current stepping plane and any unique point B with its adjacent stepping plane as the weights of unique point A-unique point B;
B, the weights of two unique points be in same stepping plane are set to infinity;
C, the weights of two unique points be in non-adjacent stepping plane are set to infinity;
(4.5) according to the shortest path first (Dijkstra's algorithm) in graph theory, take Seed Points as starting point, every bar intersection is passed through successively by clockwise, and with this Seed Points for terminal, find Seed Points and enclose the shortest path formed around all intersections one, this shortest path is as the cut-off rule of corona.
Further, step (4.5) comprising:
(4.5.1) Seed Points is obtained as current point A;
(4.5.2) from current point A, the adjacent intersection of current point place intersection is obtained by clockwise;
(4.5.3) unique point on current adjacent intersection is, obtain unique point A-unique point respectivelyweights, unique pointfor i-th unique point on current adjacent intersection, find minimum unique point A-unique pointweights, recording feature point, this minimum unique point A-unique point cumulativeweights obtain current path weights;
(4.5.4) with unique pointas new current point A, repeat step (4.5.2)-(4.5.4), until current point gets back to Seed Points, the result that weights add up is the routine weight value of this closed path;
(4.5.5) judge that this Seed Points whether its intersection does not carry out last Seed Points of path computing, if not, then obtain next Seed Points as current point A, repeat step (4.5.2)-(4.5.5);
(4.5.6) minimum routine weight value is if so, found, using the closed path corresponding with these minimal path weights as the cut-off rule of corona.
The present invention distributes without the need to the three dimensional curvature of appraising model, three-dimensional model process is converted into the process of two dimensional surface data simultaneously, reduces the interference of noise data, ensure that the success ratio of extraction; The present invention can disposable automatic extraction cut-off rule dentulous, without the need to too much human intervention, reduces the requirement to operation technician; The method drastically increases the treatment effeciency of system, meets the requirement of mass production better.
The invention has the advantages that:
1, the present invention only need operate technician on each tooth, click a center, reduces the complexity of man-machine interactively.
2, the present invention shields three-dimensional grid model Curvature Estimate process, the calculating of three-dimensional model is converted into the process of two dimensional surface data, improves stability and the availability of method.
3, the present invention adopts shortest path in graph theory theoretical, by arranging the weight between each node, seek to Seed Points for starting point gets back to the shortest closed path of a circle that Seed Points formed after all intersections one encloses, using this closed path as the cut-off rule of corona and gum, successfully solve the bottleneck problem determining cut-off rule.
4, disposable automatic extraction all tooth cut-off rules method, drastically increases system treatment effeciency.
Accompanying drawing explanation
Fig. 1 is process flow diagram of the present invention.
Fig. 2 is the schematic diagram that operator clicks the center of each tooth.
Fig. 3 is the schematic diagram of the intersection of all stepping planes of labeling of single tooth and dentognathic model.
Fig. 4 is the schematic diagram of the cut-off rule candidate point of single corona.
Fig. 5 is the schematic diagram of the cut-off rule of single corona.
Fig. 6 is the schematic diagram of the cut-off rule of all coronas.
Fig. 7 is the schematic diagram that corona extracts result.
Embodiment
With reference to accompanying drawing, further illustrate the present invention:
Based on the corona data extraction method of digital model, comprise the following steps:
1) the oral cavity data of patient, are obtained, utilize the three-dimensional dentognathic model of the oral cavity data reconstruction patient of patient, three-dimensional dentognathic model is converted to digital detal cast, digital detal cast D Triangulation M=(G, R) represent, wherein G=(V, E, F) is by vertex set V={vi}, limit collection E={ (vi, vj) }, gore collection F={ (vi, vj, vk) } connected graph that forms, and grid model meets two-dimensional manifold character;
2) tooth to be split in digital detal cast, by operator is clicked successively, using the point of operator's click as the center of current clicked tooth;
3), PCA(principal component analysis (PCA) is utilized) algorithm, calculate the least square plane of whole digital detal cast, the normal direction of this least square plane is designated as N;
4), successively each tooth is proceeded as follows:
(4.1) center of current dental and the normal of least square plane create a reference plane, with the normal of reference plane for datum line excessively; Datum line is pressed clockwise step-by-step movement to rotate, datum line often walks and all creates a stepping plane with the center of current dental further; Current stepping plane and digital detal cast ask the intersection point handed over and obtain current stepping plane and digital tooth cast, and these intersection points form the intersection of current stepping plane and digital detal cast;
(4.2), using the concave point on each intersection as unique point, adopt bilateral filtering algorithm, detect the unique point on each intersection, the set of unique point constitutes the set of the candidate point of tooth cut-off rule;
(4.3), the minimum intersection of unique point is found as seed intersection, using the unique point on this seed intersection as Seed Points;
(4.4), arrange in adjacent stepping plane between unique point weights, rule is as follows:
A, respectively using the distance of any unique point A of current stepping plane and any unique point B with its adjacent stepping plane as the weights of unique point A-unique point B;
B, the weights of two unique points be in same stepping plane are set to infinity;
C, the weights of two unique points be in non-adjacent stepping plane are set to infinity;
(4.5) according to the shortest path first (Dijkstra's algorithm) in graph theory, take Seed Points as starting point, find Seed Points gets back to the formation of this Seed Points around all intersection one circles shortest path with clockwise, this shortest path is as the cut-off rule of corona.
(4.5.1) Seed Points is obtained as current point A;
(4.5.2) from current point A, the adjacent intersection of current point place intersection is obtained by clockwise; Clockwise can be also can be counterclockwise clockwise, as long as going out to be sent back to Seed Points from Seed Points is by a clockwise stepping;
(4.5.3) unique point on current adjacent intersection is, obtain unique point A-unique point respectivelyweights, unique pointfor i-th unique point on current adjacent intersection, find minimum unique point A-unique pointweights, recording feature point, this minimum unique point A-unique point cumulativeweights obtain current path weights;
(4.5.4) with unique pointas new current point A, repeat step (4.5.2)-(4.5.4), until current point gets back to Seed Points, the result that weights add up is the routine weight value of this closed path;
(4.5.5) judge that this Seed Points whether its intersection does not carry out last Seed Points of path computing, if not, then obtain next Seed Points as current point A, repeat step (4.5.2)-(4.5.5);
(4.5.6) minimum routine weight value is if so, found, using the shortest path corresponding with these minimal path weights as the cut-off rule of corona.
The cut-off rule of gum and tooth is made up of point, and these points are included in the unique point detecting and obtain, and are subsets of unique point.Include noise characteristic point in unique point, the present invention adopts shortest path first to be exactly in order to filter out noise unique point, if closed path is the shortest, namely routine weight value is minimum, then can be considered cut-off rule.
The present invention distributes without the need to the three dimensional curvature of appraising model, three-dimensional model process is converted into the process of two dimensional surface data simultaneously, reduces the interference of noise data, ensure that the success ratio of extraction; The present invention can disposable automatic extraction cut-off rule dentulous, without the need to too much human intervention, reduces the requirement to operation technician; The method drastically increases the treatment effeciency of system, meets the requirement of mass production better.
The invention has the advantages that:
1, the present invention only need operate technician on each tooth, click a center, reduces the complexity of man-machine interactively.
2, the present invention shields three-dimensional grid model Curvature Estimate process, the calculating of three-dimensional model is converted into the process of two dimensional surface data, improves stability and the availability of method.
3, the present invention adopts shortest path in graph theory theoretical, by arranging the weight between each node, seek to Seed Points for starting point gets back to the shortest closed path of a circle that Seed Points formed after all intersections one encloses, using this closed path as the cut-off rule of corona and gum, successfully solve the bottleneck problem determining cut-off rule.
4, disposable automatic extraction all tooth cut-off rules method, drastically increases system treatment effeciency.
Content described in this instructions embodiment is only enumerating the way of realization of inventive concept; protection scope of the present invention should not be regarded as being only limitted to the concrete form that embodiment is stated, protection scope of the present invention also and conceive the equivalent technologies means that can expect according to the present invention in those skilled in the art.