CN110288638A - A kind of knochenbruch model rough registration method, system and knochenbruch Model registration method - Google Patents
A kind of knochenbruch model rough registration method, system and knochenbruch Model registration method Download PDFInfo
- Publication number
- CN110288638A CN110288638A CN201910527855.2A CN201910527855A CN110288638A CN 110288638 A CN110288638 A CN 110288638A CN 201910527855 A CN201910527855 A CN 201910527855A CN 110288638 A CN110288638 A CN 110288638A
- Authority
- CN
- China
- Prior art keywords
- knochenbruch
- point cloud
- model
- point
- cloud model
- 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.)
- Granted
Links
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
Abstract
The present invention provides a kind of knochenbruch model rough registration method, system and knochenbruch Model registration method, comprising: the CT image of fractured bones is reconstructed to obtain one group of knochenbruch model;The corresponding point cloud data extracted on each knochenbruch model, is respectively converted into corresponding knochenbruch point cloud model for each knochenbruch model;The corresponding spin matrix and transposed matrix calculated between each knochenbruch point cloud model and sample bone point cloud model;Each knochenbruch point cloud model is subjected to displacement rotating according to its corresponding spin matrix and transposed matrix respectively, obtains knochenbruch point cloud model of each knochenbruch point cloud model after displacement rotating, rough registration is completed.The present invention then helps to improve the precision of registration for realizing the registration for the knochenbruch model that can not provide axis for realizing the registration of the knochenbruch model more than one point.
Description
Technical field
The present invention relates to technical field of image processing, and in particular to a kind of knochenbruch model rough registration method, system and knochenbruch
Model registration method.
Background technique
In recent years, computer aided orthopedics technology had become a kind of popular research, was schemed by the CT of patients with fractures's bone
It extracts and carries out three-dimensional reconstruction about the information of bone, obtain the knochenbruch model of patient, then the knochenbruch model of patient is matched
Standard, thus obtaining the corresponding complete skeleton model of fractured bones.
During actual registration, need to pre-process knochenbruch model, i.e. rough registration.And currently, rough registration method master
Will there are two types of, first is that using bone axis be aligned bone, second is that utilize partial model match overall model.For utilizing bone axis
The method that line is aligned bone, the often matching to knochenbruch disconnected section, it is desirable that the position between two knochenbruch should not be apart from mistake
It far and requires in same Z-direction between knochenbruch, current feasible method is to carry out Z-direction using the axis of bone
Calibration, but for that can not provide the knochenbruch of axis, the method cannot achieve rough registration.In addition, above-mentioned utilize bone axis pair
The method of neat bone is often suitable for the knochenbruch model of one-to-two, for one point three, one point of four or even one points of more knochenbruch,
Effect is relatively bad.In addition, the above-mentioned method using partial model matching overall model, is for a part with overall model
Remove matched sample bone model itself.Be applied to skeleton model matching when, due to human skeleton have in shape it is similar, can
To use other people bone to match as sample bone, and certain effect can be obtained, but between two bones
There are the matching results that when difference in size, will lead to mistake to a certain extent.And most of situation, between different human body bone
There can be the difference of size, this influences matched precision to a certain extent.
For this purpose, the present invention provides a kind of new knochenbruch model rough registration method, system and knochenbruch Model registration method, it is used for
Solve above-mentioned technical problem.
Summary of the invention
For the above-mentioned deficiency of the prior art, the present invention provides a kind of knochenbruch model rough registration method, system and knochenbruch mould
Type method for registering, for realizing the registration of knochenbruch model when knochenbruch can not provide axis.It is also used to improve the precision of registration.
In a first aspect, the present invention provides a kind of knochenbruch model rough registration method, comprising:
The CT image of fractured bones is reconstructed to obtain one group of knochenbruch model;
The corresponding point cloud data extracted on each knochenbruch model, is respectively converted into corresponding knochenbruch point Yun Mo for each knochenbruch model
Type;
The corresponding spin matrix and transposed matrix calculated between each knochenbruch point cloud model and sample bone point cloud model;Wherein,
The sample bone point cloud model is got by the conversion of sample bone model, and sample bone model passes through to the sample bone chosen in advance
CT image is reconstructed to obtain, and the reconstructing method of the sample bone model is identical as the reconstructing method of knochenbruch model, the sample
The conversion acquisition methods of bone point cloud model are identical as the conversion acquisition methods of knochenbruch point cloud model;
Each knochenbruch point cloud model is subjected to displacement rotating according to its corresponding spin matrix and transposed matrix respectively, is obtained
To knochenbruch point cloud model of each knochenbruch point cloud model after displacement rotating, rough registration is completed.
Further, the described corresponding spin matrix calculated between each knochenbruch point cloud model and sample bone point cloud model and
Transposed matrix, comprising steps of
The corresponding point cloud key point extracted on each knochenbruch point cloud model;
For each knochenbruch point cloud model: the point cloud key point that will be extracted from knochenbruch point cloud model respectively, with use
The point cloud key point that same method is extracted from sample bone point cloud model is matched, and the point cloud of one group of successful match is obtained
Key point pair;
Geometrical consistency method is respectively adopted, to corresponding cloud key point of each knochenbruch point cloud model to screening, sieves
The corresponding matching of knochenbruch point cloud model accurately point cloud key point pair is selected, and is denoted as target point cloud key point pair;
It is based respectively on the corresponding target point cloud key point pair of filtered out knochenbruch point cloud model, it is corresponding to calculate respectively
Spin matrix and transposed matrix between knochenbruch point cloud model and sample bone point cloud model.
Further, the point cloud key point that will be extracted from knochenbruch point cloud model respectively, it is same as using
The point cloud key point that method is extracted from sample bone point cloud model is matched, and the point cloud key point of one group of successful match is obtained
It is right, including key point describe sub- generation step and point cloud key point to generation step;
Wherein, the key point describes sub- generation step, comprising:
Calculate the local surfaces attribute of the corresponding point of each point cloud data in knochenbruch point cloud model;
Submethod is described using SHOT, based on point cloud key point on extracted knochenbruch point cloud model and extracted
The local surfaces attribute of the corresponding point of each point cloud data in knochenbruch point cloud model generates each cloud on knochenbruch point cloud model and closes
The SHOT of key point describes son;
The point cloud key point is to generation step, comprising:
The SHOT of each cloud key point on the knochenbruch point cloud model that sub- generation step generates is described based on above-mentioned key point
The sub and SHOT based on cloud key point each of on the sample bone point cloud model generated using same method of description is described
Son is searched for and is generated the point Yun Guanjian of successful match on knochenbruch point cloud model and sample bone point cloud model using Kd-Tree method
Point pair.
Further, using the NormalEstimationOMP class in PCL, each cloud in knochenbruch point cloud model is calculated
The local surfaces attribute of the corresponding point of data.
Further, the local surfaces attribute, including surface normal and curvature.
Further, Geometrical consistency method is respectively adopted, to corresponding cloud key point of each knochenbruch point cloud model into
Row screening filters out the corresponding matching of knochenbruch point cloud model accurately point cloud key point pair, including to each knochenbruch point cloud
Model carries out respectively: extracting corresponding cloud key point pair of knochenbruch point cloud model one by one, and each for what is currently extracted
A cloud key point pair, is handled as follows respectively:
The point cloud key point that r31, foundation currently extract obtains a cloud key point a to cloud key point a and b at corresponding
With b in knochenbruch point cloud model and corresponding coordinate in sample bone point cloud model, midpoint cloud key point a belongs to current knochenbruch point
Cloud model, point cloud key point b belong to the sample bone point cloud model;
R32, centered on the coordinate points of cloud key point a, set radius ra, searched in current knochenbruch point cloud model
It is all to be in radius raEach point cloud key point in range constitutes key point set A={ a1, a2..., an};And with point Yun Guanjian
Radius r is set centered on the coordinate points of point bb, searched in sample bone point cloud model all in radius rbEach point cloud in range
Key point constitutes key point set B={ b1, b2..., bm};
R33, crucial point set A is traversed, and counts the matched sample bone point cloud model of point cloud key point institute in crucial point set A
On point cloud key point appear in the total degree n in crucial point set B point, if n >=N, N be preset total degree threshold value and N
>=1, then it puts cloud key point a and b matching accurately and is a target point cloud key point pair, otherwise: point cloud key point a and b matching
Mistake deletes this cloud key point pair.
Further, the corresponding point cloud key point extracted on each knochenbruch point cloud model, comprising: key point extraction is respectively adopted
Method, the corresponding point cloud key point extracted on each knochenbruch point cloud model;
Wherein, the key point extraction method, are as follows: VoxelGrid filter is used, to extracted knochenbruch point cloud model
On point cloud data be filtered and sample, obtain the point cloud key point on knochenbruch point cloud model.
Second aspect, the present invention provide a kind of knochenbruch model rough registration system, comprising:
Knochenbruch model reconstruction unit is reconstructed to obtain one group of knochenbruch model for the CT image to fractured bones;
Model conversion unit is connected with knochenbruch model reconstruction unit, for the corresponding point cloud number extracted on each knochenbruch model
According to each knochenbruch model is respectively converted into corresponding knochenbruch point cloud model;
Computing unit is connected with model conversion unit, calculates each knochenbruch point cloud model and sample bone point Yun Mo for corresponding
Spin matrix and transposed matrix between type;Wherein, the sample bone point cloud model is got by the conversion of sample bone model, sample
This bone model is reconstructed to obtain by the CT image to the sample bone chosen in advance, the reconstructing method of the sample bone model with
The reconstructing method of knochenbruch model is identical, and the conversion of the conversion acquisition methods and knochenbruch point cloud model of the sample bone point cloud model obtains
Take method identical;
Rough registration unit, is connected with computing unit, is used for each knochenbruch point cloud model respectively according to its corresponding rotation
Torque battle array and transposed matrix carry out displacement rotating, obtain knochenbruch point cloud model of each knochenbruch point cloud model after displacement rotating, slightly
Registration is completed.
The third aspect, the present invention provide a kind of knochenbruch Model registration method, including rough registration and essence registration, in which:
The method for registering that the rough registration uses is knochenbruch model rough registration method as described above;
The essence registration: using ICP algorithm, and each knochenbruch point cloud model obtained to rough registration carries out space essence registration.
Further, the use ICP algorithm, each knochenbruch point cloud model obtained to rough registration carry out space essence and match
Standard, comprising steps of
When the quantity of knochenbruch point cloud model is 2, carried out by point set of the ICP algorithm to the two knochenbruch point cloud model accurate
Registration completes the engagement of two knochenbruch;
When the quantity of knochenbruch point cloud model is greater than 2:
It is any to choose two knochenbruch point cloud models, two selected knochenbruch point cloud models are accurately matched by ICP algorithm
Standard completes the engagement of corresponding two knochenbruch, obtains a new knochenbruch point cloud model;
Circulation uses ICP algorithm later, to obtained new knochenbruch point cloud model and remainder without accuracy registration
An arbitrary knochenbruch point cloud model carry out accuracy registration, complete the combination of current corresponding knochenbruch, until it is remaining without
The quantity of the knochenbruch point cloud model of accuracy registration is zero.
The beneficial effects of the present invention are:
(1) knochenbruch model rough registration method provided by the invention, system and knochenbruch Model registration method first calculate each
Spin matrix and transposed matrix between knochenbruch point cloud model and sample bone point cloud model later distinguish each knochenbruch point cloud model
It is rotated according to its corresponding spin matrix and transposed matrix, is carried out in entire registration process without using the axis of bone
Calibration, avoids the use to fractured bones axis, helps still to can be realized knochenbruch mould when knochenbruch can not provide axis
The rough registration of type registration.
(2) knochenbruch model rough registration method provided by the invention, system and knochenbruch Model registration method, by each knochenbruch point cloud
Model carries out swing offset matching according to corresponding spin matrix and transposed matrix respectively, does not use in entire registration process
The axis of bone is calibrated, and the use to bone axis is avoided, it is seen that rough registration of the invention is not only applicable in one-to-two
Knochenbruch scene, for one point three, one point four or even one points more knochenbruch scenes it is equally applicable, it is applied widely.
(3) knochenbruch model rough registration method provided by the invention, system and knochenbruch Model registration method, pass through spin matrix
The swing offset matching of knochenbruch point cloud model is realized with transposed matrix, matching process is not related to the size of fractured bones Yu sample bone
Difference problem is easy to implement.
(4) knochenbruch model rough registration method provided by the invention, system and knochenbruch Model registration method calculate respectively corresponding
When spin matrix and transposed matrix between knochenbruch point cloud model and sample bone point cloud model, obtained on knochenbruch point cloud model before this
Cloud key point pair is successfully put with first fit on sample bone point cloud model, matching is excluded using Geometrical consistency method again later
The point cloud key point pair of mistake obtains matching accurate target point cloud key point pair, later by matching accurate target point cloud
Key point pair calculates spin matrix and transposed matrix between knochenbruch point cloud model and sample bone point cloud model, it is seen that certain
Registration accuracy is improved in degree.
In addition, design principle of the present invention is reliable, structure is simple, has very extensive application prospect.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, for those of ordinary skill in the art
Speech, without creative efforts, is also possible to obtain other drawings based on these drawings.
Fig. 1 is the schematic flow chart of the knochenbruch model rough registration method of one embodiment of the invention.
Fig. 2 is a kind of schematic flow chart of embodiment of step S3 shown in Fig. 1.
Fig. 3 is the schematic block diagram of the knochenbruch model rough registration system of one embodiment of the invention.
Fig. 4 is the schematic flow chart of one embodiment of knochenbruch Model registration method of the present invention.
Specific embodiment
Technical solution in order to enable those skilled in the art to better understand the present invention, below in conjunction with of the invention real
The attached drawing in example is applied, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described implementation
Example is only a part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, this field is common
Technical staff's every other embodiment obtained without making creative work, all should belong to protection of the present invention
Range.
The Key Term occurred in the present invention is explained below.
Fig. 1 and Fig. 2 is the schematic flow chart of the knochenbruch model rough registration method of one embodiment of the invention.
Referring to Fig. 1 and Fig. 2, a kind of knochenbruch model rough registration method described in the present embodiment, includes the following steps S1-
S4。
Step S1: the CT image of fractured bones is reconstructed to obtain one group of knochenbruch model.
Correspondingly, using same reconstructing method, the CT image of sample bone is reconstructed to obtain the sample of sample bone
Bone model.The sample bone model can reconstruct to obtain in advance for threedimensional model, can also be in this method use process
It is reconstructed, for example can reconstruct to obtain in step sl.
It should be noted that sample bone involved in the present invention, shape is identical as the shape of fractured bones or substantially
It is identical.
Step S2: each knochenbruch model is respectively converted into corresponding disconnected by the corresponding point cloud data extracted on each knochenbruch model
Bone point cloud model.
Specifically by taking knochenbruch model A, knochenbruch Model B as an example:
The point cloud data on knochenbruch model A is extracted, knochenbruch model A is converted into corresponding knochenbruch point cloud model;
The point cloud data in knochenbruch Model B is extracted, knochenbruch Model B is converted into corresponding knochenbruch point cloud model.
Correspondingly, same conversion method can be used, in advance or in step s 2: extracting the point cloud on sample bone model
Sample bone model is converted to sample bone point cloud model by data.
Step S3: the corresponding spin matrix calculated between each knochenbruch point cloud model and sample bone point cloud model and displacement square
Battle array.
Specifically, the corresponding spin matrix calculated between each knochenbruch point cloud model and sample bone point cloud model and displacement square
Battle array, includes the following steps r1-r4.
Step r1: the corresponding point cloud key point extracted on each knochenbruch point cloud model.
In the present embodiment, the corresponding point cloud key point extracted on each knochenbruch point cloud model, comprising: key point is respectively adopted and mentions
It follows the example of, the corresponding point cloud key point extracted on each knochenbruch point cloud model.
Wherein, the key point extraction method, are as follows: VoxelGrid filter is used, to extracted knochenbruch point cloud model
On point cloud data be filtered and sample, obtain the point cloud key point on knochenbruch point cloud model.
The VoxelGrid filter is realized based on the VoxelGrid class in PCL.Correspondingly, the key
When point extraction method specific implementation: being mentioned by setInputCloud () the power function input in the VoxelGrid class in PCL
Each point cloud data of the knochenbruch point cloud model taken, and pass through setLeafSize () function letter in above-mentioned VoxelGrid class
Number setting calculates radius (may be configured as 5cm), later by filter () power function in VoxelGrid class, filters out disconnected
Point cloud key point on bone point cloud model.
Specifically, by taking knochenbruch point cloud model A as an example: passing through setInputCloud () function in PCL in VoxelGrid class
Energy function inputs each point cloud data of extracted knochenbruch point cloud model A, and passes through the setLeafSize in VoxelGrid class
The setting of () power function calculates radius (being set as 5cm), passes through filter () power function in VoxelGrid class, sieve later
Select the point cloud key point on knochenbruch point cloud model A.
The extraction of point cloud key point on other knochenbruch point cloud models, executes referring to above-mentioned knochenbruch point cloud model A.
In addition, the point cloud data on extracted sample bone point cloud model is filtered and is adopted using same method
Sample, obtains the point cloud key point on sample bone point cloud model, and specific extraction step includes: by the VoxelGrid class in PCL
SetInputCloud () power function input each point cloud data of extracted sample bone point cloud model, by above-mentioned
It is (corresponding with knochenbruch point cloud model to calculate half that setLeafSize () power function setting in VoxelGrid class calculates radius
Diameter is identical, is also configured as 5cm), later by filter () power function in VoxelGrid class, filter out sample bone point cloud
Point cloud key point on model.
Wherein, PCL (Point Cloud Library, Dian Yunku): being a kind of cross-platform open source C++ programming library of large size, it
The relevant general-purpose algorithm of a large amount of point clouds and efficient data structure are realized, cloud acquisition, filtering, segmentation, registration, an inspection are related to
Rope, feature extraction, identification, tracking, curve reestablishing, visualization etc.;Support several operation systems platform, can Windows,
Linux, Android, Mac OS X, it runs in partially embedded real-time system.
Step r2: for each knochenbruch point cloud model: the point cloud key point that will be extracted respectively from knochenbruch point cloud model,
It is matched with the above-mentioned point cloud key point extracted from sample bone point cloud model using same method, obtains one group of matching
Successfully point cloud key point pair.
Wherein, the point cloud key point that will be extracted respectively from knochenbruch point cloud model, with above-mentioned using same
The point cloud key point that method is extracted from sample bone point cloud model is matched, and the point cloud key point of one group of successful match is obtained
It is right, comprising: key point describes sub- generation step and puts cloud key point to generation step.
Wherein, key point described in the present embodiment describes sub- generation step, comprising:
Calculate the local surfaces attribute of the corresponding point of each point cloud data in knochenbruch point cloud model, including surface normal and song
Rate;
Submethod is described using SHOT, based on point cloud key point on extracted knochenbruch point cloud model and extracted
The local surfaces attribute of the corresponding point of each point cloud data in knochenbruch point cloud model generates each cloud on knochenbruch point cloud model and closes
The SHOT of key point describes son.
Wherein, using the NormalEstimationOMP class in PCL, each point cloud data in knochenbruch point cloud model is calculated
The local surfaces attribute of corresponding point, specifically includes:
1. the setInputCloud () power function by NormalEstimationOMP class inputs knochenbruch point cloud model
Upper extracted each point cloud data;
2. calculating radius by setRadiusSearch () the power function setting of NormalEstimationOMP class (can
It is set as 10cm);
By executing compute () power function of NormalEstimationOMP class after 3., calculates and obtain knochenbruch
The local surfaces attribute of the corresponding point of each point cloud data in point cloud model.
Wherein, submethod is described using SHOT, based on extracted knochenbruch point cloud model point cloud key point and institute
The local surfaces attribute of the corresponding point of each point cloud data in the knochenbruch point cloud model of extraction generates each on knochenbruch point cloud model
The SHOT of point cloud key point describes son, specifically includes:
Knochenbruch point Yun Mo is inputted by the setInputCloud () power function in PCL in SHOTEstimationOMP class
All point cloud key points of type;
Knochenbruch point cloud model is inputted by the setSearchSurface () power function in SHOTEstimationOMP class
The point cloud data of upper extraction;
Calculating radius by setRadiusSearch () the power function setting in SHOTEstimationOMP class (can set
It is set to 10cm);
Knochenbruch point cloud model is inputted by the setInputNormals () power function in SHOTEstimationOMP class
In the corresponding point of each point cloud data local surfaces attribute;
Compute () power function in SHOTEstimationOMP class is executed, each point on knochenbruch point cloud model is generated
The SHOT of cloud key point describes son.
NormalEstimationOMP class is a class in PCL, is estimated using OpenMP standard parallel each
The local surfaces attribute of 3D point, such as the surface normal and curvature of 3D point.
Correspondingly, it can refer to above-mentioned key point and describe sub- generation step, generate each point on sample bone point cloud model
The SHOT of cloud key point describes son.
In the present embodiment, the point cloud key point is to generation step, comprising: is generated based on above-mentioned key point description
The SHOT description of each cloud key point is sub on the knochenbruch point cloud model that step generates and is based on generating using same method
Sample bone point cloud model on each of cloud key point SHOT describe son, searched for using Kd-Tree method and generate knochenbruch
The point cloud key point pair of successful match on point cloud model and sample bone point cloud model.
Wherein, Kd-Tree is a kind of data structure for dividing k dimension data space, is mainly used in hyperspace key number
According to search.
Wherein, described that each cloud pass on the knochenbruch point cloud model that sub- generation step generates is described based on above-mentioned key point
The SHOT description of key point is sub and based on cloud key point each of on the sample bone point cloud model generated using same method
SHOT is described, search for using Kd-Tree method and generates successful match on knochenbruch point cloud model and sample bone point cloud model
Point cloud key point pair, specifically include:
By setInputCloud () power function of KdTreeFLANN class in PCL, it is corresponding to input knochenbruch point cloud model
All descriptions;
Pass through nearestKSearch () power function of KdTreeFLANN class in PCL, input sample bone point cloud model pair
All description answered;
By executing above-mentioned nearestKSearch () power function, finds and calculate knochenbruch point cloud model and sample bone point
The difference of two squares between cloud model between each point description, and the resulting difference of two squares will be calculated less than preset difference of two squares threshold value
Corresponding the CorrespondencesPtr () power function for being transferred to KdTreeFLANN class of each group description;It wherein, will be every
Group description is denoted as a description subgroup, and description in each description subgroup group respectively constitutes knochenbruch point cloud model and sample bone
A cloud key point pair of successful match on point cloud model;
Later by above-mentioned CorrespondencesPtr () power function, description generated in each description subgroup group is right
The corresponding relationship between point cloud key point answered, deposit registration point list;It is registrated the Dian Yunguan in point list with corresponding relationship
Key point constitutes the point cloud key point pair of successful match on knochenbruch point cloud model and sample bone point cloud model.
Wherein each description subgroup is by the point cloud key point and a sample bone point cloud on a knochenbruch point cloud model
Point cloud key point composition on model describes the corresponding two cloud key points of two descriptions in subgroup group, as knochenbruch
The point cloud key point pair of successful match on point cloud model and sample bone point cloud model;It can be seen that there being knochenbruch point cloud in registration point list
All point cloud key points pair of successful match on model and sample bone point cloud model.
For describing subgroup A: description subgroup A describes sub two description by the first description and second and is constituted, and first
Point cloud key point d1, the second corresponding sample bone point searched of description on the corresponding knochenbruch point cloud model searched of description
Point cloud key point c1 on cloud model, the difference of two squares putting cloud key point d1 and putting between cloud key point c1 are less than preset put down
Variance threshold values;Point cloud key point d1 and point cloud key point c1 are stored in registration point list with corresponding to each other, and are knochenbruch point Yun Mo
A cloud key point pair of successful match in type and sample bone point cloud model.
Wherein, the value of above-mentioned difference of two squares threshold value in the present embodiment is 0.25, when specific implementation, those skilled in the art
It can modify according to actual needs to the specific value of above-mentioned difference of two squares threshold value.
Step r3: being respectively adopted Geometrical consistency method, to corresponding cloud key point of each knochenbruch point cloud model to progress
Screening filters out the corresponding matching of knochenbruch point cloud model accurately point cloud key point pair, and is denoted as target point cloud key point
It is right.Wherein, two cloud key points of target point cloud key point centering are target point cloud key point.
Wherein, Geometrical consistency method is respectively adopted, to corresponding cloud key point of each knochenbruch point cloud model to sieving
Choosing filters out the corresponding matching of knochenbruch point cloud model accurately point cloud key point pair, including to each knochenbruch point cloud model
It carries out respectively:
Corresponding cloud key point pair of knochenbruch point cloud model, and each point cloud for currently extracting are extracted one by one
Key point pair carries out the processing of following steps r31-r33 respectively:
The point cloud key point that r31, foundation currently extract obtains a cloud key point a to cloud key point a and b at corresponding
With b in knochenbruch point cloud model and corresponding coordinate in sample bone point cloud model, midpoint cloud key point a belongs to current knochenbruch point
Cloud model, point cloud key point b belong to the sample bone point cloud model;
R32, centered on the coordinate points of cloud key point a, set radius ra, searched in current knochenbruch point cloud model
It is all to be in radius raEach point cloud key point in range constitutes key point set A={ a1, a2..., an};And with point Yun Guanjian
Radius r is set centered on the coordinate points of point bb, searched in sample bone point cloud model all in radius rbEach point cloud in range
Key point constitutes key point set B={ b1, b2..., bm};
R33, crucial point set A is traversed, and counts the matched sample bone point cloud model of point cloud key point institute in crucial point set A
On point cloud key point appear in the total degree n in crucial point set B point, if n >=N, N be preset total degree threshold value and N
>=1, then it puts cloud key point a and b matching accurately and is a target point cloud key point pair, otherwise: point cloud key point a and b matching
Mistake deletes this cloud key point pair.
In the present embodiment, in step r33, the deletion this cloud key point pair, comprising: by the cloud
Key point is deleted from above-mentioned registration point list.As it can be seen that step r3 can be updated the registration point list, obtain
Updated registration point list.The updated registration point list is denoted as key point list of matches, then key point matching column
What is stored in table is knochenbruch point cloud model and the target point cloud key point pair on sample bone point cloud model and its to be registrated relationship (i.e. disconnected
The corresponding relationship of target point cloud key point on bone point cloud model and sample bone point cloud model).
Step r4: being based respectively on the corresponding target point cloud key point pair of filtered out knochenbruch point cloud model, corresponding
Calculate the spin matrix and transposed matrix between each knochenbruch point cloud model and sample bone point cloud model.
In the present embodiment, corresponding to calculate each knochenbruch using the GeometricConsistencyGrouping class in PCL
Spin matrix and transposed matrix between point cloud model and sample bone point cloud model.Specifically, for arbitrary knochenbruch point
For cloud model (lower abbreviation knochenbruch point cloud model M1):
On knochenbruch point cloud model M1 target point cloud key point (coordinate) be M11 (x1, y1, z1), M12 (x2, y2, z2) ...,
M1p (xp, yp, zp), p are the quantity of target point cloud key point on knochenbruch point cloud model M1;
Target point cloud key point M11 (x1, y1, z1) on knochenbruch point cloud model M1, M12 (x2, y2, z2) ..., M1p (xp,
Yp, zp), on sample bone point cloud model corresponding target point cloud key point be followed successively by J11 (x1 ', y1 ', z1 '), J12 (x2 ',
Y2 ', z2 ') ..., J1q (xq ', yq ', zq '), q is the quantity of the target point cloud key point on sample bone point cloud model, p=q;
It is defeated by setInputCloud () power function in PCL in GeometricConsistencyGrouping class
Enter each target point cloud key point (coordinate) on knochenbruch point cloud model M1: M11 (x1, y1, z1), M12 (x2, y2, z2) ..., M1p
(xp, yp, zp);
By setSceneCloud () power function in GeometricConsistencyGrouping class, sample is inputted
Target point cloud key point on this bone point cloud model: J11 (x1 ', y1 ', z1 '), J12 (x2 ', y2 ', z2 ') ..., J1q (xq ',
Yq ', zq ');
Pass through setModelSceneCorrespondences () function of GeometricConsistencyGrouping class
Energy function, input knochenbruch point cloud model M1 match corresponding key point with the target point cloud key point on sample bone point cloud model
List;
Recognize () power function in GeometricConsistencyGrouping class is executed, calculates and exports disconnected
Spin matrix and transposed matrix of the bone point cloud model M1 to sample bone point cloud model.
Step S4: each knochenbruch point cloud model is displaced according to its corresponding spin matrix and transposed matrix respectively
Rotation, rough registration are completed.
Based on step S4, knochenbruch point cloud model of each knochenbruch point cloud model after displacement rotating is obtained.
It should be noted that the present invention after the completion of rough registration, is actually the equal of that all knochenbruch point cloud models are mobile
Onto same Z-direction.
It is further to note that model involved in the present invention is threedimensional model.
To sum up, the knochenbruch model rough registration method, first calculates between each knochenbruch point cloud model and sample bone point cloud model
Spin matrix and transposed matrix, later by each knochenbruch point cloud model respectively according to its corresponding spin matrix and displacement square
Battle array is rotated, to complete the rough registration of each knochenbruch point cloud model, it is seen that the axis of bone is not used in entire registration process
It is calibrated: this aspect, so that the present invention helps that the registration that the knochenbruch of axis realizes knochenbruch model can not be provided;It is another
Aspect makes the registration of the knochenbruch model of the invention for being applicable not only to one-to-two, applies also for one point three, one point four, even one point
The registration of more knochenbruch models;Furthermore the present invention realizes knochenbruch by spin matrix and transposed matrix in thick matching step
The rotation matching of point cloud model is not related to the difference in size problem of fractured bones Yu sample bone in matching process, is easy to implement.
In addition, knochenbruch model rough registration method provided by the invention, obtained before this on knochenbruch point cloud model with sample bone
First fit successfully puts cloud key point pair on point cloud model, excludes the point of matching error using Geometrical consistency method again later
Cloud key point pair obtains matching accurate target point cloud key point pair, later by matching accurate target point cloud key point pair,
Calculate the spin matrix and transposed matrix between knochenbruch point cloud model and sample bone point cloud model, it is seen that improve to a certain extent
Registration accuracy.
Fig. 3 is the schematic block diagram of one embodiment of knochenbruch model rough registration system of the present invention.
Referring to Fig. 3, which includes:
Knochenbruch model reconstruction unit is reconstructed to obtain one group of knochenbruch model for the CT image to fractured bones;
Model conversion unit converts each knochenbruch model for the corresponding point cloud data extracted on each knochenbruch model respectively
For corresponding knochenbruch point cloud model;
Computing unit, for the corresponding spin matrix calculated between each knochenbruch point cloud model and sample bone point cloud model and position
Move matrix;Wherein, the sample bone point cloud model is got by the conversion of sample bone model, and sample bone model passes through to preparatory choosing
The CT image of the sample bone taken is reconstructed to obtain, the reconstructing method phase of the reconstructing method and knochenbruch model of the sample bone model
Together, the conversion acquisition methods of the sample bone point cloud model are identical as the conversion acquisition methods of knochenbruch point cloud model;
Rough registration unit is used for each knochenbruch point cloud model respectively according to its corresponding spin matrix and transposed matrix
Displacement rotating is carried out, knochenbruch point cloud model of each knochenbruch point cloud model after displacement rotating is obtained, rough registration is completed.
Each component units of the knochenbruch model rough registration system, its implementation respectively with above-mentioned knochenbruch model rough registration side
Corresponding part in method is corresponding.When the knochenbruch model rough registration system uses: first passing through knochenbruch model reconstruction unit to bone
The CT image of folding bone is reconstructed to obtain one group of knochenbruch model;It is corresponding to extract knochenbruch model later by model conversion unit
The point cloud data on each knochenbruch model that reconfiguration unit reconstructs, is then respectively converted into corresponding knochenbruch for each knochenbruch model
Point cloud model;Later by computing unit, it is corresponding calculate each knochenbruch point cloud model (being obtained by model conversion cell translation) with
Spin matrix and transposed matrix between sample bone point cloud model;Later by rough registration unit, by each knochenbruch point cloud model point
Displacement rotating is not carried out according to its corresponding spin matrix and transposed matrix (being calculated by computing unit), is obtained each
Knochenbruch point cloud model of the knochenbruch point cloud model after displacement rotating, then completes rough registration.
It is corresponding with above-mentioned knochenbruch Model registration method in view of the knochenbruch model rough registration system, there is above-mentioned knochenbruch model
The all advantages of method for registering, details are not described herein.
Based on the present invention, after rough registration, all knochenbruch models, which are equivalent to, has been moved to same Z-direction, but by
Difference in size is often deposited in sample bone model and knochenbruch model, causes the knochenbruch section between knochenbruch cannot perfect weight
It closes, for this purpose, the present invention provides a kind of knochenbruch Model registration method, for the disconnected of each knochenbruch point cloud model that rough registration is completed
Bone section is further matched.Fig. 4 is the schematic block diagram of the knochenbruch Model registration method provided by the invention.
Referring to fig. 4, which includes rough registration and essence registration, in which:
(1) rough registration
The method for registering that the rough registration of the knochenbruch Model registration method uses is knochenbruch model as described above rough registration side
Method.The rough registration of the knochenbruch Model registration method can refer to knochenbruch model rough registration method as described above and is registrated.Slightly match
After standard, each corresponding knochenbruch point cloud model is obtained.
(2) essence registration
The essence registration of the knochenbruch Model registration method: ICP algorithm, each knochenbruch point cloud model obtained to rough registration are used
Carry out space essence registration.To simplify statement, involved in the present embodiment following " 1 " and " 2 " two parts " knochenbruch point cloud model "
It is the obtained each corresponding knochenbruch point cloud model of rough registration.
1, it when the quantity of knochenbruch point cloud model is 2, is carried out by point set of the ICP algorithm to two knochenbruch point cloud models accurate
Registration.
Specifically, using ICP (iteration closest approach) algorithm to the point Yun Jinhang in two knochenbruch sections of knochenbruch point cloud model
Successive ignition calculates, and is repeatedly finely tuned to the relative position of knochenbruch point cloud model, until between the point set in two knochenbruch sections
Space length it is minimum, realize the space essence registration of knochenbruch point cloud model:
(1) two knochenbruch Section Point Clouds on two knochenbruch point cloud models are denoted as target point cloud and source point cloud respectively;
(2) target point cloud midpoint cloud is denoted as point set P={ pi, i=1,2 ..., n };
(3) for each of point set P point, the corresponding points nearest away from the point, composition and point are all found out in source point cloud
Collect the corresponding point set Q={ q of Pi, i=1,2 ..., n } so that | | pi-qi| |=min, even if invocation point qiIt is in source point cloud
The range points p found outiNearest point;
(4) it is calculated using lowest mean square root method and is registrated matrix between point set P and Q, make to obtain registration transformation matrix R and t,
Wherein R is 3 × 3 spin matrix, and t is 3 × 1 transposed matrix;
(5) it to point set P, is coordinately transformed using spin matrix R and transposed matrix t, obtains new point set P ', i.e. P '=
RP+t;
(6) the average distance d of point set P ' and point set Q are calculated;
(7) if average distance d is less than given limiting threshold value e, following (8) is continued to execute, are otherwise executed following
(9);
(8) iterative calculation terminates, and realizes target point cloud and source point using current newest spin matrix R and transposed matrix t
The disconnected knochenbruch section splicing of two knochenbruch point cloud models is completed in swing offset operation between the affiliated knochenbruch point cloud model of cloud;
(9) P is replaced with point set P ', repeated above-mentioned (3)-(7).
2, when the quantity of knochenbruch point cloud model is greater than 2:
It is any to choose two knochenbruch point cloud models, two selected knochenbruch point cloud models are accurately matched by ICP algorithm
Standard, completes the engagement of corresponding two knochenbruch, and obtains a new knochenbruch point cloud model;
Circulation uses ICP algorithm later, to obtained new knochenbruch point cloud model and remainder without accuracy registration
Any one knochenbruch point cloud model carry out accuracy registration, the combination of current corresponding knochenbruch is completed, until remaining without essence
The quantity for the knochenbruch point cloud model being really registrated is zero.
Knochenbruch model rough registration method as described above is based in view of the knochenbruch Model registration method and carries out rough registration, is had
The all advantages of above-mentioned knochenbruch Model registration method, details are not described herein.
It should be noted that part same and similar between each embodiment in this specification may refer to each other.In particular,
For system embodiments, since it is substantially similar to knochenbruch Model registration embodiment of the method, so be described relatively simple,
Related place is referring to the explanation in knochenbruch Model registration embodiment of the method.
In conclusion the present invention still can be realized the registration of knochenbruch model when knochenbruch can not provide axis, and use
Geometrical consistency method, to screening, improves registration accuracy to cloud key point to a certain extent.
Although by reference to attached drawing and combining the mode of preferred embodiment to the present invention have been described in detail, the present invention
It is not limited to this.Without departing from the spirit and substance of the premise in the present invention, those of ordinary skill in the art can be to the present invention
Embodiment carry out various equivalent modifications or substitutions, and these modifications or substitutions all should in covering scope of the invention/appoint
What those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, answer
It is included within the scope of the present invention.Therefore, protection scope of the present invention is answered described is with scope of protection of the claims
It is quasi-.
Claims (10)
1. a kind of knochenbruch model rough registration method characterized by comprising
The CT image of fractured bones is reconstructed to obtain one group of knochenbruch model;
The corresponding point cloud data extracted on each knochenbruch model, is respectively converted into corresponding knochenbruch point cloud model for each knochenbruch model;
The corresponding spin matrix and transposed matrix calculated between each knochenbruch point cloud model and sample bone point cloud model;Wherein, described
Sample bone point cloud model got by the conversion of sample bone model, sample bone model passes through CT to the sample bone chosen in advance figure
As being reconstructed to obtain, the reconstructing method of the sample bone model is identical as the reconstructing method of knochenbruch model, the sample bone point
The conversion acquisition methods of cloud model are identical as the conversion acquisition methods of knochenbruch point cloud model;
Each knochenbruch point cloud model is subjected to displacement rotating according to its corresponding spin matrix and transposed matrix respectively, is obtained each
Knochenbruch point cloud model of the knochenbruch point cloud model after displacement rotating, rough registration are completed.
2. knochenbruch model rough registration method according to claim 1, which is characterized in that corresponding each knochenbruch point of calculating
Spin matrix and transposed matrix between cloud model and sample bone point cloud model, comprising steps of
The corresponding point cloud key point extracted on each knochenbruch point cloud model;
For each knochenbruch point cloud model: the point cloud key point that will be extracted from knochenbruch point cloud model respectively, it is same as using
The point cloud key point that is extracted from sample bone point cloud model of method matched, obtain the point Yun Guanjian of one group of successful match
Point pair;
Geometrical consistency method is respectively adopted, to corresponding cloud key point of each knochenbruch point cloud model to screening, filters out
Cloud key point pair is accurately put in the corresponding matching of knochenbruch point cloud model, and is denoted as target point cloud key point pair;
It is based respectively on the corresponding target point cloud key point pair of filtered out knochenbruch point cloud model, correspondence calculates each knochenbruch
Spin matrix and transposed matrix between point cloud model and sample bone point cloud model.
3. knochenbruch model rough registration method according to claim 2, which is characterized in that described respectively will be from knochenbruch point cloud
The point cloud key point extracted on model, with the point cloud key point extracted from sample bone point cloud model using same method
It is matched, obtains the point cloud key point pair of one group of successful match, including key point describes sub- generation step and puts cloud key point
To generation step;
Wherein, the key point describes sub- generation step, comprising:
Calculate the local surfaces attribute of the corresponding point of each point cloud data in knochenbruch point cloud model;
Submethod is described using SHOT, based on extracted knochenbruch point cloud model point cloud key point and extracted knochenbruch
The local surfaces attribute of the corresponding point of each point cloud data, generates each cloud key point on knochenbruch point cloud model in point cloud model
SHOT describe son;
The point cloud key point is to generation step, comprising:
The SHOT description of each cloud key point on the knochenbruch point cloud model that sub- generation step generates is described based on above-mentioned key point
Sub and based on cloud key point each of on the sample bone point cloud model generated using same method SHOT describes son,
The point cloud key point of successful match on knochenbruch point cloud model and sample bone point cloud model is searched for and generated using Kd-Tree method
It is right.
4. knochenbruch model rough registration method according to claim 3, which is characterized in that further, using in PCL
NormalEstimationOMP class calculates the local surfaces attribute of the corresponding point of each point cloud data in knochenbruch point cloud model.
5. knochenbruch model rough registration method according to claim 3, which is characterized in that the local surfaces attribute, packet
Include surface normal and curvature.
6. knochenbruch model rough registration method according to claim 2, which is characterized in that Geometrical consistency side is respectively adopted
Method filters out knochenbruch point cloud model corresponding to corresponding cloud key point of each knochenbruch point cloud model to screening
With accurate point cloud key point pair, including each knochenbruch point cloud model is carried out respectively: it is corresponding to extract knochenbruch point cloud model one by one
Point cloud key point pair, and for currently extract each point cloud key point pair, be handled as follows respectively:
The point cloud key point that r31, foundation currently extract obtains cloud key point an a and b and exists to corresponding cloud key point a and b
Knochenbruch point cloud model belongs to current knochenbruch point Yun Mo with corresponding coordinate in sample bone point cloud model, midpoint cloud key point a
Type, point cloud key point b belong to the sample bone point cloud model;
R32, centered on the coordinate points of cloud key point a, set radius ra, searched in current knochenbruch point cloud model all
In radius raEach point cloud key point in range constitutes key point set A={ a1, a2..., an};And with cloud key point b's
Radius r is set centered on coordinate pointsb, searched in sample bone point cloud model all in radius rbEach point cloud in range is crucial
Point constitutes key point set B={ b1, b2..., bm};
R33, traverse crucial point set A, and count the point cloud key point in crucial point set A on matched sample bone point cloud model
Point cloud key point appear in the total degree n in crucial point set B point, if n >=N, N be preset total degree threshold value and N >=1,
Cloud key point a and b is then put to match accurate and be a target point cloud key point pair, otherwise: point cloud key point a and b matching error,
Delete this cloud key point pair.
7. knochenbruch model rough registration method according to claim 2, which is characterized in that corresponding to extract each knochenbruch point cloud model
On point cloud key point, comprising: be respectively adopted key point extraction method, the corresponding point Yun Guanjian extracted on each knochenbruch point cloud model
Point;
Wherein, the key point extraction method, are as follows: VoxelGrid filter is used, on extracted knochenbruch point cloud model
Point cloud data is filtered and samples, and obtains the point cloud key point on knochenbruch point cloud model.
8. a kind of knochenbruch model rough registration system characterized by comprising
Knochenbruch model reconstruction unit is reconstructed to obtain one group of knochenbruch model for the CT image to fractured bones;
Each knochenbruch model is respectively converted into pair for the corresponding point cloud data extracted on each knochenbruch model by model conversion unit
The knochenbruch point cloud model answered;
Computing unit, for the corresponding spin matrix calculated between each knochenbruch point cloud model and sample bone point cloud model and displacement square
Battle array;Wherein, the sample bone point cloud model is got by the conversion of sample bone model, and sample bone model by choosing in advance
The CT image of sample bone is reconstructed to obtain, and the reconstructing method of the sample bone model is identical as the reconstructing method of knochenbruch model,
The conversion acquisition methods of the sample bone point cloud model are identical as the conversion acquisition methods of knochenbruch point cloud model;
Rough registration unit, for carrying out each knochenbruch point cloud model according to its corresponding spin matrix and transposed matrix respectively
Displacement rotating, obtains knochenbruch point cloud model of each knochenbruch point cloud model after displacement rotating, and rough registration is completed.
9. a kind of knochenbruch Model registration method, including rough registration and essence registration, which is characterized in that
The method for registering that the rough registration uses is knochenbruch model rough registration side described in claims 1 or 2 or 3 or 4 or 5 or 6
Method;
The essence registration: using ICP algorithm, and each knochenbruch point cloud model obtained to rough registration carries out space essence registration.
10. knochenbruch Model registration method according to claim 9, which is characterized in that described uses ICP algorithm, to thick
It is registrated obtained each knochenbruch point cloud model and carries out space essence registration, comprising steps of
When the quantity of knochenbruch point cloud model is 2, accurately matched by point set of the ICP algorithm to the two knochenbruch point cloud model
Standard completes the engagement of two knochenbruch;
When the quantity of knochenbruch point cloud model is greater than 2:
It is any to choose two knochenbruch point cloud models, accuracy registration is carried out to two selected knochenbruch point cloud models by ICP algorithm, it is complete
At the engagement of corresponding two knochenbruch, a new knochenbruch point cloud model is obtained;
Circulation uses ICP algorithm later, to obtained new knochenbruch point cloud model and remaining appointing without accuracy registration
One knochenbruch point cloud model of meaning carries out accuracy registration, the combination of current corresponding knochenbruch is completed, until remaining without accurate
The quantity of the knochenbruch point cloud model of registration is zero.
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 true CN110288638A (en) | 2019-09-27 |
CN110288638B 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) |
Cited By (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 |
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 |
CN112764651A (en) * | 2021-02-01 | 2021-05-07 | 飞燕航空遥感技术有限公司 | Method and system for drawing three-dimensional point cloud profile of browser |
CN113284252A (en) * | 2021-03-03 | 2021-08-20 | 山东交通学院 | Fracture model resetting method and system based on VTK |
CN113345112A (en) * | 2021-05-25 | 2021-09-03 | 上海大学 | Long bone fracture section 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 |
CN114274139A (en) * | 2020-09-27 | 2022-04-05 | 西门子股份公司 | Automatic spraying method, device, system and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150087965A1 (en) * | 2013-09-20 | 2015-03-26 | Junichi Tokuda | System and method for automatic detection and registration of medical images |
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 |
-
2019
- 2019-06-18 CN CN201910527855.2A patent/CN110288638B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150087965A1 (en) * | 2013-09-20 | 2015-03-26 | Junichi Tokuda | System and method for automatic detection and registration of medical images |
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 |
Cited By (11)
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 |
CN114274139A (en) * | 2020-09-27 | 2022-04-05 | 西门子股份公司 | Automatic spraying method, device, system and storage medium |
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 |
CN112764651A (en) * | 2021-02-01 | 2021-05-07 | 飞燕航空遥感技术有限公司 | Method and system for drawing three-dimensional point cloud profile of browser |
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 |
CN113284252A (en) * | 2021-03-03 | 2021-08-20 | 山东交通学院 | Fracture model resetting method and system based on VTK |
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 |
CN113345112A (en) * | 2021-05-25 | 2021-09-03 | 上海大学 | Long bone fracture section 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 |
Also Published As
Publication number | Publication date |
---|---|
CN110288638B (en) | 2022-07-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110288638A (en) | A kind of knochenbruch model rough registration method, system and knochenbruch Model registration method | |
CN110363858B (en) | Three-dimensional face reconstruction method and system | |
CN101339670B (en) | Computer auxiliary three-dimensional craniofacial rejuvenation method | |
CN104851126B (en) | Threedimensional model dividing method and device based on generalized cylinder | |
CN104463894A (en) | Overall registering method for global optimization of multi-view three-dimensional laser point clouds | |
CN112562082A (en) | Three-dimensional face reconstruction method and system | |
Zhang et al. | Computer-aided design of dental inlay restoration based on dual-factor constrained deformation | |
Sinko et al. | 3D registration of the point cloud data using ICP algorithm in medical image analysis | |
CN102254317A (en) | Method for automatically extracting dental arch curved surface in dental implantation navigation | |
CN108038862A (en) | A kind of Interactive medical image intelligent scissor modeling method | |
CN103632371A (en) | Compatibility mesh segmentation based skeleton parameter computation method | |
CN109242951A (en) | A kind of face's real-time three-dimensional method for reconstructing | |
CN109903319A (en) | A kind of iteratively faster closest approach registration Algorithm based on multiresolution | |
Campen et al. | A Practical Guide to Polygon Mesh Repairing. | |
CN115619773B (en) | Three-dimensional tooth multi-mode data registration method and system | |
Feng et al. | Repair volume extraction method for damaged parts in remanufacturing repair | |
CN109118455B (en) | Ancient human skull craniofacial interactive restoration method based on modern soft tissue distribution | |
CN107123163A (en) | A kind of plant three-dimensional reconstruction system based on various visual angles stereoscopic vision | |
CN105261062A (en) | Character segmented modeling method | |
CN112802046A (en) | Image generation system for generating pseudo CT from multi-sequence MR based on deep learning | |
CN105809730A (en) | Long bone fracture surface point cloud extraction method | |
CN108898578A (en) | A kind of processing method of medical image, device and computer storage medium | |
CN112102491A (en) | Skin damage surface three-dimensional reconstruction method based on surface structured light | |
CN117274339A (en) | Point cloud registration method based on improved ISS-3DSC characteristics combined with ICP | |
CN111275811A (en) | Repair model creation method, system and storage medium |
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 |