CN115797416A - Image reconstruction method, device and equipment based on point cloud image and storage medium - Google Patents
Image reconstruction method, device and equipment based on point cloud image and storage medium Download PDFInfo
- Publication number
- CN115797416A CN115797416A CN202211633652.XA CN202211633652A CN115797416A CN 115797416 A CN115797416 A CN 115797416A CN 202211633652 A CN202211633652 A CN 202211633652A CN 115797416 A CN115797416 A CN 115797416A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- image
- images
- target
- processed
- 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.)
- Pending
Links
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention relates to the field of data processing, and discloses an image reconstruction method, an image reconstruction device, image reconstruction equipment and a storage medium based on a point cloud image, which are used for improving the accuracy of image reconstruction according to point cloud data. The method comprises the following steps: based on a preset image acquisition rule, carrying out point cloud image acquisition on the head position of a target patient through a preset binocular camera to obtain a plurality of groups of point cloud images; carrying out registration processing on a plurality of groups of point cloud images through a preset iterative algorithm to obtain candidate point cloud images; carrying out accuracy correction on the candidate point cloud image to obtain a target point cloud image; performing target point position analysis on the target point cloud image to determine a plurality of target point positions; and transmitting the positions of the target points to a preset data processing terminal for visual display.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to an image reconstruction method, device and equipment based on a point cloud image and a storage medium.
Background
With the development of internet technology, the transcranial magnetic stimulation technology plays an irreplaceable role in treatment of senile nervous system diseases in recent years due to the advantages of being noninvasive, simple to operate and the like, and the main principle is that three-dimensional reconstruction of a head model is carried out by using a nuclear magnetic resonance image of a patient based on a personalized three-dimensional image, and accurate positioning of an image guide stimulation coil is achieved through registration and registration of the three-dimensional image and the space where the patient is located.
However, in actual clinical practice, in some cases, for example, a magnetic or mechanical implantable device exists in a body, or a patient suffers from a disease such as epilepsy, so that a nuclear magnetic resonance image of the patient cannot be acquired, and in the current reconstruction method of reconstructing a three-dimensional image by point cloud data to replace a nuclear magnetic resonance image, there is a problem of insufficient accuracy when the point cloud image is subjected to data registration.
Disclosure of Invention
In view of this, embodiments of the present invention provide an image reconstruction method, an image reconstruction device, and an image reconstruction system based on a point cloud image, which solve the technical problem of insufficient accuracy when performing image reconstruction by using a point cloud image.
The invention provides an image reconstruction method based on a point cloud image, which comprises the following steps: based on a preset image acquisition rule, carrying out point cloud image acquisition on the head position of a target patient through a preset binocular camera to obtain a plurality of groups of point cloud images; registering the multiple groups of point cloud images through a preset iterative algorithm to obtain candidate point cloud images; carrying out accuracy correction on the candidate point cloud image to obtain a target point cloud image; performing target point position analysis on the target point cloud image to determine a plurality of target point positions; and transmitting the positions of the target points to a preset data processing terminal for visual display.
The image reconstruction method based on the point cloud image is based on the preset image acquisition rule, the preset binocular camera is used for acquiring the point cloud image of the head of the target patient to obtain a plurality of groups of point cloud images, and the acquired point cloud images are acquired by acquiring one group of data at intervals of 10 degrees, so that each group of point cloud can be ensured to have a better point cloud structure and two adjacent groups of point clouds can have better coincidence rate, and the final registration result is improved on the original data. Registering the multiple groups of point cloud images through a preset iterative algorithm to obtain candidate point cloud images; the candidate point cloud image is corrected in accuracy to obtain a target point cloud image, the candidate point cloud image is corrected in accuracy, namely outliers in the candidate point cloud image are processed to enable global convergence to be optimal, the accuracy of point cloud data registration is further improved,
with reference to the first aspect, in a first implementation manner of the first aspect, before the step of performing point cloud image acquisition on the head position of a target patient through a preset binocular camera based on a preset image acquisition rule to obtain multiple groups of point cloud images, the method further includes: constructing a target coordinate system, carrying out position analysis on the head position of a target patient, and determining head coordinate information of the head position in the target coordinate system; according to a preset distance, an image acquisition track is constructed through the head coordinate information, and a target image acquisition track is obtained; determining image acquisition point positions of the target image acquisition track through a preset angle to obtain a plurality of image acquisition point positions; and generating an image acquisition rule through the plurality of image acquisition point positions and the target image acquisition track to obtain the image acquisition rule.
With reference to the implementation manner of the first aspect, in a second implementation manner of the first aspect, the registering the multiple groups of point cloud images through a preset iterative algorithm to obtain a target point cloud image includes: carrying out image matching on the plurality of groups of point cloud images to obtain a plurality of groups of point cloud images to be processed; and carrying out image registration processing on the plurality of groups of point cloud images to be processed to obtain a target point cloud image.
With reference to the second implementation manner of the first aspect, in a third implementation manner of the first aspect, the step of performing image registration processing on the multiple groups of point cloud images to be processed to obtain candidate point cloud images includes: performing overlapping part analysis on each group of point cloud images to be processed in the plurality of groups of point cloud images to be processed, and determining overlapping part point cloud data sets corresponding to each group of point cloud images to be processed; respectively carrying out point cloud registration on the point cloud data sets of the overlapped parts corresponding to each group of point cloud images to be processed based on a preset registration function to obtain a registered point cloud data set; and performing point cloud image splicing through the registered point cloud data set to generate a candidate point cloud image.
According to the scheme, the server performs sub-classification on the rough classification result according to the adjacent point cloud sets corresponding to each group of point cloud images to be processed, determines the overlapped part point cloud data sets corresponding to each group of point cloud images to be processed, solves the problem that the overlapped part in the adjacent point cloud images is not accurately segmented in the existing classification method, improves the classification accuracy, provides an accurate data base for subsequent point cloud image registration, and further improves the accuracy of point cloud data registration.
With reference to the third implementation manner of the first aspect, in a fourth implementation manner of the first aspect, the step of performing a coincidence analysis on each group of point cloud images to be processed in the plurality of groups of point cloud images to be processed and determining a coincidence point cloud data set corresponding to each group of point cloud images to be processed includes: carrying out adjacent point cloud set analysis on the multiple groups of point cloud images to be processed to obtain adjacent point cloud sets corresponding to each group of point cloud images to be processed; determining a coincidence part based on the adjacent point cloud sets corresponding to each group of point cloud images to be processed, and determining a coincidence part point cloud data set corresponding to each group of point cloud images to be processed.
With reference to the third implementation manner of the first aspect, in a fifth implementation manner of the first aspect, the performing, based on a preset registration function, point cloud registration on the point cloud data sets of the overlapped portions corresponding to each group of point cloud images to be processed respectively to obtain registered point cloud data sets includes: carrying out data grouping on the overlapped part point cloud data sets corresponding to each group of point cloud images to be processed, and determining an adjacent point cloud set corresponding to each group of point cloud images to be processed; respectively carrying out corresponding point matching on adjacent point cloud sets corresponding to each group of point cloud images to be processed to obtain multiple groups of corresponding points corresponding to the adjacent point cloud sets corresponding to each group of point cloud images to be processed and generate corresponding point sets; and performing coordinate conversion on each group of corresponding points in the corresponding point set through the registration function to obtain a registered point cloud data set.
With reference to the first aspect, in a sixth implementation manner of the first aspect, the candidate point cloud image is subjected to outlier analysis, and an outlier set in the candidate point cloud image is determined; registering the outlier set in the candidate point cloud image through a preset correction error function to obtain a corrected outlier set; and generating an image based on the corrected outlier set and the candidate point cloud data to obtain a target point cloud image.
In the scheme, the influence of outliers on the registration effect is achieved by defining the outliers and weakening the weight of the outliers, so that the registration precision and speed are improved, meanwhile, the iterative reweighted least square method is adopted in the scheme of the application, so that the weight of the iterative error of the outliers can be reduced, the registration precision is improved,
according to a second aspect, an embodiment of the present invention provides an image reconstruction apparatus based on a point cloud image, including:
the acquisition module is used for acquiring point cloud images of the head position of a target patient through a preset binocular camera based on a preset image acquisition rule to obtain a plurality of groups of point cloud images;
the registration module is used for carrying out registration processing on the multiple groups of point cloud images through a preset iterative algorithm to obtain candidate point cloud images;
the correction module is used for correcting the accuracy of the candidate point cloud image to obtain a target point cloud image;
the analysis module is used for carrying out target point position analysis on the target point cloud image and determining a plurality of target point positions;
and the transmission module is used for transmitting the positions of the target points to a preset data processing terminal for visual display.
According to a third aspect, an embodiment of the present invention provides an electronic device, including: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing therein computer instructions, and the processor executing the computer instructions to perform the method for reconstructing an image based on a point cloud image according to the aspect or any one of the embodiments of the aspect, or to perform the method for reconstructing an image based on a point cloud image according to the second aspect or any one of the embodiments of the second aspect.
According to a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute an image reconstruction method based on a point cloud image as described in the aspect or any one of the aspects, or execute an image reconstruction method based on a point cloud image as described in the second aspect or any one of the aspects.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a flowchart of an image reconstruction method based on a point cloud image according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating registration of multiple sets of point cloud images according to an embodiment of the present disclosure;
FIG. 3 is a flowchart illustrating an image registration process performed on a plurality of sets of point cloud images to be processed according to an embodiment of the present invention;
FIG. 4 is a flowchart illustrating accuracy correction of a candidate point cloud image according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of an image reconstruction apparatus based on a point cloud image according to an embodiment of the present invention;
fig. 6 is a schematic diagram of a hardware structure of an electronic device according to an embodiment of the present invention.
Reference numerals:
501. an acquisition module; 502. a registration module; 503. a correction module; 504. an analysis module; 505. a transmission module; 601. a processor; 602. a memory.
Detailed Description
The technical solutions of the present invention will be described clearly and completely with reference to the accompanying drawings, and it should be understood that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present invention. Furthermore, the terms "" second, "" third, "and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
In addition, the technical features involved in the different embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
For convenience of understanding, a detailed flow of an embodiment of the present invention is described below, please refer to fig. 1, fig. 1 is a flow chart of an image reconstruction method based on a point cloud image according to an embodiment of the present invention, as shown in fig. 1, the flow chart includes the following steps:
step S101: based on a preset image acquisition rule, carrying out point cloud image acquisition on the head position of a target patient through a preset binocular camera to obtain a plurality of groups of point cloud images;
it is to be understood that the executing subject of the present invention may be an image reconstruction apparatus based on a point cloud image, and may also be a terminal or a server, which is not limited herein. The embodiment of the present invention is described by taking a server as an execution subject.
It should be noted that the image acquisition rule is determined based on the head position of the target patient, and after the head position coordinates of the target patient are determined in a preset coordinate system, specifically, when the target patient sits on a chair, the binocular camera is fixed by the support to shoot obliquely above the head of the patient, the position of the camera is converted to shoot for the second time after a set of data is shot, and after the point cloud image acquisition is performed on the head position of the target patient according to the image acquisition rule, a plurality of sets of point cloud images are obtained.
Step S102: registering a plurality of groups of point cloud images through a preset iterative algorithm to obtain candidate point cloud images;
it should be noted that, after the target patient is photographed at different positions and angles, a co-view and adjacent relationship exists between two sets of collected point cloud data. In order to obtain a complete three-dimensional point cloud image of a target patient, the acquired data needs to be processed, and all the point cloud data are spliced together. The registration algorithm adopted in the scheme is a closest point iterative algorithm, wherein in the registration process of point cloud data, as the acquired point cloud images acquire one group of data at an interval of 10 degrees, each group of point cloud can be ensured to have a better point cloud structure and a better coincidence rate of two adjacent groups of point clouds, and thus the final registration result is improved on the original data. Specifically, the server determines a plurality of groups of near points by analyzing the near points of a plurality of groups of point cloud images, and further, the server respectively solves a rotation matrix and a translation matrix which minimize errors in each group of near points by finding a least square formula of the near point construction errors, and then completes the registration of each group of near points to obtain candidate point cloud images, wherein the least square formula is as follows:
wherein, therein
For corresponding points in the two sets of point cloud data, N p The logarithm of corresponding points in every two groups of point cloud data is obtained, N is the group number of the point clouds, an objective function represents the square sum of Euclidean distances between all corresponding points, R and T are a rotation matrix and a translation matrix respectively, the optimal R and T are solved, so that the objective function obtains the minimum value, and therefore the registration between the two groups of point cloud data is completed.
Step S103: carrying out accuracy correction on the candidate point cloud image to obtain a target point cloud image;
it should be noted that, because R and T obtained during further registration do not necessarily satisfy the previously registered point cloud data, a global ICP algorithm is usually adopted when performing accuracy correction on the point cloud data at present, where an error formula of the global ICP algorithm is;
where K is the number of all point cloud data, N mn The number of the m-th point cloud and the n-th point cloud, R m 、R n The rotation matrices of the mth and nth point clouds, t m And t n The global ICP algorithm can meet the requirement of multi-view registration, but in the process of data acquisition, due to certain external interference, outliers exist, and the global ICP algorithm is not robust to the outliers, so that the outliers are processed to enable global convergence to be optimal. For this reason, we introduce a heteroscedastic weighting matrix to process outliers, so the improved error formula becomes:
wherein, ω is a distance weight function introduced, and the iterative weighted least square method can reduce the weight of the outlier iterative error and improve the registration accuracy, and the value of ω changes according to the error E iterated each time. Two common ω -functions are:
wherein, gamma is a wave-absorbing function, generally, a value of gamma needs to be preset, which is used for defining outliers in data, and the influence of the outliers on the registration effect is achieved by defining the outliers and weakening the weight of the outliers, so as to improve the precision and the speed of the registration.
Step S104: performing target point position analysis on the target point cloud image to determine a plurality of target point positions;
specifically, when the target point position analysis is performed in the scheme of the application, the server determines the coordinates of the stimulation target point according to the brain partition in the head position of the target patient, specifically, the server performs coordinate point matching on the target point cloud image through preset brain partition information, and after the coordinates of the target point are determined, the coordinates of the target point are subjected to coordinate transformation and projected to a preset target point coordinate system to obtain a plurality of target point positions.
Step S105: and transmitting the positions of the target points to a preset data processing terminal for visual display.
By executing the steps, the point cloud image acquisition is carried out on the head position of the target patient through the preset binocular camera based on the preset image acquisition rule, so that a plurality of groups of point cloud images are obtained, and as the acquired point cloud images acquire one group of data at intervals of 10 degrees, each group of point cloud can be ensured to have a better point cloud structure and a better coincidence rate of two adjacent groups of point clouds, so that the final registration result is improved on the original data. Registering the multiple groups of point cloud images through a preset iterative algorithm to obtain candidate point cloud images; and carrying out accuracy correction on the candidate point cloud image to obtain a target point cloud image, carrying out accuracy correction on the candidate point cloud image, namely processing outliers in the candidate point cloud image to enable the overall convergence to be optimal, further improving the accuracy of point cloud data registration, carrying out target point position analysis on the target point cloud image, determining a plurality of target point positions, and transmitting the target point positions to a preset data processing terminal for visual display.
In a specific embodiment, before executing step S101, the following steps may be further specifically included:
(1) Constructing a target coordinate system, carrying out position analysis on the head position of a target patient, and determining head coordinate information of the head position in the target coordinate system;
(2) According to a preset distance, an image acquisition track is constructed through head coordinate information to obtain a target image acquisition track;
(3) Determining image acquisition point positions of a target image acquisition track through a preset angle to obtain a plurality of image acquisition point positions;
(4) And generating an image acquisition rule through the plurality of image acquisition point positions and the target image acquisition track to obtain the image acquisition rule.
Specifically, when a target coordinate system is constructed, a server firstly determines the head position of a target patient, determines a coordinate origin based on the head position, establishes a three-dimensional coordinate system with the head position as the coordinate origin, and simultaneously determines head coordinate information of the head position in the target coordinate system as an origin position coordinate, further, the server determines an acquisition track according to a preset distance, wherein the preset distance can be subjected to parameter setting according to actual conditions, further, when the track is determined, after the distance is determined, the head position coordinate is taken as the origin, the distance is taken as a radius to determine a circular image acquisition track, and simultaneously, image acquisition point location determination is performed on the target image acquisition track according to a preset angle, wherein the acquisition angle of data is required to be shot once per 10 degrees of rotation, the number of the shooting data is 36 groups, the data format is a polygon model data format, so that each two adjacent groups of data have a better overlapping area and the same structure of the target image acquisition track, and meanwhile, the central position of the camera is required to be recorded to be changed every time, a plurality of acquisition angle acquisition rules are obtained as a plurality of point cloud acquisition rules after the acquisition of the target image acquisition tracks are generated.
In a specific embodiment, as shown in fig. 2, the process of executing step S102 may specifically include the following steps:
s201: carrying out image matching on the multiple groups of point cloud images to obtain multiple groups of point cloud images to be processed;
s202: and carrying out image registration processing on a plurality of groups of point cloud images to be processed to obtain a target point cloud image.
Specifically, the plurality of groups of point cloud data are obtained by shooting the head position of the target patient at different positions and angles, and a co-view and adjacent relation exists between two adjacent groups of point cloud data. In order to obtain a complete three-dimensional point cloud image of a patient, acquired data needs to be processed, image matching is carried out on multiple groups of point cloud images to obtain multiple groups of point cloud images to be processed, when image matching is carried out, a server carries out adjacent connection relation analysis on the multiple groups of point cloud images, the point cloud images with the common-view adjacent connection relation are taken as one group to obtain multiple groups of point cloud images to be processed, and further the server carries out image registration processing on each group of point cloud images to be processed respectively to obtain a target point cloud image.
In an embodiment, as shown in fig. 3, the step S202 includes the following steps:
s301: performing coincidence part analysis on each group of point cloud images to be processed in the plurality of groups of point cloud images to be processed, and determining a coincidence part point cloud data set corresponding to each group of point cloud images to be processed;
the method comprises the steps of carrying out brightness analysis on each group of point cloud images to be processed, determining the image brightness value of each point cloud image in each group of point cloud images to be processed, carrying out binarization processing on each group of point cloud images to be processed based on the image brightness value of each point cloud image in each group of point cloud images to be processed to obtain a corresponding binarization image set, carrying out matrix conversion on each binarization image in the binarization image set to obtain a corresponding matrix set, carrying out similarity calculation on the matrix set to obtain a corresponding similarity calculation result, and further determining an overlapped part point cloud data set corresponding to each group of point cloud images to be processed by a server according to the similarity calculation result.
S302: respectively carrying out point cloud registration on the overlapped part point cloud data sets corresponding to each group of point cloud images to be processed based on a preset registration function to obtain a registered point cloud data set;
s303: and carrying out point cloud image splicing through the registered point cloud data set to generate a candidate point cloud image.
Specifically, the registration function is the least squares equation described above, as shown below
Wherein, therein
For corresponding points in the two sets of point cloud data, N p The logarithm of corresponding points in every two groups of point cloud data, N is the group number of point clouds, the target function represents the square sum of Euclidean distances between all corresponding points, R and T are respectively a rotation matrix and a translation matrix, and the optimal R and T are solved to ensure that the target function obtains the minimum value, so that the operation is finishedAnd finally, the server performs point cloud image splicing through the registered point cloud data set to generate a candidate point cloud image.
It should be noted that, in order to obtain a complete head model, further registration is still required to be performed between the two-by-two registered point cloud data, so as to complete further accuracy correction of the two-by-two registered point cloud data.
In an embodiment, the step S301 includes the following steps:
(1) Carrying out adjacent point cloud set analysis on a plurality of groups of point cloud images to be processed to obtain an adjacent point cloud set corresponding to each group of point cloud images to be processed;
(2) And determining the overlapped part based on the adjacent point cloud sets corresponding to each group of point cloud images to be processed, and determining the overlapped part point cloud data sets corresponding to each group of point cloud images to be processed.
Specifically, the server performs rasterization processing on three-dimensional point cloud data in the multiple groups of point cloud images to be processed, divides the three-dimensional point cloud data into three-dimensional point cloud data to be processed in different areas, further performs adjacent position analysis on the three-dimensional point cloud data to be processed in the different areas, and screens the point cloud data meeting a preset distance threshold value to obtain an adjacent point cloud set corresponding to each group of point cloud images to be processed.
The server performs angle classification on a plurality of groups of point cloud images to be processed, determines an angle label of each group of point cloud images to be processed, obtains a plurality of groups of point cloud images to be processed with the angle classification labels, performs rough classification on the coincident parts according to the angle classification labels to obtain rough classification results, performs fine classification on the rough classification results according to the adjacent point cloud sets corresponding to each group of point cloud images to be processed, determines a coincident part point cloud data set corresponding to each group of point cloud images to be processed, solves the problem that the segmentation of the coincident parts in the adjacent point cloud images is inaccurate in the existing classification method, improves the classification accuracy, provides an accurate data base for subsequent point cloud image registration, and further improves the accuracy of point cloud data registration.
In an embodiment, the step S302 specifically includes the following steps:
(1) Carrying out data grouping on the coincident part point cloud data sets corresponding to each group of point cloud images to be processed, and determining an adjacent point cloud set corresponding to each group of point cloud images to be processed;
(2) Respectively carrying out corresponding point matching on the adjacent point cloud sets corresponding to each group of point cloud images to be processed to obtain a plurality of groups of corresponding points corresponding to the adjacent point cloud sets corresponding to each group of point cloud images to be processed and generate corresponding point sets;
(3) And performing coordinate conversion on each group of corresponding points in the corresponding point set through a registration function to obtain a registered point cloud data set.
Specifically, data grouping is carried out on the overlapped part of point cloud data sets corresponding to each group of point cloud images to be processed, and an adjacent point cloud set corresponding to each group of point cloud images to be processed is determined, wherein when data grouping is carried out, a server acquires a preset adjacent point cloud position corresponding table, and a table name coding column is added to the adjacent point cloud position corresponding table; and performing data screening on the overlapped part point cloud data set corresponding to each group of point cloud images to be processed according to preset position screening conditions and an adjacent point cloud position corresponding table, and determining the adjacent point cloud set corresponding to each group of point cloud images to be processed.
Further, the server respectively performs corresponding point matching on adjacent point cloud sets corresponding to each group of point cloud images to be processed to obtain multiple groups of corresponding points corresponding to the adjacent point cloud sets corresponding to each group of point cloud images to be processed and generate corresponding point sets, wherein the corresponding points refer to point groups which are substantially the same point in the adjacent point cloud images, and finally performs coordinate conversion on each group of corresponding points in the corresponding point sets through a registration function to obtain a registered point cloud data set.
In an embodiment, as shown in fig. 4, the step S103 includes the following steps:
s401: performing outlier analysis on the candidate point cloud image to determine an outlier set in the candidate point cloud image;
s402: carrying out registration processing on the outlier set in the candidate point cloud image through a preset correction error function to obtain a corrected outlier set;
s403: and performing image generation based on the corrected outlier set and the candidate point cloud data to obtain a target point cloud image.
It should be noted that, the server performs outlier analysis on the candidate point cloud image, where in the present application, the server obtains outliers in the candidate point cloud image by taking points with fewer neighborhood points as outliers, and further, the server performs registration processing on an outlier set in the candidate point cloud image according to the error correction function to obtain a corrected outlier set, where the error correction function is the above improved error formula, as follows:
wherein, ω is a distance weight function introduced, and the iterative weighted least square method can reduce the weight of the outlier iterative error and improve the registration accuracy, and the value of ω changes according to the error E iterated each time. Two common ω -functions are:
the method comprises the steps that gamma is a wave elimination function, the numerical value of the gamma is generally required to be preset and is used for defining outliers in data, the outliers are defined and the weight of the outliers is weakened to achieve the effect of the outliers on registration, so that the registration precision and the registration speed are improved, meanwhile, the iterative weighted least square method can be used for reducing the weight of iterative errors of the outliers and improving the registration precision, specifically, the outliers in a candidate point cloud image are subjected to registration processing through a preset correction error function to obtain a corrected outlier set, it is required to be noted that the outliers are generally far away from a main body part of the point cloud data and few neighborhood points exist around the outlier set, and finally, a server performs image generation based on the corrected outlier set and the candidate point cloud data to obtain a target point cloud image.
Optionally, because the acquired data volume is large, a situation of low speed is easily caused when only a global ICP algorithm is used for image registration, so that an Anderson acceleration algorithm can be introduced to improve the registration speed on the basis of the improved global ICP algorithm, the Anderson acceleration algorithm is different from a standard iteration method in the ICP algorithm, the result of the last iteration is not only dependent on the result of the last iteration, but is a linear combination of all iteration results, so that the convergence speed of the method is higher, meanwhile, the influence of outliers is reduced, and the instability in the Anderson acceleration process is solved to a certain extent, which is the process of the improvement of the global ICP algorithm and the registration of multi-view point cloud images.
An embodiment of the present invention further provides an image reconstruction apparatus based on a point cloud image, as shown in fig. 5, the image reconstruction apparatus based on a point cloud image specifically includes:
the acquisition module 501 is used for acquiring a point cloud image of the head position of a target patient through a preset binocular camera based on a preset image acquisition rule to obtain a plurality of groups of point cloud images;
a registration module 502, configured to perform registration processing on multiple groups of point cloud images through a preset iterative algorithm to obtain candidate point cloud images;
a correction module 503, configured to perform accuracy correction on the candidate point cloud image to obtain a target point cloud image;
an analysis module 504, configured to perform target position analysis on the target point cloud image, and determine a plurality of target positions;
and the transmission module 505 is configured to transmit the positions of the multiple target points to a preset data processing terminal for visual display.
Further functional descriptions of the above modules are the same as those of the corresponding method embodiments, and are not repeated herein.
Through the cooperative cooperation of the components, on the basis of a preset image acquisition rule, point cloud image acquisition is carried out on the head position of a target patient through a preset binocular camera to obtain a plurality of groups of point cloud images, and as the acquired point cloud images acquire a group of data at intervals of 10 degrees, each group of point cloud can be ensured to have a better point cloud structure and a better coincidence rate of two adjacent groups of point clouds, so that the final registration result is improved on the original data. Registering the multiple groups of point cloud images through a preset iterative algorithm to obtain candidate point cloud images; carrying out accuracy correction on the candidate point cloud image to obtain a target point cloud image, carrying out accuracy correction on the candidate point cloud image, namely processing outliers in the candidate point cloud image to enable the overall convergence to be optimal, further improving the accuracy of point cloud data registration, carrying out target position analysis on the target point cloud image, and determining a plurality of target positions; and transmitting the positions of the target points to a preset data processing terminal for visual display.
An embodiment of the present invention further provides an electronic device, as shown in fig. 6, the electronic device may include a processor 601 and a memory 602, where the processor 601 and the memory 602 may be connected by a bus or in another manner, and fig. 6 illustrates an example of a connection by a bus.
The memory 602 may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created by the processor 601, and the like. Further, the memory 602 may include high speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory 602 may optionally include memory located remotely from the processor 601, which may be connected to the processor 601 via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
One or more modules are stored in the memory 602 and, when executed by the processor 601, perform the above-described methods.
Embodiments of the present invention further provide a non-transitory computer storage medium, where computer-executable instructions are stored, and the computer-executable instructions may execute the people counting method in any of the method embodiments. The storage medium may be a magnetic Disk, an optical Disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a Flash Memory (Flash Memory), a Hard Disk (Hard Disk Drive, abbreviated as HDD), a Solid State Drive (SSD), or the like; the storage medium may also comprise a combination of memories of the kind described above.
Those skilled in the art can understand that all or part of the processes in the method of the above embodiments may be implemented by instructing related hardware through a computer program, and the program may be stored in a computer readable storage medium, where the storage medium may be a magnetic Disk, an optical Disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a Flash Memory (Flash Memory), a Hard Disk (Hard Disk Drive, abbreviated as HDD), or a Solid-State Drive (SSD); the storage medium may also comprise a combination of memories of the kind described above.
The above embodiments are only for illustrating the technical solutions of the present invention and not for limiting the same, and although the present invention is described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications and equivalents may be made to the embodiments of the invention without departing from the spirit and scope of the invention, which is to be covered by the claims.
Claims (10)
1. An image reconstruction method based on a point cloud image is characterized by comprising the following steps:
based on a preset image acquisition rule, carrying out point cloud image acquisition on the head position of a target patient through a preset binocular camera to obtain a plurality of groups of point cloud images;
registering the multiple groups of point cloud images through a preset iterative algorithm to obtain candidate point cloud images;
carrying out accuracy correction on the candidate point cloud image to obtain a target point cloud image;
performing target point position analysis on the target point cloud image to determine a plurality of target point positions;
and transmitting the positions of the target points to a preset data processing terminal for visual display.
2. The method for reconstructing an image based on a point cloud image according to claim 1, wherein before the step of acquiring a point cloud image of the head of a target patient by using a preset binocular camera based on a preset image acquisition rule to obtain a plurality of groups of point cloud images, the method further comprises:
constructing a target coordinate system, carrying out position analysis on the head position of a target patient, and determining head coordinate information of the head position in the target coordinate system;
constructing an image acquisition track through the head coordinate information according to a preset distance to obtain a target image acquisition track;
determining image acquisition point positions of the target image acquisition track through a preset angle to obtain a plurality of image acquisition point positions;
and generating an image acquisition rule through the plurality of image acquisition point positions and the target image acquisition track to obtain the image acquisition rule.
3. The method for reconstructing an image based on a point cloud image according to claim 1, wherein the step of registering the plurality of groups of point cloud images by a preset iterative algorithm to obtain a target point cloud image comprises:
carrying out image matching on the plurality of groups of point cloud images to obtain a plurality of groups of point cloud images to be processed;
and carrying out image registration processing on the multiple groups of point cloud images to be processed to obtain target point cloud images.
4. The method of claim 3, wherein the step of performing image registration on the plurality of groups of point cloud images to be processed to obtain candidate point cloud images comprises:
performing overlapping part analysis on each group of point cloud images to be processed in the plurality of groups of point cloud images to be processed, and determining overlapping part point cloud data sets corresponding to each group of point cloud images to be processed;
respectively carrying out point cloud registration on the point cloud data sets of the overlapped parts corresponding to each group of point cloud images to be processed based on a preset registration function to obtain a registered point cloud data set;
and carrying out point cloud image splicing through the registered point cloud data set to generate a candidate point cloud image.
5. The point cloud image-based image reconstruction method according to claim 4, wherein the step of performing a coincidence analysis on each group of point cloud images to be processed in the plurality of groups of point cloud images to be processed to determine a coincidence point cloud data set corresponding to each group of point cloud images to be processed comprises:
carrying out adjacent point cloud set analysis on the plurality of groups of point cloud images to be processed to obtain adjacent point cloud sets corresponding to each group of point cloud images to be processed;
determining a coincidence part based on the adjacent point cloud sets corresponding to each group of point cloud images to be processed, and determining a coincidence part point cloud data set corresponding to each group of point cloud images to be processed.
6. The point cloud image-based image reconstruction method according to claim 4, wherein the step of performing point cloud registration on the point cloud data sets of the overlapped parts corresponding to each group of point cloud images to be processed respectively based on a preset registration function to obtain the registered point cloud data sets comprises:
carrying out data grouping on the overlapped part point cloud data sets corresponding to each group of point cloud images to be processed, and determining an adjacent point cloud set corresponding to each group of point cloud images to be processed;
respectively carrying out corresponding point matching on adjacent point cloud sets corresponding to each group of point cloud images to be processed to obtain multiple groups of corresponding points corresponding to the adjacent point cloud sets corresponding to each group of point cloud images to be processed and generate corresponding point sets;
and performing coordinate conversion on each group of corresponding points in the corresponding point set through the registration function to obtain a registered point cloud data set.
7. The method of claim 1, wherein the step of performing accuracy correction on the candidate point cloud image to obtain a target point cloud image comprises:
performing outlier analysis on the candidate point cloud image, and determining an outlier set in the candidate point cloud image;
registering the outlier set in the candidate point cloud image through a preset correction error function to obtain a corrected outlier set;
and performing image generation based on the corrected outlier set and the candidate point cloud data to obtain a target point cloud image.
8. An image reconstruction apparatus based on a point cloud image for performing the image reconstruction method based on a point cloud image according to any one of claims 1 to 7, comprising:
the acquisition module is used for acquiring point cloud images of the head position of a target patient through a preset binocular camera based on a preset image acquisition rule to obtain a plurality of groups of point cloud images;
the registration module is used for carrying out registration processing on the plurality of groups of point cloud images through a preset iterative algorithm to obtain candidate point cloud images;
the correction module is used for correcting the accuracy of the candidate point cloud image to obtain a target point cloud image;
the analysis module is used for carrying out target point position analysis on the target point cloud image and determining a plurality of target point positions;
and the transmission module is used for transmitting the positions of the target points to a preset data processing terminal for visual display.
9. An electronic device, comprising:
a memory and a processor, wherein the memory and the processor are connected with each other in communication, the memory stores computer instructions, and the processor executes the computer instructions to execute the image reconstruction method based on the point cloud image according to any one of claims 1 to 7.
10. A computer-readable storage medium storing computer instructions for causing a computer to execute the method for image reconstruction based on a point cloud image according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211633652.XA CN115797416A (en) | 2022-12-19 | 2022-12-19 | Image reconstruction method, device and equipment based on point cloud image and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211633652.XA CN115797416A (en) | 2022-12-19 | 2022-12-19 | Image reconstruction method, device and equipment based on point cloud image and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115797416A true CN115797416A (en) | 2023-03-14 |
Family
ID=85425815
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211633652.XA Pending CN115797416A (en) | 2022-12-19 | 2022-12-19 | Image reconstruction method, device and equipment based on point cloud image and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115797416A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116563297A (en) * | 2023-07-12 | 2023-08-08 | 中国科学院自动化研究所 | Craniocerebral target positioning method, device and storage medium |
-
2022
- 2022-12-19 CN CN202211633652.XA patent/CN115797416A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116563297A (en) * | 2023-07-12 | 2023-08-08 | 中国科学院自动化研究所 | Craniocerebral target positioning method, device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11657509B2 (en) | Method for precisely and automatically positioning reference line for integrated images | |
| CN107578376B (en) | Image splicing method based on feature point clustering four-way division and local transformation matrix | |
| CN111291825B (en) | Focus classification model training method, apparatus, computer device and storage medium | |
| KR102450931B1 (en) | Image registration method and associated model training method, apparatus, apparatus | |
| CN112328715B (en) | Visual positioning method, training method of related model, related device and equipment | |
| CN109741395B (en) | Dual-chamber quantification method and device, electronic equipment and storage medium | |
| CN111488872B (en) | Image detection method, image detection device, computer equipment and storage medium | |
| JP2021533505A (en) | Deep model training methods and their equipment, electronics and storage media | |
| CN110796135A (en) | Target positioning method and device, computer equipment and computer storage medium | |
| CN115797416A (en) | Image reconstruction method, device and equipment based on point cloud image and storage medium | |
| CN114758337A (en) | Semantic instance reconstruction method, device, equipment and medium | |
| CN114219855A (en) | Point cloud normal vector estimation method and device, computer equipment and storage medium | |
| CN113034684B (en) | Three-dimensional reconstruction method, electronic device, and computer-readable storage medium | |
| CN111986242B (en) | Method and device for determining brain tissue partition, storage medium and electronic equipment | |
| CN112215955A (en) | Rigid body mark point screening method, device, system, equipment and storage medium | |
| CN115620264B (en) | Vehicle positioning method and device, electronic equipment and computer readable medium | |
| CN112581460B (en) | Scanning planning method, device, computer equipment and storage medium | |
| CN115457093A (en) | Tooth image processing method and device, electronic equipment and storage medium | |
| CN113643328A (en) | Calibration object reconstruction method and device, electronic equipment and computer readable medium | |
| CN111080734B (en) | Method and terminal for processing Positron Emission Tomography (PET) data | |
| CN108875520B (en) | Method, device and system for positioning face shape point and computer storage medium | |
| CN111784705A (en) | Method, device and equipment for delineating region of interest and storage medium | |
| CN114288023B (en) | Calibration method, device, equipment and medium for positioning guide plate in operation navigation equipment | |
| CN112950582B (en) | 3D lung focus segmentation method and device based on deep learning | |
| CN117437288B (en) | Photogrammetry method, device, equipment 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 |