CN110245671A - A kind of endoscopic images characteristic point matching method and system - Google Patents

A kind of endoscopic images characteristic point matching method and system Download PDF

Info

Publication number
CN110245671A
CN110245671A CN201910521138.9A CN201910521138A CN110245671A CN 110245671 A CN110245671 A CN 110245671A CN 201910521138 A CN201910521138 A CN 201910521138A CN 110245671 A CN110245671 A CN 110245671A
Authority
CN
China
Prior art keywords
characteristic point
matching
point
endoscopic images
image
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
Application number
CN201910521138.9A
Other languages
Chinese (zh)
Inventor
杨峰
李文杰
李旭
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ari Mai Di Technology Shijiazhuang Co Ltd
Original Assignee
Ari Mai Di Technology Shijiazhuang Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Ari Mai Di Technology Shijiazhuang Co Ltd filed Critical Ari Mai Di Technology Shijiazhuang Co Ltd
Priority to CN201910521138.9A priority Critical patent/CN110245671A/en
Publication of CN110245671A publication Critical patent/CN110245671A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/36Image preprocessing, i.e. processing the image information without deciding about the identity of the image
    • G06K9/46Extraction of features or characteristics of the image
    • G06K9/4604Detecting partial patterns, e.g. edges or contours, or configurations, e.g. loops, corners, strokes, intersections
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KRECOGNITION OF DATA; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for reading or recognising printed or written characters or for recognising patterns, e.g. fingerprints
    • G06K9/62Methods or arrangements for recognition using electronic means
    • G06K9/6201Matching; Proximity measures
    • G06K9/6202Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
    • G06K9/6203Shifting or otherwise transforming the patterns to accommodate for positional errors
    • G06K9/6211Matching configurations of points or features, e.g. constellation matching
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00Image enhancement or restoration
    • G06T5/001Image restoration
    • G06T5/002Denoising; Smoothing
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/13Edge detection
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/136Segmentation; Edge detection involving thresholding
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/20Analysis of motion

Abstract

The embodiment of the present invention provides a kind of endoscopic images characteristic point matching method and system.This method includes extracting image characteristic point based on two endoscopic images to be matched, and complete Image Feature Point Matching based on Feature Descriptor similarity, obtains initial matching point to set;Local distance constraint is carried out to set to initial matching point, characteristic point information is extended in conjunction with affine parameter and character pair point motion information, similarity boundary is estimated, obtains the characteristic point correspondence set with Movement consistency;Space length perception is optimized based on characteristic point correspondence set, bilateral affine motion consistency model is generated, the adaptive distance threshold parameters of bilateral moving boundaries is set, the corresponding interior point matching set of global image is obtained, realizes Feature Points Matching.The embodiment of the present invention finds reliable corresponding relationship, and guaranteeing high-precision while retaining enough Feature Points Matchings pair by the Movement consistency method kept based on locality from given initial matching pair.

Description

A kind of endoscopic images characteristic point matching method and system
Technical field
The present invention relates to field of image processing more particularly to a kind of endoscopic images characteristic point matching method and systems.
Background technique
Endoscope is a kind of optical instrument, is shown by cold light source camera lens, fiber optic conducting wire, image delivering system, screen The composition such as system, it can expand surgical field of view, and the outstanding feature using endoscope is that operative incision is small, and incisional scar is unobvious, After-operation response is light, and bleeding, livid purple and swelling time can greatly reduce, and restores also fast compared with traditional operation.Passing through area of computer aided Medical endoscopes field of surgery, it can be common that carry out associated picture processing to guide Minimally Invasive Surgery, build between images Founding reliable corresponding relationship is the critical issue in numerous clinical applications, for example, tissue surface rebuild, camera motion estimation with And in the application such as surgical navigational.
Existing endoscopic images characteristic point matching method is generally special using common detection and description image local first Algorithm is levied, such as utilizes affine scale invariant feature conversion method (Affine Scale Invariant Feature Transform, hereinafter referred to as ASIFT) etc. extract the characteristic point of image, and using corresponding Feature Descriptor come to characteristic point Match the feature point correspondence between available image, but due to the general texture of institutional framework in endoscopic images Information is weak, exists and such as blocks and the problems such as deformation, simple dependence characteristics describe sub- similarity be difficult to obtain it is satisfactory Matching result.
Existing endoscopic images characteristic point matching method is primarily present following problem: extracting characteristic point in the case of weak texture Number deficiency, the matching result lazy weight finally retained are difficult to realize careful surface or structural remodeling;Change threshold value to mention When taking fully measure feature point, since texture information is weak, there are factors such as deformation, the matching number of mistake is more, a large amount of outlier without Method is effectively rejected, and matching precision is caused to decline.
For clinical medicine image procossing, obtains sufficient amount and high-precision matching result is that image guidance is minimally invasive The key of successful surgery, it is therefore desirable to a kind of method for proposing high-precision endoscopic images Feature Points Matching.
Summary of the invention
The embodiment of the present invention provides a kind of endoscopic images characteristic point matching method and system, to solve in the prior art Image characteristic point extracts lazy weight, and matching precision is low, the problems such as influence by a large amount of interference noise points.
In a first aspect, the embodiment of the present invention provides a kind of endoscopic images characteristic point matching method, comprising:
S1 extracts image characteristic point based on two endoscopic images to be matched, and complete based on Feature Descriptor similarity At described image Feature Points Matching, the initial matching point of described image characteristic point is obtained to set;
S2, based on the unknown deformation and described image characteristic point local neighborhood between two endoscopic images to be matched Corresponding relationship and partial structurtes between structure carry out local distance constraint to set to the initial matching point, in conjunction with affine Parameter and character pair point motion information are extended characteristic point information, estimate similarity boundary, obtain having movement consistent The characteristic point correspondence set of property;
S3 optimizes space length perception based on the characteristic point correspondence set, generates bilateral affine motion The adaptive distance threshold parameters of bilateral moving boundaries are arranged in consistency model, obtain the corresponding interior set of matches of global image It closes, realizes the Feature Points Matching between two endoscopic images to be matched.
Wherein, the step of S2 is specifically included:
S21 establishes the unknown deformation between two endoscopic images to be matched and described image characteristic point part neighbour Corresponding relationship between domain structure;
S22 is based on each described image characteristic point, and corresponding 6 closest approaches construct the office in corresponding point set Portion's structure;
S23, based between two endoscopic images to be matched unknown deformation and described image characteristic point part it is adjacent Corresponding relationship between domain structure, and any the distance between the initial matching point pair and its neighborhood point are fixed, and threshold value is arranged Parameter calculates and obtains the interior point set kept with locality, realizes local distance constraint;
S24, the characteristic point information extracted to the two images to be matched are extended, and addition characteristic point is corresponding Motion information obtain the characteristic point correspondence set with the Movement consistency using similarity boundary function.
Wherein, the step S22 includes being calculated based on Euclidean distance formula, is obtained based on each described image Characteristic point and its in corresponding point set corresponding 6 closest approaches come the partial structurtes that construct.
Wherein, similarity boundary function is applied in the step S24, obtains the characteristic point pair with Movement consistency Set of relationship is answered, is based on obtained by the interior point set kept with locality.
Wherein, the step of S3 is specifically included:
S31 is based on the characteristic point correspondence set, using the affine motion boundary of bilateral variation, obtains described double Side affine motion consistency model;
The adaptive space threshold parameter of the bilateral moving boundaries is arranged in S32, and it is corresponding interior to obtain the global image Point matching set, realizes the Feature Points Matching between two endoscopic images to be matched.
Wherein, the adaptive space threshold parameter of the bilateral moving boundaries is set in the step S32, is specifically included:
Model is kept in conjunction with partial structurtes, sets the adaptive space threshold parameter of the bilateral moving boundaries to and institute The distance restraint threshold parameter for stating partial structurtes is consistent.
Wherein, the corresponding interior point matching set of the global image is calculated and obtained in the step S32, is specifically included:
It is special by the way that estimation result and described image of the described image characteristic point in bilateral moving boundaries both direction is arranged The distance between sign spot noise observation data threshold value is gathered to obtain the corresponding interior point matching of the global image.
Second aspect, the embodiment of the present invention provide a kind of endoscopic images Feature Points Matching system, comprising:
First processing module for extracting image characteristic point based on two endoscopic images to be matched, and is based on feature It describes sub- similarity and completes described image Feature Points Matching, obtain the initial matching point of described image characteristic point to set;
Second processing module, for based on the unknown deformation and described image between two endoscopic images to be matched Corresponding relationship and partial structurtes between characteristic point local neighborhood structure to the initial matching point to set carry out part away from From constraint, characteristic point information is extended in conjunction with affine parameter and character pair point motion information, similarity boundary is estimated, obtains To the characteristic point correspondence set with Movement consistency;
Third processing module, it is raw for being optimized based on the characteristic point correspondence set to space length perception At bilateral affine motion consistency model, the adaptive distance threshold parameters of bilateral moving boundaries are set, global image pair is obtained The interior point matching set answered, realizes the Feature Points Matching between two endoscopic images to be matched.
The third aspect, the embodiment of the present invention provide a kind of electronic equipment, comprising:
Memory, processor and storage on a memory and the computer program that can run on a processor, the processing Device realizes a kind of the step of any one endoscopic images characteristic point matching method when executing described program.
Fourth aspect, the embodiment of the present invention provide a kind of non-transient computer readable storage medium, are stored thereon with calculating Machine program any one of realizes a kind of endoscopic images characteristic point matching method when computer program is executed by processor Step.
A kind of endoscopic images characteristic point matching method provided in an embodiment of the present invention and system, by being protected based on locality The Movement consistency method held, finds reliable corresponding relationship from the given initial matching pair comprising a large amount of outliers, and Guarantee high-precision while retaining enough Feature Points Matchings pair, helps to realize finer tissue surface and rebuild, is more quasi- The application such as true camera motion estimation and surgical navigational.
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, the accompanying drawings in the following description is the present invention Some embodiments for those of ordinary skill in the art without creative efforts, can also basis These attached drawings obtain other attached drawings.
Fig. 1 is a kind of endoscopic images characteristic point matching method flow chart provided in an embodiment of the present invention;
Fig. 2 is step S2 detailed process in a kind of endoscopic images characteristic point matching method provided in an embodiment of the present invention Figure;
Fig. 3 is step S3 detailed process in a kind of endoscopic images characteristic point matching method provided in an embodiment of the present invention Figure;
Fig. 4 is a kind of endoscopic images Feature Points Matching system structure diagram provided in an embodiment of the present invention;
Fig. 5 is a kind of Second processing module sub- knot of endoscopic images Feature Points Matching system provided in an embodiment of the present invention Structure schematic diagram;
Fig. 6 is a kind of third processing module sub- knot of endoscopic images Feature Points Matching system provided in an embodiment of the present invention Structure schematic diagram;
Fig. 7 is matching algorithm flow chart provided in an embodiment of the present invention;
Fig. 8 is the Movement consistency bounding algorithm flow chart provided in an embodiment of the present invention kept based on partial structurtes;
Fig. 9 is provided in an embodiment of the present invention based on the bilateral moving boundaries bounding algorithm flow chart of adaptive threshold;
Figure 10 is the structural block diagram of electronic equipment provided in an embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art Every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
There are the following problems for existing endoscopic images characteristic point matching method: characteristic point number is extracted in the case of weak texture Deficiency, the matching result lazy weight finally retained are difficult to realize careful surface or structural remodeling;Change threshold value and extracts foot When measure feature point, since texture information is weak, there are factors such as deformation, the matching number of mistake is more, and a large amount of outlier can not be by It effectively rejects, matching precision is caused to decline.
Aiming at the problems existing in the prior art, the embodiment of the invention provides a kind of endoscopic images Feature Points Matching sides Method, Fig. 1 are a kind of endoscopic images characteristic point matching method flow chart provided in an embodiment of the present invention, as shown in Figure 1, comprising:
S1 extracts image characteristic point based on two endoscopic images to be matched, and complete based on Feature Descriptor similarity At described image Feature Points Matching, the initial matching point of described image characteristic point is obtained to set;
S2, based on the unknown deformation and described image characteristic point local neighborhood between two endoscopic images to be matched Corresponding relationship and partial structurtes between structure carry out local distance constraint to set to the initial matching point, in conjunction with affine Parameter and character pair point motion information are extended characteristic point information, estimate similarity boundary, obtain having movement consistent The characteristic point correspondence set of property;
S3 optimizes space length perception based on the characteristic point correspondence set, generates bilateral affine motion The adaptive distance threshold parameters of bilateral moving boundaries are arranged in consistency model, obtain the corresponding interior set of matches of global image It closes, realizes the Feature Points Matching between two endoscopic images to be matched.
Specifically, two endoscopic images characteristic points to be matched are extracted first in step sl, it is assumed that in one pair given Sight glass image is respectively Ir(reference picture) and It(template image), using image characteristics extraction algorithm, the present embodiment is used ASIFT algorithm, the feature extracted in two images respectively obtain character pair point set FrAnd Ft.Due to group in endoscopic images It knits that smooth, texture information is weak, therefore to consider all possible matching result, obtain one using the similarity of Feature Descriptor Initial feature point correspondence, is defined asPurpose is in order to by rejecting SoIn outlier To obtain reliable corresponding relationship.Unknown deformation between two endoscopic images to be matched and described is initially set up in step S2 Corresponding relationship between image characteristic point local neighborhood structure, and based on above-mentioned corresponding relationship and partial structurtes to being obtained in step S1 The initial matching point taken carries out local distance constraint to set, then introduces affine parameter and the movement of character pair point to characteristic point Information is extended, then estimates similarity boundary to realize the consistency constraint to characteristic point movement.Step S3 is in step s 2 Space length perception is optimized on the basis of all set of characteristic points for meeting similarity boundary generated, is obtained bilateral imitative Movement consistency model is penetrated, while using the average distance of partial structurtes as the adaptive apart from threshold of bilateral moving boundaries model Value, carries out judgement calculating, it is therefore an objective to reject the noise outlier in image, and obtain final Feature Points Matching result.
Whole matching algorithm process is as shown in fig. 7, Fig. 7 is matching algorithm flow chart provided in an embodiment of the present invention.
The embodiment of the present invention is by the Movement consistency method that is kept based on locality, from given comprising a large amount of outliers Reliable corresponding relationship is found in initial matching pair, and is being guaranteed high-precision while being retained enough Feature Points Matchings pair, Help to realize the applications such as finer tissue surface reconstruction, the estimation of more accurate camera motion and surgical navigational.
On that basi of the above embodiments, the step S2 specific steps refer to Fig. 2, and Fig. 2 is provided in an embodiment of the present invention Step S2 specific flow chart in a kind of endoscopic images characteristic point matching method, as shown in Figure 2, comprising:
S21 establishes the unknown deformation between two endoscopic images to be matched and described image characteristic point part neighbour Corresponding relationship between domain structure;
S22 is based on each described image characteristic point, and corresponding 6 closest approaches construct the office in corresponding point set Portion's structure;
S23, based between two endoscopic images to be matched unknown deformation and described image characteristic point part it is adjacent Corresponding relationship between domain structure, and any the distance between the initial matching point pair and its neighborhood point are fixed, and threshold value is arranged Parameter calculates and obtains the interior point set kept with locality, realizes local distance constraint;
S24, the characteristic point information extracted to the two images to be matched are extended, and addition characteristic point is corresponding Motion information obtain the characteristic point correspondence set with the Movement consistency using similarity boundary function.
Wherein, the step S22 includes being calculated based on Euclidean distance formula, is obtained based on each described image Characteristic point and its in corresponding point set corresponding 6 closest approaches come the partial structurtes that construct.
Wherein, similarity boundary function is applied in the step S24, obtains the characteristic point pair with Movement consistency Set of relationship is answered, is based on obtained by the interior point set kept with locality.
Specifically, step S21 first assumes that unknown deformation has occurred between the image of a pair of of endoscope, and character pair point Local neighborhood structure can't freely change, can only constrain local neighborhood structure here, and find two width and wait for The corresponding relationship between unknown deformation and two image characteristic point local neighborhood structures to be matched between matched image, we It is the optimal solution t for finding locality loss function C by the task presentation for establishing above-mentioned reliable corresponding relationship, with following formula table Show:
Wherein t is the binary set of N × 1, works as tn=1 and tnCorresponding points (x is respectively indicated when=0n,yn) be interior point or Outlier, defined parameters ξ is used to the weight of summation first item and Section 2 in balanced double-rope C, and ξ > 0, parameter N are indicated just The number of beginning matching double points, parameter cnLocal distance structure is calculated by following formula:
Wherein d is the distance matrix for defining two-value, NxAnd NyRespectively indicate the neighborhood point set of characteristic point x and y.
Step S22, the calculation formula based on Euclidean distance, by finding each characteristic point phase in corresponding point set 6 closest approaches answering construct partial structurtes, this is defined herein as k, whenWhen d (xn, xm)=0, whenWhen d (xn, xm)=1, d (yn, ym) definition similarly.
It is due to characteristics of image point correspondence initial in the step s 21 it has been determined that and any first in step 23 Beginning matching double points and with the distance between its neighborhood point be it is fixed, judged by the way that threshold parameter ζ is arranged, when meeting cn When≤ξ, tn=1, to solve the optimal solution t of locality loss function C, so that the interior point set that there is locality to keep is obtained, WithIndicate the interior point set, as follows:
Next in step s 24, by addition motion information and radiology information, by the description information of image characteristic point from Only location information xn=[xn, yn] expand to xn=[xn, yn;vn;on], wherein n is index a little, vn=[un, vn] indicate The movement of character pair point in x and y direction, on=[scale, orientation, tilt, rotation] is ASIFT algorithm The affine information of middle Feature Descriptor.In the set kept with localityOn the basis of, using a similarity boundary functionIt realizes Movement consistency, obtains following formula:
Wherein H () indicates Huber function, for punishing between the function prediction value of estimation and the observation " 1 " of hypothesis Deviation.G (i, j) is oneSymmetrical gram square matrix, wherein γ be Standard deviation, λ are the weight of smooth item.It is the Gaussian kernel weight vectors of M dimension,Indicate that M characterization cluster centre exists Characteristic point { xjWhere space distribution.Therefore, the minimum value in formula is estimated that optimized parameterAs a result, we can be with Passing through willIt brings intoExpression formula in calculate similarity boundaryValue, then by verifying all initial matchings As a result SoWhether meet the threshold condition under similarity boundary, is set as εlikelihood, obtain the corresponding relationship with Movement consistency Set
Step 21 is mentioned to the complete algorithm flow chart of step 24 referring to Fig. 8, Fig. 8 for the embodiment of the present invention in above-described embodiment The Movement consistency bounding algorithm flow chart kept based on partial structurtes supplied.
The embodiment of the present invention carries out local distance constraint to set by the initial matching point to acquisition, realizes based on part Structure-preserved Movement consistency constraint, can be from the endoscopic images feature point correspondence there are much noise, accurately Erroneous matching is rejected on ground, while considering the consistency of invariance and global motion of the partial structurtes in deformation, can be with robust Ground solves the matching problem under different types of deformation.
On that basi of the above embodiments, the step S3 specific steps refer to Fig. 3, and Fig. 3 is provided in an embodiment of the present invention Step S3 specific flow chart in a kind of endoscopic images characteristic point matching method, as shown in Figure 3, comprising:
S31 is based on the characteristic point correspondence set, using the affine motion boundary of bilateral variation, obtains described double Side affine motion consistency model;
The adaptive space threshold parameter of the bilateral moving boundaries is arranged in S32, and it is corresponding interior to obtain the global image Point matching set, realizes the Feature Points Matching between two endoscopic images to be matched.
Wherein, the adaptive space threshold parameter of the bilateral moving boundaries is set in the step S32, is specifically included:
Model is kept in conjunction with partial structurtes, sets the adaptive space threshold parameter of the bilateral moving boundaries to and institute The distance restraint threshold parameter for stating partial structurtes is consistent.
Wherein, the corresponding interior point matching set of the global image is calculated and obtained in the step S32, is specifically included:
It is special by the way that estimation result and described image of the described image characteristic point in bilateral moving boundaries both direction is arranged The distance between sign spot noise observation data threshold value is gathered to obtain the corresponding interior point matching of the global image.Specifically, it is Model is set to include subtle spatial perception ability, in step S31, in the correspondence set with Movement consistencyThe affine motion boundary of a bilateral variation is cascaded, afterwards to obtain more Accurate world model, above-mentioned set SlbIt is to obtain in step 24, is expressed as in x and y directionWithIt can estimate to obtain by following optimization problem:
Wherein OkIt is the deviation of scalar, k is the index of different smooth functions.By the loss for minimizing loss function Value, is calculated optimal solutionWithEach smooth functionIt can be by the way that optimal solution be brought intoExpression formula In be calculated, while also just obtainingWith
In step 32, in order to make local distance constraint that can play a role in entirely matching process, bilateral moving boundaries Word space threshold be also required to be consistent therewith.Therefore, keep model in conjunction with partial structurtes, be arranged one it is following adaptive Capacity-threshold is answered to enable global motion that locality is preferably combined to keep strategy, as follows:
The distance between data threshold value d is observed with noise by the estimation result in setting both directionlp, we can obtain There is high-precision interior point set .S to followingmc:
Finally realize the Feature Points Matching between two endoscopic images to be matched.
Step 31 is mentioned to the complete algorithm flow chart of step 32 referring to Fig. 9, Fig. 9 for the embodiment of the present invention in above-described embodiment Supply based on the bilateral moving boundaries bounding algorithm flow chart of adaptive threshold.
The embodiment of the present invention is all had adaptively by the application of adaptive threshold in the Deformation Types in face of different scale Constraint, be effectively adapted to the endoscopic images of unknown deformation size, can retain in the case where guaranteeing high-precision situation enough More proper characteristics point correspondence, convenient for being applied in the operation such as subsequent structural remodeling.
Fig. 4 is a kind of endoscopic images Feature Points Matching system structure diagram provided in an embodiment of the present invention, such as Fig. 4 institute Show, comprising: first processing module 41, Second processing module 42 and third processing module 43, in which: first processing module 41 is used for Image characteristic point is extracted based on two endoscopic images to be matched, and described image spy is completed based on Feature Descriptor similarity Sign point matching, obtains the initial matching point of described image characteristic point to set;Second processing module 42 is used to be based on two width Corresponding relationship between unknown deformation between endoscopic images to be matched and described image characteristic point local neighborhood structure and Partial structurtes carry out local distance constraint to set to the initial matching point, in conjunction with affine parameter and character pair point movement letter Breath is extended characteristic point information, estimates similarity boundary, obtains the characteristic point correspondence set with Movement consistency; Third processing module 43 is used to optimize space length perception based on the characteristic point correspondence set, generates bilateral imitative Movement consistency model is penetrated, the adaptive distance threshold parameters of bilateral moving boundaries are set, obtains the corresponding interior point of global image Matching set, realizes the Feature Points Matching between two endoscopic images to be matched.
System provided in an embodiment of the present invention for executing above-mentioned corresponding method, specific embodiment and method Embodiment is consistent, and the algorithm flow being related to is identical as corresponding method some algorithm process, and details are not described herein again.
The embodiment of the present invention carries out local distance constraint to set by the initial matching point to acquisition, realizes based on part Structure-preserved Movement consistency constraint, can be from the endoscopic images feature point correspondence there are much noise, accurately Erroneous matching is rejected on ground, while considering the consistency of invariance and global motion of the partial structurtes in deformation, can be with robust Ground solves the matching problem under different types of deformation.
On that basi of the above embodiments, Fig. 5 is a kind of endoscopic images Feature Points Matching system provided in an embodiment of the present invention The Second processing module minor structure schematic diagram of system, as shown in figure 5, Second processing module 42 specifically includes: matched sub-block 421, Construct submodule 422, the first computational submodule 423 and the second computational submodule 424, in which:
Matched sub-block 421 is used for the unknown deformation and described image established between two endoscopic images to be matched Corresponding relationship between characteristic point local neighborhood structure;It constructs submodule 422 to be used to be based on each described image characteristic point, right Corresponding 6 closest approaches are answered in point set to construct the partial structurtes;First computational submodule 423 is used to be based on two width Corresponding relationship between unknown deformation between endoscopic images to be matched and described image characteristic point local neighborhood structure, and appoint Anticipating, the distance between the initial matching point pair and its neighborhood point are fixed, and setting threshold parameter is calculated and obtained described with office The interior point set that portion's property is kept realizes local distance constraint;Second computational submodule 424 is used for the two width figures to be matched The characteristic point information that picture extracts is extended, and the corresponding motion information of addition characteristic point is obtained using similarity boundary function To the characteristic point correspondence set with the Movement consistency.
System provided in an embodiment of the present invention for executing above-mentioned corresponding method, specific embodiment and method Embodiment is consistent, and the algorithm flow being related to is identical as corresponding algorithm process, and details are not described herein again.
The embodiment of the present invention carries out local distance constraint to set by the initial matching point to acquisition, realizes based on part Structure-preserved Movement consistency constraint, can be from the endoscopic images feature point correspondence there are much noise, accurately Erroneous matching is rejected on ground, while considering the consistency of invariance and global motion of the partial structurtes in deformation, can be with robust Ground solves the matching problem under different types of deformation.
On that basi of the above embodiments, Fig. 6 is a kind of endoscopic images Feature Points Matching system provided in an embodiment of the present invention The third processing module minor structure schematic diagram of system, as shown in fig. 6, third processing module 43 specifically includes: third computational submodule 431 and the 4th computational submodule 432, in which:
Third computational submodule 431 is used to be based on the characteristic point correspondence set, using the affine fortune of bilateral variation Moving boundary obtains the bilateral affine motion consistency model;4th computational submodule 432 is for being arranged the bilateral movement side The adaptive space threshold parameter on boundary obtains the corresponding interior point matching set of the global image, and realization described two to be matched Endoscopic images between Feature Points Matching.
System provided in an embodiment of the present invention for executing above-mentioned corresponding method, specific embodiment and method Embodiment is consistent, and the algorithm flow being related to is identical as corresponding algorithm process, and details are not described herein again.
The embodiment of the present invention is all had adaptively by the application of adaptive threshold in the Deformation Types in face of different scale Constraint, be effectively adapted to the endoscopic images of unknown deformation size, can retain in the case where guaranteeing high-precision situation enough More proper characteristics point correspondence, convenient for being applied in the operation such as subsequent structural remodeling.
Figure 10 illustrates a kind of entity structure schematic diagram of server, and as shown in Figure 10, which may include: processing Device (processor) 1010, communication interface (Communications Interface) 1020,1030 He of memory (memory) Communication bus 1040, wherein processor 1010, communication interface 1020, memory 1030 are completed mutually by communication bus 1040 Between communication.Processor 1010 can call the logical order in memory 1030, to execute following method: be waited for based on two width The endoscopic images matched extract image characteristic point, and complete described image Feature Points Matching based on Feature Descriptor similarity, obtain Take the initial matching point of described image characteristic point to set;Based on the unknown deformation between two endoscopic images to be matched Corresponding relationship and partial structurtes between described image characteristic point local neighborhood structure are to the initial matching point to set Local distance constraint is carried out, characteristic point information is extended in conjunction with affine parameter and character pair point motion information, estimates phase Like degree boundary, the characteristic point correspondence set with Movement consistency is obtained;Based on the characteristic point correspondence set pair Space length perception optimizes, and generates bilateral affine motion consistency model, the adaptive distance of bilateral moving boundaries is arranged Threshold parameter obtains the corresponding interior point matching set of global image, realizes between two endoscopic images to be matched Feature Points Matching.
In addition, the logical order in above-mentioned memory 1030 can be realized by way of SFU software functional unit and conduct Independent product when selling or using, can store in a computer readable storage medium.Based on this understanding, originally Substantially the part of the part that contributes to existing technology or the technical solution can be in other words for the technical solution of invention The form of software product embodies, which is stored in a storage medium, including some instructions to So that a computer equipment (can be personal computer, server or the network equipment etc.) executes each implementation of the present invention The all or part of the steps of example the method.And storage medium above-mentioned include: USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic or disk etc. it is various It can store the medium of program code.
On the other hand, the embodiment of the present invention also provides a kind of non-transient computer readable storage medium, is stored thereon with meter Calculation machine program, which is implemented to carry out the various embodiments described above offer method when being executed by processor, for example, Image characteristic point is extracted based on two endoscopic images to be matched, and described image spy is completed based on Feature Descriptor similarity Sign point matching, obtains the initial matching point of described image characteristic point to set;Based on two endoscopic images to be matched Between unknown deformation and described image characteristic point local neighborhood structure between corresponding relationship and partial structurtes to described initial Matching double points set carries out local distance constraint, carries out in conjunction with affine parameter and character pair point motion information to characteristic point information Extension estimates similarity boundary, obtains the characteristic point correspondence set with Movement consistency;It is corresponding based on the characteristic point Set of relationship optimizes space length perception, generates bilateral affine motion consistency model, bilateral moving boundaries are arranged Adaptive distance threshold parameters obtain the corresponding interior point matching set of global image, realize two endoscopes to be matched Feature Points Matching between image.
The apparatus embodiments described above are merely exemplary, wherein described, unit can as illustrated by the separation member It is physically separated with being or may not be, component shown as a unit may or may not be physics list Member, it can it is in one place, or may be distributed over multiple network units.It can be selected according to the actual needs In some or all of the modules achieve the purpose of the solution of this embodiment.Those of ordinary skill in the art are not paying creativeness Labour in the case where, it can understand and implement.
Through the above description of the embodiments, those skilled in the art can be understood that each embodiment can It realizes by means of software and necessary general hardware platform, naturally it is also possible to pass through hardware.Based on this understanding, on Stating technical solution, substantially the part that contributes to existing technology can be embodied in the form of software products in other words, should Computer software product may be stored in a computer readable storage medium, such as ROM/RAM, magnetic disk, CD, including several fingers It enables and using so that a computer equipment (can be personal computer, server or the network equipment etc.) executes each implementation Method described in certain parts of example or embodiment.
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations;Although Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be used To modify the technical solutions described in the foregoing embodiments or equivalent replacement of some of the technical features; And these are modified or replaceed, technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution spirit and Range.

Claims (10)

1. a kind of endoscopic images characteristic point matching method characterized by comprising
S1 extracts image characteristic point based on two endoscopic images to be matched, and completes institute based on Feature Descriptor similarity Image Feature Point Matching is stated, obtains the initial matching point of described image characteristic point to set;
S2, based on the unknown deformation and described image characteristic point local neighborhood structure between two endoscopic images to be matched Between corresponding relationship and partial structurtes, to the initial matching point to set carry out local distance constraint, and combine it is affine Parameter and character pair point motion information, are extended characteristic point information, estimate similarity boundary, obtain having movement consistent The characteristic point correspondence set of property;
S3 optimizes space length perception based on the characteristic point correspondence set, it is consistent to generate bilateral affine motion Property model, the adaptive distance threshold parameters of bilateral moving boundaries are set, the corresponding interior point matching set of global image is obtained, it is real Feature Points Matching between existing two endoscopic images to be matched.
2. a kind of endoscopic images characteristic point matching method according to claim 1, which is characterized in that the step of the S2 It specifically includes:
S21 establishes the unknown deformation between two endoscopic images to be matched and described image characteristic point local neighborhood knot Corresponding relationship between structure;
S22 is based on each described image characteristic point, and corresponding 6 closest approaches are tied in corresponding point set to construct the part Structure;
S23, based between two endoscopic images to be matched unknown deformation and described image characteristic point local neighborhood knot Corresponding relationship between structure, and any the distance between the initial matching point pair and its neighborhood point are fixed, and threshold parameter is arranged, The interior point set kept with locality is calculated and obtained, realizes local distance constraint;
S24, the characteristic point information extracted to the two images to be matched are extended, the corresponding fortune of addition characteristic point Dynamic information obtains the characteristic point correspondence set with the Movement consistency using similarity boundary function.
3. a kind of endoscopic images characteristic point matching method according to claim 2, which is characterized in that the step S22 Including being calculated based on Euclidean distance formula, obtain based on each described image characteristic point and its in the corresponding point set Corresponding 6 closest approaches are in conjunction come the partial structurtes that construct.
4. a kind of endoscopic images characteristic point matching method according to claim 3, which is characterized in that the step S24 Middle application similarity boundary function obtains the characteristic point correspondence set with Movement consistency, is based on the tool Obtained by the interior point set for thering is locality to keep.
5. a kind of endoscopic images characteristic point matching method according to claim 2, which is characterized in that the step of the S3 It specifically includes:
S31 is based on the characteristic point correspondence set, using the affine motion boundary of bilateral variation, obtains described bilateral imitative Penetrate Movement consistency model;
S32 is arranged the adaptive space threshold parameter of the bilateral moving boundaries, obtains the corresponding interior point of the global image With set, the Feature Points Matching between two endoscopic images to be matched is realized.
6. a kind of endoscopic images characteristic point matching method according to claim 5, which is characterized in that the step S32 The adaptive space threshold parameter of the middle setting bilateral moving boundaries, specifically includes:
Model is kept in conjunction with partial structurtes, sets the adaptive space threshold parameter of the bilateral moving boundaries to and the office The distance restraint threshold parameter of portion's structure is consistent.
7. a kind of endoscopic images characteristic point matching method according to claim 6, which is characterized in that the step S32 Middle calculating simultaneously obtains the corresponding interior point matching set of the global image, specifically includes:
Pass through estimation result and described image characteristic point of the setting described image characteristic point in bilateral moving boundaries both direction The distance between noise observation data threshold value is gathered to obtain the corresponding interior point matching of the global image.
8. a kind of endoscopic images Feature Points Matching system characterized by comprising
First processing module for extracting image characteristic point based on two endoscopic images to be matched, and is described based on feature Sub- similarity completes described image Feature Points Matching, obtains the initial matching point of described image characteristic point to set;
Second processing module, for based on the unknown deformation and described image feature between two endoscopic images to be matched Corresponding relationship and partial structurtes between point local neighborhood structure carry out local distance about to set to the initial matching point Beam is extended characteristic point information in conjunction with affine parameter and character pair point motion information, estimates similarity boundary, is had There is the characteristic point correspondence set of Movement consistency;
Third processing module is generated double for being optimized based on the characteristic point correspondence set to space length perception Side affine motion consistency model is arranged the adaptive distance threshold parameters of bilateral moving boundaries, it is corresponding to obtain global image Interior point matching set, realizes the Feature Points Matching between two endoscopic images to be matched.
9. a kind of electronic equipment including memory, processor and stores the calculating that can be run on a memory and on a processor Machine program, which is characterized in that the processor is realized as described in any one of claim 1 to 7 when executing described program in one kind The step of sight glass image characteristic point matching method.
10. a kind of non-transient computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer Realizing a kind of endoscopic images characteristic point matching method as described in any one of claim 1 to 7 when program is executed by processor Step.
CN201910521138.9A 2019-06-17 2019-06-17 A kind of endoscopic images characteristic point matching method and system Pending CN110245671A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910521138.9A CN110245671A (en) 2019-06-17 2019-06-17 A kind of endoscopic images characteristic point matching method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910521138.9A CN110245671A (en) 2019-06-17 2019-06-17 A kind of endoscopic images characteristic point matching method and system

Publications (1)

Publication Number Publication Date
CN110245671A true CN110245671A (en) 2019-09-17

Family

ID=67887435

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910521138.9A Pending CN110245671A (en) 2019-06-17 2019-06-17 A kind of endoscopic images characteristic point matching method and system

Country Status (1)

Country Link
CN (1) CN110245671A (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101161201A (en) * 2007-11-08 2008-04-16 珠海友通科技有限公司 Method for registrating external circumstance DSA elasticity automatically
US20150098659A1 (en) * 2012-10-26 2015-04-09 Calex Llc Method and apparatus for image retrieval
CN109008909A (en) * 2018-07-13 2018-12-18 宜宾学院 A kind of low-power consumption capsule endoscope Image Acquisition and three-dimensional reconstruction system
CN109697692A (en) * 2018-12-29 2019-04-30 安徽大学 One kind being based on the similar feature matching method of partial structurtes

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101161201A (en) * 2007-11-08 2008-04-16 珠海友通科技有限公司 Method for registrating external circumstance DSA elasticity automatically
US20150098659A1 (en) * 2012-10-26 2015-04-09 Calex Llc Method and apparatus for image retrieval
CN109008909A (en) * 2018-07-13 2018-12-18 宜宾学院 A kind of low-power consumption capsule endoscope Image Acquisition and three-dimensional reconstruction system
CN109697692A (en) * 2018-12-29 2019-04-30 安徽大学 One kind being based on the similar feature matching method of partial structurtes

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
郭晓君 等: "基于特征点的内窥镜图像和CT影像配准方法", 《现代商贸工业》 *

Similar Documents

Publication Publication Date Title
JP6255463B2 (en) Feature extraction and matching and template update for biometric authentication
US9202280B2 (en) Position estimation based rotation of switched off light source
Revaud et al. Epicflow: Edge-preserving interpolation of correspondences for optical flow
Delaunoy et al. Photometric bundle adjustment for dense multi-view 3d modeling
Wang et al. Occlusion-aware depth estimation using light-field cameras
Gotardo et al. Computing smooth time trajectories for camera and deformable shape in structure from motion with occlusion
US10217234B2 (en) Modeling method and apparatus using three-dimensional (3D) point cloud
ES2680825T3 (en) Eye look tracking based on correlation of adaptive homography
Yang et al. Spatial-depth super resolution for range images
US10048749B2 (en) Gaze detection offset for gaze tracking models
Buchanan et al. Damped newton algorithms for matrix factorization with missing data
Ruiz et al. Information theory in computer vision and pattern recognition
KR101278430B1 (en) Method and circuit arrangement for recognising and tracking eyes of several observers in real time
Arth et al. Wide area localization on mobile phones
Zuo et al. On generation and analysis of synthetic iris images
US6954544B2 (en) Visual motion analysis method for detecting arbitrary numbers of moving objects in image sequences
Torr Bayesian model estimation and selection for epipolar geometry and generic manifold fitting
Alvarez et al. Dense disparity map estimation respecting image discontinuities: A PDE and scale-space based approach
US9177381B2 (en) Depth estimate determination, systems and methods
Sun et al. Super-resolution from internet-scale scene matching
WO2015161816A1 (en) Three-dimensional facial recognition method and system
CN102834837B (en) Using the iris identification device and method of multiple iris templates
KR20190065287A (en) Prediction of depth from image data using statistical model
US20140010407A1 (en) Image-based localization
US10466778B2 (en) Eye image selection

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