CN110288638B - Broken bone model rough registration method and system and broken bone model registration method - Google Patents
Broken bone model rough registration method and system and broken bone model registration method Download PDFInfo
- Publication number
- CN110288638B CN110288638B CN201910527855.2A CN201910527855A CN110288638B CN 110288638 B CN110288638 B CN 110288638B CN 201910527855 A CN201910527855 A CN 201910527855A CN 110288638 B CN110288638 B CN 110288638B
- Authority
- CN
- China
- Prior art keywords
- point cloud
- model
- bone
- point
- key
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T19/00—Manipulating 3D models or images for computer graphics
- G06T19/20—Editing of 3D images, e.g. changing shapes or colours, aligning objects or positioning parts
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
- G06T7/344—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods involving models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/04—Indexing scheme for image data processing or generation, in general involving 3D image data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30008—Bone
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2219/00—Indexing scheme for manipulating 3D models or images for computer graphics
- G06T2219/20—Indexing scheme for editing of 3D models
- G06T2219/2004—Aligning objects, relative positioning of parts
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2219/00—Indexing scheme for manipulating 3D models or images for computer graphics
- G06T2219/20—Indexing scheme for editing of 3D models
- G06T2219/2016—Rotation, translation, scaling
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Architecture (AREA)
- Computer Graphics (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Processing Or Creating Images (AREA)
Abstract
The invention provides a broken bone model rough registration method, a broken bone model rough registration system and a broken bone model registration method, which comprise the following steps: reconstructing the CT image of the fractured bone to obtain a group of fractured bone models; correspondingly extracting point cloud data on each broken bone model, and respectively converting each broken bone model into a corresponding broken bone point cloud model; correspondingly calculating a rotation matrix and a displacement matrix between each broken bone point cloud model and the sample bone point cloud model; and (4) performing displacement rotation on each broken bone point cloud model according to the rotation matrix and the displacement matrix which respectively correspond to the broken bone point cloud models to obtain the broken bone point cloud model after displacement rotation of each broken bone point cloud model, and completing coarse registration. The invention is used for realizing the registration of the fractured bone model which can not provide the axis, is used for realizing the registration of the fractured bone model with one to many parts, and is further beneficial to improving the registration precision.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a broken bone model coarse registration method and system and a broken bone model registration method.
Background
In recent years, computer-aided orthopedics technology has become a popular research, which extracts information about a bone from a CT image of a fractured bone of a patient to perform three-dimensional reconstruction, thereby obtaining a fractured bone model of the patient, and then registers the fractured bone model of the patient, thereby obtaining a complete bone model corresponding to the fractured bone.
In the actual registration process, the fracture model needs to be preprocessed, namely, coarse registration. At present, the rough registration method mainly comprises two methods, namely, the bone is aligned by using the bone axis, and the whole model is matched by using a local model. For the method of aligning the bone by using the bone axis, which is often the matching of the broken bone sections, the position between two broken bones is not suitable to be too far, and the broken bones are required to be in the same Z-axis direction. In addition, the above method of aligning bones by using the axes of bones is often suitable for a two-in-one broken bone model, and has relatively poor effect on a three-in-one, four-in-one, or even more broken bones. In addition, the above-mentioned method of matching the entire model using the local model is directed to matching the sample bone model itself using a part of the entire model. When applied to bone model matching, as human bones are similar in shape, bones of other people can be used as sample bones for matching and certain effects can be obtained, but when a size difference exists between two bones, a wrong matching result can be caused to a certain extent. In most cases, there will be a difference in size between different human bones, which affects the accuracy of the matching to some extent.
Therefore, the invention provides a novel broken bone model rough registration method, a system and a broken bone model registration method, which are used for solving the technical problems.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a broken bone model coarse registration method, a broken bone model coarse registration system and a broken bone model registration method, which are used for realizing the registration of a broken bone model when a broken bone cannot provide an axis. But also to improve the accuracy of the registration.
In a first aspect, the present invention provides a coarse registration method for a fractured bone model, including:
reconstructing the CT image of the fractured bone to obtain a group of fractured bone models;
correspondingly extracting point cloud data on each broken bone model, and respectively converting each broken bone model into a corresponding broken bone point cloud model;
correspondingly calculating a rotation matrix and a displacement matrix between each broken bone point cloud model and the sample bone point cloud model; the method comprises the steps that a sample bone point cloud model is obtained through conversion of the sample bone model, the sample bone model is obtained through reconstruction of a CT image of a pre-selected sample bone, the reconstruction method of the sample bone model is the same as that of a broken bone model, and the conversion obtaining method of the sample bone point cloud model is the same as that of the broken bone point cloud model;
and (4) performing displacement rotation on each broken bone point cloud model according to the rotation matrix and the displacement matrix which respectively correspond to the broken bone point cloud models to obtain the broken bone point cloud model after displacement rotation of each broken bone point cloud model, and completing coarse registration.
Further, the method correspondingly calculates a rotation matrix and a displacement matrix between each broken bone point cloud model and the sample bone point cloud model, and comprises the following steps:
correspondingly extracting point cloud key points on each broken bone point cloud model;
for each broken bone point cloud model: respectively matching the point cloud key points extracted from the broken bone point cloud model with the point cloud key points extracted from the sample bone point cloud model by the same method to obtain a group of successfully matched point cloud key point pairs;
respectively screening point cloud key point pairs corresponding to each broken bone point cloud model by adopting a geometric consistency method, screening out point cloud key point pairs which correspond to the broken bone point cloud models and are accurately matched, and recording as target point cloud key point pairs;
and correspondingly calculating a rotation matrix and a displacement matrix between each broken bone point cloud model and the sample bone point cloud model respectively based on the target point cloud key pairs corresponding to the screened broken bone point cloud models respectively.
Further, the point cloud key points extracted from the broken bone point cloud model are respectively matched with the point cloud key points extracted from the sample bone point cloud model by the same method to obtain a group of successfully matched point cloud key point pairs, and the method comprises a key point descriptor generation step and a point cloud key point pair generation step;
wherein, the key point descriptor generating step includes:
calculating local surface attributes of points corresponding to each point cloud data in the broken bone point cloud model;
generating a SHOT descriptor of each point cloud key point on the broken bone point cloud model by adopting a SHOT descriptor method based on the point cloud key points on the extracted broken bone point cloud model and the local surface attribute of the point corresponding to each point cloud data in the extracted broken bone point cloud model;
the point cloud key point pair generation step comprises the following steps:
and searching and generating a point cloud key point pair which is successfully matched on the broken bone point cloud model and the sample bone point cloud model by adopting a Kd-Tree method based on the SHOT descriptor of each point cloud key point on the broken bone point cloud model generated in the step of generating the key point descriptor and the SHOT descriptor of each point cloud key point on the sample bone point cloud model generated by adopting the same method.
Further, local surface attributes of points corresponding to each point cloud data in the broken bone point cloud model are calculated by adopting a normallestimationOMP class in PCL.
Further, the local surface properties include surface normal and curvature.
Further, respectively adopting a geometric consistency method to screen point cloud key point pairs corresponding to each fractured bone point cloud model, screening out point cloud key point pairs which correspond to the fractured bone point cloud models and are accurately matched, wherein the screening comprises the following steps of: extracting point cloud key point pairs corresponding to the broken bone point cloud model one by one, and respectively performing the following processing on each point cloud key point pair extracted currently:
r31, obtaining corresponding coordinates of the point cloud key points a and b in the broken bone point cloud model and the sample bone point cloud model according to the point cloud key points a and b corresponding to the currently extracted point cloud key point pairs, wherein the point cloud key point a belongs to the current broken bone point cloud model, and the point cloud key point b belongs to the sample bone point cloud model;
r32, setting radius r with coordinate point of point cloud key point a as centeraSearching all the points in the current broken bone point cloud model at the radius raEach cloud key point in the range forms a key point set A ═ a1,a2,...,an}; and setting a radius r by taking a coordinate point of the point cloud key point b as a centerbSearching all the positions at the radius r in the sample bone point cloud modelbEach point cloud key point in the range forms a key point set B ═ B1,b2,...,bm};
r33, traversing the key point set A, counting the total times N of the point cloud key points on the sample bone point cloud model matched with the point cloud key points in the key point set A appearing in the key point set B, if N is more than or equal to N, N is a preset total times threshold and N is more than or equal to 1, matching the point cloud key points a and B accurately and is a target point cloud key point pair, otherwise: and c, matching the point cloud key points a and b wrongly, and deleting the point cloud key point pair.
Further, correspondingly extracting point cloud key points on each broken bone point cloud model, including: correspondingly extracting point cloud key points on each broken bone point cloud model by respectively adopting a key point extraction method;
wherein, the key point extraction method comprises the following steps: and filtering and sampling the point cloud data on the extracted point cloud model of the fractured bone by adopting a Voxelgrid filter to obtain point cloud key points on the point cloud model of the fractured bone.
In a second aspect, the present invention provides a bone fracture model coarse registration system, comprising:
the fractured bone model reconstruction unit is used for reconstructing the CT image of the fractured bone to obtain a group of fractured bone models;
the model conversion unit is connected with the fractured bone model reconstruction unit and used for correspondingly extracting point cloud data on each fractured bone model and respectively converting each fractured bone model into a corresponding fractured bone point cloud model;
the calculation unit is connected with the model conversion unit and is used for correspondingly calculating a rotation matrix and a displacement matrix between each broken bone point cloud model and the sample bone point cloud model; the method comprises the steps that a sample bone point cloud model is obtained through conversion of the sample bone model, the sample bone model is obtained through reconstruction of a CT image of a pre-selected sample bone, the reconstruction method of the sample bone model is the same as that of a broken bone model, and the conversion obtaining method of the sample bone point cloud model is the same as that of the broken bone point cloud model;
and the rough registration unit is connected with the calculation unit and is used for performing displacement rotation on each broken bone point cloud model according to the rotation matrix and the displacement matrix which respectively correspond to the broken bone point cloud models to obtain the broken bone point cloud model after each broken bone point cloud model is subjected to displacement rotation, and the rough registration is completed.
In a third aspect, the present invention provides a method for registering a fractured bone model, including coarse registration and fine registration, wherein:
the coarse registration adopts a registration method, which is the coarse registration method of the fractured bone model;
the fine registration comprises the following steps: and performing spatial fine registration on each broken bone point cloud model obtained by coarse registration by adopting an ICP (inductively coupled plasma) algorithm.
Further, the method for performing spatial fine registration on each broken bone point cloud model obtained by coarse registration by adopting an ICP (inductively coupled plasma) algorithm comprises the following steps:
when the number of the broken bone point cloud models is 2, accurately registering the point sets of the two broken bone point cloud models through an ICP (inductively coupled plasma) algorithm to complete the joint of the two broken bones;
when the number of the broken bone point cloud models is more than 2:
randomly selecting two broken bone point cloud models, accurately registering the two selected broken bone point cloud models through an ICP (inductively coupled plasma) algorithm, and completing the joint of two corresponding broken bones to obtain a new broken bone point cloud model;
and then, circularly adopting an ICP (inductively coupled plasma) algorithm to accurately register the obtained new point cloud model of the fractured bone and any residual point cloud model of the fractured bone which is not accurately registered, and finishing the combination of the current corresponding fractured bones until the number of the residual point cloud models of the fractured bone which are not accurately registered is zero.
The invention has the beneficial effects that:
(1) according to the fracture model rough registration method, the fracture model rough registration system and the fracture model registration method, the rotation matrix and the displacement matrix between each fracture point cloud model and the sample bone point cloud model are calculated firstly, then each fracture point cloud model is rotated according to the corresponding rotation matrix and displacement matrix, the axis of the bone is not used for calibration in the whole registration process, the use of the axis of the fractured bone is avoided, and the rough registration of the fracture model can still be realized when the fractured bone cannot provide the axis.
(2) According to the fracture model rough registration method, the fracture model rough registration system and the fracture model registration method, the point cloud models of the fracture are subjected to rotary displacement matching according to the corresponding rotary matrix and displacement matrix respectively, the axis of the bone is not used for calibration in the whole registration process, the use of the axis of the bone is avoided, and the rough registration method is suitable for the fracture scene of one to two, is also suitable for the fracture scene of one to three, one to four and even more fracture scenes of one to more and has a wide application range.
(3) According to the fracture model rough registration method, the fracture model rough registration system and the fracture model registration method, the rotation displacement matching of the fracture point cloud model is realized through the rotation matrix and the displacement matrix, the matching process does not involve the size difference problem of the fracture bone and the sample bone, and the method and the system are convenient to realize.
(4) According to the method and the system for roughly registering the fractured bone model and the fractured bone model registering method provided by the invention, when the rotation matrix and the displacement matrix between each fractured bone point cloud model and the sample bone point cloud model are correspondingly calculated, the point cloud key point pair which is successfully matched with the sample bone point cloud model for the first time on the fractured bone point cloud model is firstly obtained, then the point cloud key point pair which is wrongly matched is eliminated by adopting a geometric consistency method, the accurately matched target point cloud key point pair is obtained, and then the rotation matrix and the displacement matrix between the fractured bone point cloud model and the sample bone point cloud model are calculated by matching the accurately matched target point cloud key point pair, so that the registering precision is improved to a certain extent.
In addition, the invention has reliable design principle, simple structure and very wide application prospect.
Drawings
In order to more clearly illustrate the embodiments or technical solutions in the prior art of the present invention, the drawings used in the description of the embodiments or prior art will be briefly described below, and it is obvious for those skilled in the art that other drawings can be obtained based on these drawings without creative efforts.
FIG. 1 is a schematic flow chart of a method for coarse registration of a fractured bone model according to an embodiment of the invention.
FIG. 2 is a schematic flow chart diagram for one embodiment of step S3 shown in FIG. 1.
FIG. 3 is a schematic block diagram of a fracture model coarse registration system according to an embodiment of the present invention.
FIG. 4 is a schematic flow chart diagram illustrating one embodiment of a method for registering a fractured bone model in accordance with the present invention.
Detailed Description
In order to make those skilled in the art better understand the technical solution of the present invention, the technical solution in the embodiment of the present invention will be clearly and completely described below with reference to the drawings in the embodiment of the present invention, and it is obvious that the described embodiment is only a part of the embodiment of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The following explains key terms appearing in the present invention.
Fig. 1 and 2 are schematic flow charts of a broken bone model coarse registration method according to an embodiment of the invention.
Referring to fig. 1 and 2, the method for coarse registration of a fractured bone model in the present embodiment includes the following steps S1-S4.
Step S1: and reconstructing the CT image of the fractured bone to obtain a group of fractured bone models.
Correspondingly, the sample bone model of the sample bone is obtained by reconstructing the CT image of the sample bone by the same reconstruction method. The sample bone model is a three-dimensional model, and may be reconstructed in advance, or may be reconstructed during the using process of the method, for example, in step S1.
The sample bone according to the present invention has the same or substantially the same shape as the fractured bone.
Step S2: and correspondingly extracting point cloud data on each broken bone model, and respectively converting each broken bone model into a corresponding broken bone point cloud model.
Specifically, taking a broken bone model A and a broken bone model B as examples:
extracting point cloud data on the fractured bone model A, and converting the fractured bone model A into a corresponding fractured bone point cloud model;
and extracting point cloud data on the fractured bone model B, and converting the fractured bone model B into a corresponding fractured bone point cloud model.
Correspondingly, the same conversion method may be employed, either in advance or in step S2: and extracting point cloud data on the sample bone model, and converting the sample bone model into a sample bone point cloud model.
Step S3: and correspondingly calculating a rotation matrix and a displacement matrix between each broken bone point cloud model and the sample bone point cloud model.
Specifically, a rotation matrix and a displacement matrix between each broken bone point cloud model and the sample bone point cloud model are correspondingly calculated, and the method comprises the following steps of r1-r 4.
Step r 1: and correspondingly extracting point cloud key points on each broken bone point cloud model.
In the embodiment, the corresponding extraction of the point cloud key points on each broken bone point cloud model comprises the step of correspondingly extracting the point cloud key points on each broken bone point cloud model by respectively adopting a key point extraction method.
Wherein, the key point extraction method comprises the following steps: and filtering and sampling the point cloud data on the extracted point cloud model of the fractured bone by adopting a Voxelgrid filter to obtain point cloud key points on the point cloud model of the fractured bone.
The VoxelGrid filter is realized based on a VoxelGrid class in PCL. Correspondingly, when the key point extraction method is specifically implemented: inputting each point cloud data of the extracted fractured bone point cloud model through a setInputCloud () function in a VoxelGrid class in the PCL, setting a calculation radius (which can be set to 5cm) through a setLeafSize () function in the VoxelGrid class, and then screening out point cloud key points on the fractured bone point cloud model through a filter () function in the VoxelGrid class.
Specifically, taking a point cloud model a of a fractured bone as an example: inputting each point cloud data of the extracted broken bone point cloud model A through a setInputCloud () function in a VoxelGrid class in PCL, setting a calculation radius (set to be 5cm) through a setLeafSize () function in the VoxelGrid class, and screening out point cloud key points on the broken bone point cloud model A through a filter () function in the VoxelGrid class.
And extracting point cloud key points on other broken bone point cloud models, and executing according to the broken bone point cloud model A.
In addition, the same method is adopted to filter and sample the point cloud data on the extracted sample bone point cloud model to obtain point cloud key points on the sample bone point cloud model, and the specific extraction steps comprise: inputting each point cloud data of the extracted sample bone point cloud model through a setInputCloud () function in a VoxelGrid class in the PCL, setting a calculation radius (the calculation radius corresponding to the broken bone point cloud model is the same and is also set to be 5cm) through a setLeafSize () function in the VoxelGrid class, and then screening out point cloud key points on the sample bone point cloud model through a filter () function in the VoxelGrid class.
Wherein, PCL (Point Cloud Library): the method is a large cross-platform open source C + + programming library, realizes a large number of point cloud related general algorithms and efficient data structures, and relates to point cloud acquisition, filtering, segmentation, registration, retrieval, feature extraction, identification, tracking, curved surface reconstruction, visualization and the like; and the system supports various operating system platforms and can run on Windows, Linux, Android, Mac OS X and partially embedded real-time systems.
Step r 2: for each broken bone point cloud model: and respectively matching the point cloud key points extracted from the self-broken bone point cloud model with the point cloud key points extracted from the sample bone point cloud model by the same method to obtain a group of successfully matched point cloud key point pairs.
The method for matching the point cloud key points extracted from the self-breaking bone point cloud model with the point cloud key points extracted from the sample bone point cloud model by the same method to obtain a group of successfully matched point cloud key point pairs comprises the following steps: and a key point descriptor generating step and a point cloud key point pair generating step.
The key point descriptor generating step described in this embodiment includes:
calculating local surface attributes of points corresponding to each point cloud data in the broken bone point cloud model, wherein the local surface attributes comprise a surface normal and a curvature;
and generating a SHOT descriptor of each point cloud key point on the broken bone point cloud model by adopting a SHOT descriptor method based on the point cloud key points on the extracted broken bone point cloud model and the local surface attributes of the points corresponding to each point cloud data in the extracted broken bone point cloud model.
The method includes the steps that a normals optimization OMP type in PCL is adopted to calculate local surface attributes of points corresponding to each point cloud data in a broken bone point cloud model, and the method specifically includes the following steps:
firstly, inputting cloud data of each point extracted on a broken bone point cloud model through a setInputCloud () function of a normalEstimationOMP class;
setting a calculation radius (which can be set to 10cm) through a setradius search () function of the normals optimization OMP class;
and thirdly, calculating and obtaining the local surface attribute of the point corresponding to each point cloud data in the broken bone point cloud model by executing a computer () function of a normals optimization OMP type.
The method comprises the following steps of adopting a SHOT descriptor method, generating a SHOT descriptor of each point cloud key point on the broken bone point cloud model based on the point cloud key points on the extracted broken bone point cloud model and the local surface attribute of the point corresponding to each point cloud data in the extracted broken bone point cloud model, and specifically comprising the following steps:
inputting all point cloud key points of the broken bone point cloud model through a setInputCloud () function in a SHOTEStatemationOMP class in PCL;
inputting point cloud data extracted from a broken bone point cloud model through a setSearchSurface () function in a SHOTETIMationOMP class;
calculating the radius (which may be set to 10cm) by setradiuussearch () function setting in the SHOTEstimationOMP class;
inputting local surface attributes of points corresponding to cloud data of each point in the broken bone point cloud model through a setInputNormals () function in the SHOTETIMINOMP class;
and executing a computer () function in the SHOTETIMationOMP class to generate a SHOT descriptor of each point cloud key point on the fractured bone point cloud model.
The normallestimationomp class, a class in PCL, estimates local surface properties of each 3D point, such as surface normal and curvature of the 3D point, in parallel using the OpenMP standard.
Correspondingly, the SHOT descriptor of each point cloud key point on the sample bone point cloud model can be generated by referring to the key point descriptor generation step.
In this embodiment, the step of generating the point cloud key point pairs includes: and searching and generating a point cloud key point pair which is successfully matched on the broken bone point cloud model and the sample bone point cloud model by adopting a Kd-Tree method based on the SHOT descriptor of each point cloud key point on the broken bone point cloud model generated in the step of generating the key point descriptor and the SHOT descriptor of each point cloud key point on the sample bone point cloud model generated by adopting the same method.
The Kd-Tree is a data structure for dividing k-dimensional data space and is mainly applied to searching of key data of multi-dimensional space.
The method comprises the following steps of generating a point cloud model of a fractured bone, wherein the step of generating the point cloud model of the fractured bone based on the point cloud key of each point cloud key point generated in the step of generating the key point descriptor and the step of generating the point cloud key point pairs of each point cloud key point on the sample bone point cloud model based on the same method are used for searching and generating the point cloud key point pairs successfully matched on the fractured bone point cloud model and the sample bone point cloud model by adopting a Kd-Tree method, and specifically comprises the following steps:
inputting all descriptors corresponding to the broken bone point cloud model through a setInputCloud () function of a KdTreeFLANN class in the PCL;
inputting all descriptors corresponding to the sample bone point cloud model through a nearestKSearch () function of a KdTreeFLANN class in PCL;
by executing the nearestKSearch () function, finding and calculating the square difference between each point descriptor between the broken bone point cloud model and the sample bone point cloud model, and correspondingly transmitting each group of descriptors with the square difference smaller than a preset square difference threshold value to a CorrespondansPtr () function of Kdtree FLANN class; recording each group of descriptors as a description subgroup, wherein the descriptors in each description subgroup group respectively form a point cloud key point pair which is successfully matched on the broken bone point cloud model and the sample bone point cloud model;
then, generating a corresponding relation between point cloud key points corresponding to the descriptors in each description subgroup group by the coresponsonsPtr () function, and storing the corresponding relation into a registration point list; and point cloud key points with corresponding relations in the registration point list form point cloud key point pairs which are successfully matched on the broken bone point cloud model and the sample bone point cloud model.
Each description subgroup consists of point cloud key points on a broken bone point cloud model and point cloud key points on a sample bone point cloud model, and the two point cloud key points corresponding to the two descriptors in the description subgroup group are point cloud key point pairs which are successfully matched on the broken bone point cloud model and the sample bone point cloud model; all point cloud key point pairs successfully matched on the broken bone point cloud model and the sample bone point cloud model exist in the visible registration point list.
Taking the description of subgroup a as an example: the description subgroup A is composed of a first descriptor and a second descriptor, the first descriptor corresponds to a point cloud key point d1 on the searched broken bone point cloud model, the second descriptor corresponds to a point cloud key point c1 on the searched sample bone point cloud model, and the square difference between the point cloud key point d1 and the point cloud key point c1 is smaller than a preset square difference threshold value; the point cloud key point d1 and the point cloud key point c1 are stored in the registration point list in a mutually corresponding manner, and are a point cloud key point pair which is successfully matched on the broken bone point cloud model and the sample bone point cloud model.
The value of the square error threshold in this embodiment is 0.25, and when the square error threshold is specifically implemented, a person skilled in the art may modify the specific value of the square error threshold according to actual needs.
Step r 3: and respectively screening point cloud key point pairs corresponding to each broken bone point cloud model by adopting a geometric consistency method, screening out point cloud key point pairs which correspond to the broken bone point cloud models and are accurately matched, and recording as target point cloud key point pairs. And the two point cloud key points in the target point cloud key point pair are both target point cloud key points.
The method comprises the following steps of screening point cloud key point pairs corresponding to each fractured bone point cloud model by respectively adopting a geometric consistency method, screening out the point cloud key point pairs which correspond to the fractured bone point cloud models and are accurately matched, wherein the screening comprises the following steps of:
extracting point cloud key point pairs corresponding to the broken bone point cloud model one by one, and respectively carrying out the following steps r31-r33 on each point cloud key point pair extracted currently:
r31, obtaining corresponding coordinates of the point cloud key points a and b in the broken bone point cloud model and the sample bone point cloud model according to the point cloud key points a and b corresponding to the currently extracted point cloud key point pairs, wherein the point cloud key point a belongs to the current broken bone point cloud model, and the point cloud key point b belongs to the sample bone point cloud model;
r32, setting radius r with coordinate point of point cloud key point a as centeraSearching all the points in the radius r in the current broken bone point cloud modelaEach cloud key point in the range forms a key point set A ═ a1,a2,...,an}; and setting a radius r by taking a coordinate point of the point cloud key point b as a centerbSearching all the positions at the radius r in the sample bone point cloud modelbEach point cloud key point in the range forms a key point set B ═ B1,b2,...,bm};
r33, traversing the key point set A, counting the total times N of the point cloud key points on the sample bone point cloud model matched with the point cloud key points in the key point set A appearing in the key point set B, if N is more than or equal to N, N is a preset total time threshold and N is more than or equal to 1, matching the point cloud key points a and B accurately and is a target point cloud key point pair, otherwise: and c, matching the point cloud key points a and b wrongly, and deleting the point cloud key point pair.
In this embodiment, in step r33, the deleting the point cloud key point pair includes: and deleting the point cloud key point pair from the registration point list. It can be seen that step r3 can update the list of registration points to obtain an updated list of registration points. And recording the updated registration point list as a key point matching list, wherein the key point matching list stores the key point pairs of the target point clouds on the fractured bone point cloud model and the sample bone point cloud model and the registration relationship thereof (namely the corresponding relationship between the fractured bone point cloud model and the target point cloud key points on the sample bone point cloud model).
Step r 4: and correspondingly calculating a rotation matrix and a displacement matrix between each broken bone point cloud model and the sample bone point cloud model respectively based on the target point cloud key pairs corresponding to the screened broken bone point cloud models respectively.
In this embodiment, rotation matrices and displacement matrices between each broken bone point cloud model and the sample bone point cloud model are correspondingly calculated by using a geometricconsistencylgrouping class in PCL. Specifically, for any one of the broken bone point cloud models (hereinafter referred to as the broken bone point cloud model M1):
key points (coordinates) of target point clouds on the broken bone point cloud model M1 are M11(x1, y1, z1), M12(x2, y2, z2), …, M1p (xp, yp, zp), and p is the number of the key points of the target point clouds on the broken bone point cloud model M1;
target point cloud key points M11(x1, y1, z1), M12(x2, y2, z2), … and M1p (xp, yp and zp) on a broken bone point cloud model M1, wherein the corresponding target point cloud key points on the sample bone point cloud model are J11(x1 ', y1 ', z1 '), J12(x2 ', y2 ', z2 '), …, J1q (xq ', yq ', zq '), and q is the number of the target point cloud key points on the sample bone point cloud model, and p is q;
inputting each target point cloud key point (coordinate) on the broken bone point cloud model M1 through a setInputCloud () function in a GeometricConsistencyGrouping class in the PCL: m11(x1, y1, z1), M12(x2, y2, z2), …, M1p (xp, yp, zp);
inputting a target point cloud key point on the sample bone point cloud model through a setSceneCloud () function in a GeometricConsistencyGrouping class: j11(x1 ', y1 ', z1 '), J12(x2 ', y2 ', z2 '), …, J1q (xq ', yq ', zq ');
inputting a key point matching list corresponding to the broken bone point cloud model M1 and a target point cloud key point pair on the sample bone point cloud model through a setModelSceneCorresponsendes () function of a GeometricConsistencyGrouping class;
and executing a recognize () function in the GeometricConsistencyGrouping class, and calculating and outputting a rotation matrix and a displacement matrix from the broken bone point cloud model M1 to the sample bone point cloud model.
Step S4: and (4) performing displacement rotation on each broken bone point cloud model according to the rotation matrix and the displacement matrix corresponding to each broken bone point cloud model, and completing coarse registration.
And based on the step S4, obtaining the broken bone point cloud model of each broken bone point cloud model after displacement rotation.
It should be noted that, after the coarse registration is completed, the method actually corresponds to moving all the point cloud models of the fractured bones to the same Z-axis direction.
In addition, the models referred to in the present invention are all three-dimensional models.
In conclusion, according to the method for roughly registering the fractured bone models, the rotation matrix and the displacement matrix between each fractured bone point cloud model and the sample bone point cloud model are calculated firstly, and then each fractured bone point cloud model is rotated according to the corresponding rotation matrix and displacement matrix, so that the rough registration of each fractured bone point cloud model is completed, and the axis of the bone is not used for calibration in the whole registration process: in this respect, the invention is helpful for realizing the registration of the fracture model for the fracture which can not provide an axis; on the other hand, the method is not only suitable for the registration of the fracture model with one minute and two minutes, but also suitable for the registration of the fracture model with one minute and three, one minute and four, even more than one minute; furthermore, the invention realizes the rotation matching of the broken bone point cloud model through the rotation matrix and the displacement matrix in the rough matching step, does not relate to the size difference problem of the broken bone and the sample bone in the matching process, and is convenient to realize.
In addition, the rough registration method of the fractured bone model provided by the invention firstly obtains the point cloud key point pairs which are successfully matched with the sample bone point cloud model for the first time on the fractured bone point cloud model, then eliminates the point cloud key point pairs which are wrongly matched by adopting a geometric consistency method to obtain the target point cloud key point pairs which are accurately matched, and then calculates the rotation matrix and the displacement matrix between the fractured bone point cloud model and the sample bone point cloud model by matching the accurate target point cloud key point pairs, thereby improving the registration precision to a certain extent.
FIG. 3 is a schematic block diagram of a coarse registration system of a fractured bone model according to an embodiment of the invention.
Referring to fig. 3, the fracture model coarse registration system includes:
the fractured bone model reconstruction unit is used for reconstructing the CT image of the fractured bone to obtain a group of fractured bone models;
the model conversion unit is used for correspondingly extracting point cloud data on each broken bone model and respectively converting each broken bone model into a corresponding broken bone point cloud model;
the calculation unit is used for correspondingly calculating a rotation matrix and a displacement matrix between each broken bone point cloud model and the sample bone point cloud model; the method comprises the steps that a sample bone point cloud model is obtained through conversion of the sample bone model, the sample bone model is obtained through reconstruction of a CT image of a pre-selected sample bone, the reconstruction method of the sample bone model is the same as that of a broken bone model, and the conversion obtaining method of the sample bone point cloud model is the same as that of the broken bone point cloud model;
and the rough registration unit is used for performing displacement rotation on each broken bone point cloud model according to the rotation matrix and the displacement matrix which respectively correspond to the broken bone point cloud models to obtain the broken bone point cloud model after displacement rotation of each broken bone point cloud model, and finishing rough registration.
The implementation methods of all the components of the fracture model rough registration system respectively correspond to the corresponding parts in the fracture model rough registration method. When the broken bone model coarse registration system is used: firstly, reconstructing a CT image of a fractured bone through a fractured bone model reconstruction unit to obtain a group of fractured bone models; then, correspondingly extracting point cloud data on each broken bone model reconstructed by the broken bone model reconstruction unit through a model conversion unit, and then respectively converting each broken bone model into a corresponding broken bone point cloud model; then, correspondingly calculating a rotation matrix and a displacement matrix between each broken bone point cloud model (obtained by conversion of the model conversion unit) and the sample bone point cloud model through a calculation unit; and then, respectively carrying out displacement rotation on each fractured bone point cloud model according to the corresponding rotation matrix and displacement matrix (obtained by calculation of a calculation unit) by a rough registration unit to obtain the fractured bone point cloud model after each fractured bone point cloud model is subjected to displacement rotation, and further finishing rough registration.
Since the coarse registration system of the fractured bone model corresponds to the fractured bone model registration method, the method has all the advantages of the fractured bone model registration method, and is not described herein again.
Based on the method, after rough registration, all the broken bone models are equivalent to be moved to the same Z-axis direction, but the broken bone sections between the broken bones cannot be perfectly superposed due to the difference between the sample bone model and the broken bone models in size, so the method for registering the broken bone models is provided for further matching the broken bone sections of the point cloud models of the broken bones after the rough registration is completed. Fig. 4 is a schematic structural block diagram of the method for registering a fractured bone model provided by the invention.
Referring to fig. 4, the method for registering a fractured bone model includes coarse registration and fine registration, wherein:
coarse registration
The coarse registration method adopted by the fracture model registration method is the above-mentioned fracture model coarse registration method. The coarse registration of the fracture model registration method can be performed by referring to the fracture model coarse registration method. And after coarse registration, obtaining each corresponding broken bone point cloud model.
(II) Fine registration
The fine registration of the broken bone model registration method comprises the following steps: and performing spatial fine registration on each broken bone point cloud model obtained by coarse registration by adopting an ICP (inductively coupled plasma) algorithm. For simplicity, the "point cloud model of fractured bone" referred to in the following two parts "1" and "2" of the present embodiment are each a corresponding point cloud model of fractured bone obtained by rough registration.
1. And when the number of the broken bone point cloud models is 2, accurately registering the point sets of the two broken bone point cloud models through an ICP (inductively coupled plasma) algorithm.
Specifically, an ICP (iterative closest point) algorithm is adopted to carry out iterative computation on the point clouds of two broken bone sections of the broken bone point cloud model for multiple times, fine adjustment is carried out on the relative position of the broken bone point cloud model for multiple times until the space distance between the point sets of the two broken bone sections is minimum, and space fine registration of the broken bone point cloud model is realized:
(1) respectively recording point clouds of cross sections of the two broken bones on the cloud model of the two broken bones as a target point cloud and a source point cloud;
(2) recording the point cloud in the target point cloud as pointsSet P ═ Pi,i=1,2,...,n};
(3) For each point in the point set P, finding the corresponding point closest to the point in the source point cloud, and forming a point set Q corresponding to the point set P { Q ═ Q { (Q) }i1, 2,., n }, such that | p |i-qiI.e. get a point q | ═ miniIs a distance point p found in the cloud of source pointsiThe closest point;
(4) calculating a registration matrix between the point sets P and Q by adopting a minimum root mean square method, so that registration transformation matrixes R and t are obtained, wherein R is a rotation matrix of 3 multiplied by 3, and t is a displacement matrix of 3 multiplied by 1;
(5) performing coordinate transformation on the point set P by using the rotation matrix R and the displacement matrix t to obtain a new point set P ', namely P' ═ RP + t;
(6) calculating the average distance d between the point set P' and the point set Q;
(7) if the average distance d is less than a given limit threshold e, continuing to perform (8), otherwise performing (9);
(8) after the iterative computation is finished, the current latest rotation matrix R and displacement matrix t are used for realizing the rotation displacement operation between the point cloud models of the fractured bones to which the target point cloud and the source point cloud belong, and the fractured bone section splicing of the cloud models of the two fractured bones is finished;
(9) the above (3) to (7) are repeated with the point set P' in place of P.
2. When the number of the broken bone point cloud models is more than 2:
randomly selecting two broken bone point cloud models, accurately registering the two selected broken bone point cloud models through an ICP (inductively coupled plasma) algorithm, completing the joint of the two corresponding broken bones, and obtaining a new broken bone point cloud model;
and then, circularly adopting an ICP (inductively coupled plasma) algorithm to accurately register the obtained new point cloud model of the fractured bone and any one of the remaining point cloud models of the fractured bone which is not accurately registered, and finishing the combination of the current corresponding fractured bones until the number of the remaining point cloud models of the fractured bone which are not accurately registered is zero.
Since the fracture model registration method performs coarse registration based on the fracture model coarse registration method, the fracture model registration method has all the advantages of the fracture model registration method, and details are not repeated herein.
It should be noted that the same and similar parts in the various embodiments in this specification may be referred to each other. In particular, for the system embodiment, since it is substantially similar to the embodiment of the bone fracture model registration method, the description is simple, and the relevant points can be referred to the description in the embodiment of the bone fracture model registration method.
In conclusion, the method can still realize the registration of the fractured bone model when the fractured bone cannot provide the axis, and the point cloud key point pairs are screened by adopting a geometric consistency method, so that the registration accuracy is improved to a certain extent.
Although the present invention has been described in detail by referring to the drawings in connection with the preferred embodiments, the present invention is not limited thereto. Various equivalent modifications or substitutions can be made on the embodiments of the present invention by those skilled in the art without departing from the spirit and scope of the present invention, and these modifications or substitutions are within the scope of the present invention/any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (7)
1. A method for registering a fractured bone model comprises coarse registration and fine registration, which is characterized in that,
the coarse registration adopts a registration method which is a broken bone model coarse registration method;
the fine registration comprises the following steps: performing spatial fine registration on each broken bone point cloud model obtained by coarse registration by adopting an ICP (inductively coupled plasma) algorithm;
the ICP algorithm is adopted to carry out spatial fine registration on each broken bone point cloud model obtained by coarse registration, and the method comprises the following steps:
when the number of the broken bone point cloud models is 2, accurately registering the point sets of the two broken bone point cloud models through an ICP (inductively coupled plasma) algorithm to complete the joint of the two broken bones;
when the number of the broken bone point cloud models is more than 2:
randomly selecting two broken bone point cloud models, accurately registering the two selected broken bone point cloud models through an ICP (inductively coupled plasma) algorithm, and completing the joint of two corresponding broken bones to obtain a new broken bone point cloud model;
then, the ICP algorithm is circularly adopted to carry out accurate registration on the obtained new point cloud model of the fractured bone and any remaining point cloud model of the fractured bone which is not subjected to accurate registration, and the combination of the current corresponding fractured bone is completed until the number of the remaining point cloud models of the fractured bone which are not subjected to accurate registration is zero;
the coarse registration method of the fractured bone model comprises the following steps:
reconstructing the CT image of the fractured bone to obtain a group of fractured bone models;
correspondingly extracting point cloud data on each broken bone model, and respectively converting each broken bone model into a corresponding broken bone point cloud model;
correspondingly calculating a rotation matrix and a displacement matrix between each broken bone point cloud model and the sample bone point cloud model; the method comprises the steps that a sample bone point cloud model is obtained through conversion of the sample bone model, the sample bone model is obtained through reconstruction of a CT image of a pre-selected sample bone, the reconstruction method of the sample bone model is the same as that of a broken bone model, and the conversion obtaining method of the sample bone point cloud model is the same as that of the broken bone point cloud model;
and (4) performing displacement rotation on each broken bone point cloud model according to the rotation matrix and the displacement matrix which respectively correspond to the broken bone point cloud models to obtain the broken bone point cloud model after displacement rotation of each broken bone point cloud model, and completing coarse registration.
2. The method for registering a fractured bone model according to claim 1, wherein the rotating matrix and the displacement matrix between each fractured bone point cloud model and the sample bone point cloud model are correspondingly calculated, and the method comprises the following steps:
correspondingly extracting point cloud key points on each broken bone point cloud model;
for each broken bone point cloud model: respectively matching the point cloud key points extracted from the broken bone point cloud model with the point cloud key points extracted from the sample bone point cloud model by the same method to obtain a group of successfully matched point cloud key point pairs;
respectively screening point cloud key point pairs corresponding to each broken bone point cloud model by adopting a geometric consistency method, screening out point cloud key point pairs which correspond to the broken bone point cloud models and are accurately matched, and recording as target point cloud key point pairs;
and correspondingly calculating a rotation matrix and a displacement matrix between each broken bone point cloud model and the sample bone point cloud model respectively based on the target point cloud key pairs corresponding to the screened broken bone point cloud models respectively.
3. The fractured bone model registration method according to claim 2, wherein the point cloud key points extracted from the fractured bone point cloud model are matched with the point cloud key points extracted from the sample bone point cloud model by the same method to obtain a group of successfully matched point cloud key point pairs, and the method comprises a key point descriptor generation step and a point cloud key point pair generation step;
wherein, the key point descriptor generating step includes:
calculating local surface attributes of points corresponding to each point cloud data in the broken bone point cloud model;
generating a SHOT descriptor of each point cloud key point on the broken bone point cloud model by adopting a SHOT descriptor method based on the point cloud key points on the extracted broken bone point cloud model and the local surface attribute of the point corresponding to each point cloud data in the extracted broken bone point cloud model;
the point cloud key point pair generation step comprises the following steps:
and searching and generating a point cloud key point pair which is successfully matched on the broken bone point cloud model and the sample bone point cloud model by adopting a Kd-Tree method based on the SHOT descriptor of each point cloud key point on the broken bone point cloud model generated in the step of generating the key point descriptor and the SHOT descriptor of each point cloud key point on the sample bone point cloud model generated by adopting the same method.
4. The fractured bone model registration method according to claim 3, further comprising calculating local surface attributes of points corresponding to each point cloud data in the fractured bone point cloud model by using a normals optimization OMP class in PCL; the PCL is a point cloud library.
5. A method of registering a fractured bone model according to claim 3 wherein the local surface attributes include surface normal and curvature.
6. The registration method of the fractured bone model according to claim 2, wherein the geometrical consistency method is adopted to screen the point cloud key point pairs corresponding to each fractured bone point cloud model, and the screening of the point cloud key point pairs corresponding to each fractured bone point cloud model and accurately matched with each other comprises the following steps of: extracting point cloud key point pairs corresponding to the broken bone point cloud model one by one, and respectively performing the following processing on each point cloud key point pair extracted currently:
r31, obtaining corresponding coordinates of the point cloud key points a and b in the broken bone point cloud model and the sample bone point cloud model according to the point cloud key points a and b corresponding to the currently extracted point cloud key point pairs, wherein the point cloud key point a belongs to the current broken bone point cloud model, and the point cloud key point b belongs to the sample bone point cloud model;
r32, setting radius r with coordinate point of point cloud key point a as centeraSearching all the points in the current broken bone point cloud model at the radius raEach cloud key point in the range forms a key point set A ═ a1,a2,…,an}; and setting a radius r by taking a coordinate point of the point cloud key point b as a centerbSearching all the positions at the radius r in the sample bone point cloud modelbEach point cloud key point in the range forms a key point set B ═ B1,b2,…,bm};
r33, traversing the key point set A, counting the total times N of the point cloud key points on the sample bone point cloud model matched with the point cloud key points in the key point set A appearing in the key point set B, if N is more than or equal to N, N is a preset total times threshold and N is more than or equal to 1, matching the point cloud key points a and B accurately and is a target point cloud key point pair, otherwise: and c, matching the point cloud key points a and b wrongly, and deleting the point cloud key point pair.
7. The registration method of the fractured bone model according to claim 2, wherein the corresponding extraction of the point cloud key points on each fractured bone point cloud model comprises the steps of respectively adopting a key point extraction method to correspondingly extract the point cloud key points on each fractured bone point cloud model;
wherein, the key point extraction method comprises the following steps: and filtering and sampling the point cloud data on the extracted point cloud model of the fractured bone by adopting a Voxelgrid filter to obtain point cloud key points on the point cloud model of the fractured bone.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910527855.2A CN110288638B (en) | 2019-06-18 | 2019-06-18 | Broken bone model rough registration method and system and broken bone model registration method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910527855.2A CN110288638B (en) | 2019-06-18 | 2019-06-18 | Broken bone model rough registration method and system and broken bone model registration method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110288638A CN110288638A (en) | 2019-09-27 |
CN110288638B true CN110288638B (en) | 2022-07-01 |
Family
ID=68004071
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910527855.2A Active CN110288638B (en) | 2019-06-18 | 2019-06-18 | Broken bone model rough registration method and system and broken bone model registration method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110288638B (en) |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111476832A (en) * | 2020-03-23 | 2020-07-31 | 杭州柳叶刀机器人有限公司 | Acetabular cup registration method |
CN111724420A (en) * | 2020-05-14 | 2020-09-29 | 北京天智航医疗科技股份有限公司 | Intraoperative registration method and device, storage medium and server |
CN114274139B (en) * | 2020-09-27 | 2024-04-19 | 西门子股份公司 | Automatic spraying method, device, system and storage medium |
CN112381922A (en) * | 2020-10-27 | 2021-02-19 | 济南大学 | Method, system and terminal for obtaining skeleton model of human skeleton missing part |
CN112764651B (en) * | 2021-02-01 | 2022-03-08 | 飞燕航空遥感技术有限公司 | Method and system for drawing three-dimensional point cloud profile of browser |
CN113284252B (en) * | 2021-03-03 | 2023-05-05 | 山东交通学院 | Resetting method and resetting system of fractured bone model based on VTK |
CN113345112B (en) * | 2021-05-25 | 2023-06-13 | 上海大学 | Long bone fracture surface point cloud preprocessing and registering method |
CN113409301A (en) * | 2021-07-12 | 2021-09-17 | 上海精劢医疗科技有限公司 | Point cloud segmentation-based femoral neck registration method, system and medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105869149A (en) * | 2016-03-24 | 2016-08-17 | 大连理工大学 | Principal vector analysis based broken bone section segmentation and broken bone model registration method |
CN108537805A (en) * | 2018-04-16 | 2018-09-14 | 中北大学 | A kind of target identification method of feature based geometry income |
CN109816705A (en) * | 2019-01-03 | 2019-05-28 | 咸阳师范学院 | A kind of Characteristic points match method lacking skull |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9993211B2 (en) * | 2013-09-20 | 2018-06-12 | The Brigham And Women's Hospital, Inc. | System and method for automatic detection and registration of medical images |
-
2019
- 2019-06-18 CN CN201910527855.2A patent/CN110288638B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105869149A (en) * | 2016-03-24 | 2016-08-17 | 大连理工大学 | Principal vector analysis based broken bone section segmentation and broken bone model registration method |
CN108537805A (en) * | 2018-04-16 | 2018-09-14 | 中北大学 | A kind of target identification method of feature based geometry income |
CN109816705A (en) * | 2019-01-03 | 2019-05-28 | 咸阳师范学院 | A kind of Characteristic points match method lacking skull |
Also Published As
Publication number | Publication date |
---|---|
CN110288638A (en) | 2019-09-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110288638B (en) | Broken bone model rough registration method and system and broken bone model registration method | |
CN107122705B (en) | Face key point detection method based on three-dimensional face model | |
CN106803267B (en) | Kinect-based indoor scene three-dimensional reconstruction method | |
US7953271B2 (en) | Enhanced object reconstruction | |
Remondino et al. | 3D modeling of close-range objects: photogrammetry or laser scanning? | |
KR101532864B1 (en) | Planar mapping and tracking for mobile devices | |
Mellado et al. | Relative scale estimation and 3D registration of multi-modal geometry using growing least squares | |
Mian et al. | Automatic correspondence for 3D modeling: An extensive review | |
CN110226806B (en) | Sole gluing track generation method and device | |
Sinko et al. | 3D registration of the point cloud data using ICP algorithm in medical image analysis | |
CN107330281B (en) | Full-automatic fracture steel plate model personalized reconstruction method | |
CN110176079B (en) | Three-dimensional model deformation algorithm based on quasi-conformal mapping | |
CN111275811A (en) | Repair model creation method, system and storage medium | |
CN110910433A (en) | Point cloud matching method based on deep learning | |
Feng et al. | Repair volume extraction method for damaged parts in remanufacturing repair | |
CN109965979A (en) | A kind of steady Use of Neuronavigation automatic registration method without index point | |
Zhang | CuFusion2: accurate and denoised volumetric 3D object reconstruction using depth cameras | |
CN116958233A (en) | Skin burn area calculation method based on multiband infrared structured light system | |
Balzer et al. | Volumetric reconstruction applied to perceptual studies of size and weight | |
Zheng et al. | A new 3D contour extraction method for tooth cavity in a dental CAD/CAM system | |
US20240000512A1 (en) | Calculating range of motion | |
CN113284252B (en) | Resetting method and resetting system of fractured bone model based on VTK | |
Yan et al. | Research on splicing method of digital relic fragment model | |
CN113593049B (en) | Virtual-real fusion method for geometric consistency of real object and virtual object in scene | |
Graber | Realtime 3D reconstruction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |