CN103927785B - A kind of characteristic point matching method towards up short stereoscopic image data - Google Patents
A kind of characteristic point matching method towards up short stereoscopic image data Download PDFInfo
- Publication number
- CN103927785B CN103927785B CN201410162951.9A CN201410162951A CN103927785B CN 103927785 B CN103927785 B CN 103927785B CN 201410162951 A CN201410162951 A CN 201410162951A CN 103927785 B CN103927785 B CN 103927785B
- Authority
- CN
- China
- Prior art keywords
- triangle
- region
- matched
- characteristic point
- image points
- 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.)
- Expired - Fee Related
Links
Landscapes
- Image Analysis (AREA)
Abstract
The present invention provides a kind of characteristic point matching method towards up short stereoscopic image data, comprise the steps: image of the same name successively carrying out characteristic point forward coupling with subregion and triangle for constraints and obtaining forward corresponding image points group, image of the same name is successively inversely mated and obtains reverse corresponding image points group carrying out characteristic point with subregion and triangle for constraints, retain the corresponding image points result matched in forward corresponding image points group and reverse corresponding image points group, the corresponding image points result finally mated；The present invention uses subregion and triangle to carry out feature point detection as constraints successively, substantially increases the time efficiency of feature point detection and improves correct match point number, has important using value in up short stereoscopic image mates.
Description
Technical field
The invention belongs to numeral up short field, relate to a kind of characteristic point matching method.
Background technology
Along with the development of modern survey and draw technology, digital photogrammetry can be digital earth, cybercity construction etc. provide data and
Support.The initial data that digital photogrammetry is obtained, important is the feature of digitized video, including point-like character,
Line feature and planar feature.It is the basis of image analysing computer and Image Matching to the extracting method of these features, available various calculations
Son is carried out, such as: Moravec operator, Forstner operator and Harris operator all can be used to extract characteristic point.To numeral
Conspicuous object in image, not only needs to identify them, in addition it is also necessary to determine their position.The amount of photogrammetric neutral body picture pair
Survey is the basis extracting object three-dimensional information.Digital photogrammetry replaces traditional artificial observation with Image Matching, reaches certainly
The dynamic purpose determining corresponding image points.Image Matching substantially identifies corresponding image points between two width or several images, and it is numeral
Photogrammetric and the key problem of computer vision.Matching process based on gray scale is a kind of more ripe matching process, mainly wraps
Including based on the detection of gray scale similarity and the method for Least-Square Matching, they are all based on image similarity of the same name.
Scale invariant feature conversion (Scale-invariant feature transform, SIFT) is that the local that David Lowe proposes is special
Levy description, its spot detection, characteristic vector are generated and the step such as characteristic matching search be fully incorporated in together be optimized,
Reach close to real-time arithmetic speed.SIFT feature matching algorithm can process translate between two width images, rotate, imitative
Penetrate the matching problem under change situation, there is the strongest matching capacity.But, when obtaining image data, owing to being subject to
Shadow to factors such as precipitation reflection, soil body motion, the setting of high speed camera focal length and calibration precision, shooting angle and scene layouts
Ringing, easily there are quality problems such as obscuring, feature is inconspicuous, texture is unintelligible in image data, now, according to conventional shadow
As matching process, it may appear that the correct situation that match point is few, Mismatching point is many.
Summary of the invention
The present invention is directed to the deficiencies in the prior art, it is therefore intended that provide one can improve correct match point number, reduce error hiding
Count out and improve the characteristic point matching method towards up short stereoscopic image data of time efficiency of feature point detection.
For reaching above-mentioned purpose, the solution of the present invention is:
A kind of matching process towards up short stereoscopic image data characteristic point, comprises the steps:
(1), use the mode of parallel photography treat coupling scene shoot, select two cameras of synchronization shoot respectively two
Width image, as the second width image of piece image sum, chooses the first region to be matched in piece image, at the second width figure
In Xiang and choose the second to be matched region corresponding with the first region to be matched；
(2), SIFT algorithm is utilized to detect characteristic point and the characteristic point in the second region to be matched in the first region to be matched respectively, with
On the basis of characteristic point in first region to be matched respectively by the first region to be matched and the second region segmentation to be matched be multiple one by one
Corresponding subregion；
(3), respectively with each sub regions in the first region to be matched about restraint condition, utilize nearest neighbor distance algorithm to carry out by first
The subregion in region to be matched mates to the characteristic point forward of the corresponding subregion in the second region to be matched, finally matches each height
The corresponding image points in region；
(4), using the corresponding image points of each sub regions in the first region to be matched and the characteristic point that is positioned on subregion edge as triangle
Three summits of shape are split subregions and build triangular network；
(5), with each triangle in step (4) gained triangular network as constraints, nearest neighbor distance algorithm is utilized to carry out
Circulated forward coupling by the triangle in the first region to be matched to the characteristic point of the corresponding triangle in the second region to be matched, just obtain
To corresponding image points group；
(6), respectively with each sub regions in the second region to be matched about restraint condition, utilize nearest neighbor distance algorithm to carry out by second
The subregion in region to be matched inversely mates to the characteristic point of the corresponding subregion in the first region to be matched, finally matches each height
The corresponding image points in region；
(7), using the corresponding image points of each sub regions in the second region to be matched and the characteristic point that is positioned on subregion edge as triangle
Three summits of shape are split subregions and build triangular network；
(8), with each triangle in step (7) gained triangular network as constraints, nearest neighbor distance algorithm is utilized to carry out
Circulated reverse coupling by the triangle in the second region to be matched to the characteristic point of the corresponding triangle in the first region to be matched, obtain inverse
To corresponding image points group；
(9), the corresponding image points result correctly mated according to forward corresponding image points group and reverse corresponding image points group.
In step (2), treat according to the first region to be matched and the second respective gradient in region to be matched and slope aspect and first
Join the region characteristic point the most identical with the second region to be matched cut-point as subregion, the first region to be matched and second
Region to be matched is intactly divided into the subregion of several correspondence.Further, the first region to be matched or the second district to be matched
Territory overlaps between respective adjacent subarea territory.
Characteristic point circulation forward coupling in step (5) specifically includes:
The first step, setting preset loop number of times and triangle size threshold value,；
Second step, using each triangle in step (4) gained triangular network as first order triangle, with first order triangle
For constraints, nearest neighbor distance algorithm is utilized to carry out the first order triangle by the first region to be matched to the second region to be matched
The characteristic point forward coupling of corresponding triangle, obtain the corresponding image points of each first order triangle, n is set as 1, and m sets
It is 0；
3rd step, n=n+1, utilize the cut-point of corresponding image points and the subregion correctly mated in (n-1)th grade of triangle as triangle
Three summits of shape build n-th grade of triangle；
4th step, with n-th grade of triangle as constraints, utilize nearest neighbor distance algorithm to carry out by the of the first region to be matched
N level triangle mates to the characteristic point forward of the corresponding triangle in the second region to be matched, and by repeatedly circulation RANSAC algorithm
Reject error hiding characteristic point, draw the corresponding image points of correct coupling, m=m+1；
5th step, to judge whether whether m be respectively less than triangle not less than the size of preset loop number of times and n-th grade of triangle the biggest
Little threshold value, when m is respectively less than triangle size threshold value not less than the size of preset loop number of times and n-th grade of triangle, obtains
Forward corresponding image points group, otherwise returns and performs the 3rd step,
Wherein, n is the rank of triangle, and m is cycle-index.
The reverse coupling of characteristic point circulation in step (8) specifically includes:
The first step, setting preset loop number of times and triangle size threshold value；
Second step, using each triangle in step (7) gained triangular network as first order triangle, with first order triangle
For constraints, nearest neighbor distance algorithm is utilized to carry out the first order triangle by the second region to be matched to the first region to be matched
The characteristic point of corresponding triangle inversely mate, obtain the corresponding image points of each first order triangle, n' is set as 1, and m' sets
It is 0；
3rd step, n'=n'+1', utilize the cut-point of corresponding image points and the subregion correctly mated in the n-th '-1 grade triangle as three
Three summits of dihedral build n-th ' level triangle；
4th step, with n-th ' level triangle as constraints, utilize nearest neighbor distance algorithm to carry out by the of the second region to be matched
N' level triangle inversely mates to the characteristic point of the corresponding triangle in the first region to be matched, and by repeatedly circulation RANSAC algorithm
Reject error hiding characteristic point, draw the corresponding image points of correct coupling, m'=m'+1；
5th step, judge m' whether not less than preset loop number of times and n-th ' that whether the size of level triangle is respectively less than triangle is big
Little threshold value, when m' not less than preset loop number of times and n-th ' the size of level triangle be respectively less than triangle size threshold value time, obtain
Reverse corresponding image points group, otherwise returns and performs the 3rd step,
Wherein, n' is the rank of triangle, and m' is cycle-index.
Owing to using such scheme, the invention has the beneficial effects as follows:
First, this method uses subregion and Delaunay triangle to carry out feature point detection as constraints successively, significantly carries
The high time efficiency of feature point detection；Secondly, this method is employed many times circulation RANSAC algorithm and rejects the characteristic point of error hiding,
Thus reduce the probability rejecting correct corresponding image points；It addition, this method utilizes SIFT operator feature based to mate, do not relate to
And the internal and external orientation of image, it is to avoid camera parameter is unknown or the relatively big shadow to matching result precision of camera calibration parameter error
Ring；Finally, this method takes binding characteristic point region step by step to mate, it is to avoid the error hiding caused because of feature similarity.
Accompanying drawing explanation
Fig. 1 is the flow chart of the characteristic point matching method towards up short stereoscopic image data in the embodiment of the present invention.
Fig. 2 is the flow chart of the characteristic point forward coupling that the circulation in the embodiment of the present invention is carried out.
Fig. 3 is the flow chart that the characteristic point that the circulation in the embodiment of the present invention is carried out inversely is mated.
Detailed description of the invention
Below in conjunction with accompanying drawing illustrated embodiment, the present invention is further illustrated.
Embodiment
Present embodiments providing a kind of characteristic point matching method towards up short stereoscopic image data, it mainly utilizes subregion
With Delaunay triangle as constraints, SIFT operator is used to carry out feature point detection and coupling, and along with repeatedly circulating
RANSAC operator (Random Sample Consensus) deletes the corresponding image points of error hiding, thus improves correct match point
Count and reduce error hiding and count.
As it is shown in figure 1, the characteristic point matching method towards up short stereoscopic image data in the present embodiment comprises the steps:
The first step, obtain image of the same name to and choose region to be matched, specifically include:
High speed camera is used to use the mode of parallel photography that same scene to be matched is shot, it is thus achieved that several digital pictures, from
The two width images that two high speed cameras of middle selection synchronization shoot respectively respectively as piece image and the second width image (namely
Constitute image of the same name to).In order to improve the time efficiency of feature point detection, select to treat from piece image according to specific needs
The region joined, as the first region to be matched, selects the region corresponding with the first region to be matched as second from the second width image
Region to be matched.
Second step, SIFT algorithm is utilized to detect characteristic point and the characteristic point in the second region to be matched in the first region to be matched respectively.
SIFT algorithm includes feature point detection and corresponding image points coupling, and wherein, feature point detection is to search out in the metric space of image
The Local Extremum of image, then removes the low extreme point of contrast and unstable skirt response point, so that it is determined that the spy of image
Levy a little, afterwards characteristic point peripheral region is carried out image block, calculate each piece of interior histogram of gradients, generate unique vector
Descriptor, the Feature Descriptor of i.e. 128 dimensions, the position of each characteristic point, yardstick and directional information are described.
On the basis of 3rd step, characteristic point in the first region to be matched, respectively by the first region to be matched and the second district to be matched
Regional partition is many sub regions so that the subregion in the first region to be matched and the subregion in the second region to be matched are one_to_one corresponding
Relation.When segmentation, the first region to be matched and the feature such as the second respective gradient in region to be matched and slope aspect to be considered,
And using characteristic point the most identical in the first region to be matched and the second region to be matched as the cut-point of subregion, the
One region to be matched and the second region to be matched are intactly divided into the subregion of several correspondence.To ensure during segmentation that first is to be matched
Overlap, to ensure in subregion intersection characteristic point between region or the second respective adjacent subarea territory, region to be matched
Successful match.
4th step, about restraint condition with each sub regions in the first region to be matched respectively, utilize nearest neighbor distance algorithm carry out by
The subregion in the first region to be matched to the unidirectional coupling of characteristic point (forward coupling) of the corresponding subregion in the second region to be matched, and
Reject the characteristic point of error hiding by repeatedly circulating RANSAC algorithm, thus match the first region to be matched and second to be matched
The corresponding image points of each sub regions in region.
In this step, the corresponding image points coupling in SIFT algorithm uses nearest neighbor distance algorithm, and this nearest neighbor distance algorithm is with sub-district
Territory is as constraints, it is possible to limit the hunting zone of corresponding image points, not only increases the time efficiency of corresponding image points coupling, and
And also significantly lower in error hiding rate.Under this constraints, according to the characteristic point detected in each sub regions, point
The other characteristic point to each sub regions uses nearest neighbor distance algorithm to carry out Feature Points Matching, i.e. uses and sample characteristics point arest neighbors
The Euclidean distance of characteristic point and the ratio of the Euclidean distance of time neighbour's characteristic point and setting threshold ratio relatively, if ratio is less than threshold value,
Then it is considered characteristic point.But, to the characteristic point of coupling in each sub regions, error hiding can be there is as a result, it is possible to pass through repeatedly
Circulation RANSAC algorithm rejects the characteristic point of error hiding.
RANSAC algorithm is according to one group of sample data set comprising abnormal data, calculates the mathematical model parameter of data, obtains
The algorithm of effective sample data.The basic assumption of this algorithm is to comprise just data in sample, also comprises abnormal data, when given
, there is the method that can calculate the model parameter meeting these data in one group of correct data.First RANSAC algorithm is at sample
Notebook data is concentrated and is randomly drawed a subset and be initialized as model, it is judged that whether other data in data set meet this model,
If meeting, constituting and unanimously collecting and calculate new model, until algorithm terminates.Principle according to this algorithm, it can be deduced that: due to
The data subset randomly drawed is different, and its initial model built also can be different, and the most correct data are owing to being unsatisfactory for this mould
Type is mistaken as the data of mistake and rejects, so constitute unanimously concentrates the sample data comprised and the knot obtained through this algorithm
Fruit also can be otherwise varied.
If whole region to be matched being carried out feature point detection and corresponding image points coupling merely with SIFT operator, then in matching result
There is a lot of Mismatching points, this is accomplished by utilizing RANSAC operator to carry out Mismatching point rejecting.If only result being run one
Secondary RANSAC operator, though corresponding image points the most correct in result, substantial amounts of correct match point is also likely to be considered and is
Mismatching point is disallowable, it is therefore desirable to circulation RANSAC algorithm；When circulating RANSAC algorithm to hundreds of times, owing to following every time
The initial model of ring is the most different, therefore can reduce the probability rejecting correct corresponding image points.
5th step, using the cut-point of the corresponding image points having mated out in each sub regions and subregion as three tops of triangle
Point, builds Delaunay tri-in each sub regions in each sub regions in the first region to be matched and the second region to be matched respectively
Dihedral, and Delaunay triangle is collectively forming first order triangular network as first order triangle, multiple first order trianglees.
6th step, with each first order triangle in the 5th step gained first order triangular network as initializing constraint, utilize
Nearest neighbor distance algorithm carries out the triangle by the first region to be matched and circulates to the characteristic point of the corresponding triangle in the second region to be matched
Unidirectional coupling (forward coupling) and the structure of n-th grade of triangle, finally give forward corresponding image points group.
Wherein, as in figure 2 it is shown, the 6th step specifically includes the most step by step:
Step 1-1, setting preset loop number of times and triangle size threshold value；
Step 1-2, with the 5th step gained first order triangle as constraints, utilize nearest neighbor distance algorithm to carry out by first and treat
Join the first order triangle unidirectional coupling of characteristic point (forward coupling) to the corresponding triangle in the second region to be matched in region, and lead to
Cross repeatedly circulation RANSAC algorithm and reject error hiding characteristic point, draw the corresponding image points of each first order triangle of correct coupling,
Rank n of triangle is set as 1, and the value of cycle-index m is set as 0；
Step 1-3, make rank n=n+1 of triangle, utilize corresponding image points and the Zi Qu of correct coupling in (n-1)th grade of triangle
The cut-point in territory builds n-th grade of triangle as three summits of triangle；
Step 1-4, with n-th grade of triangle as constraints, utilize nearest neighbor distance algorithm to carry out by the of the first region to be matched
N level triangle is to the unidirectional coupling of characteristic point (forward coupling) of the corresponding triangle in the second region to be matched, and passes through repeatedly to circulate
RANSAC algorithm rejects error hiding characteristic point, draws the corresponding image points of correct coupling, cycle-index m=m+1；
Step 1-5, judge that cycle-index m is whether not less than preset loop number of times m0And size Tn of n-th grade of triangle is
No respectively less than triangle size threshold value T0, when cycle-index m is equal not less than the size of preset loop number of times and n-th grade of triangle
During less than triangle size threshold value, obtain forward corresponding image points group, then terminate the 6th step, otherwise return and perform step 1-3.
7th step, about restraint condition with each sub regions in the second region to be matched respectively, utilize nearest neighbor distance algorithm carry out by
The subregion in the second region to be matched to the unidirectional coupling of characteristic point (inversely coupling) of the corresponding subregion in the first region to be matched, and
Reject the characteristic point of error hiding by repeatedly circulating RANSAC algorithm, thus match the first region to be matched and second to be matched
The corresponding image points of each sub regions in region.
8th step, using the cut-point of the corresponding image points having mated out in each sub regions and subregion as three tops of triangle
Point, builds Delaunay tri-in each sub regions in each sub regions in the first region to be matched and the second region to be matched respectively
Dihedral, and this Delaunay triangle is collectively forming the first order triangulation network as first order triangle, multiple first order trianglees
Network.
9th step, with each first order triangle in the 8th step gained first order triangular network as initializing constraint, utilize
Nearest neighbor distance algorithm carries out the triangle by the second region to be matched and circulates to the characteristic point of the corresponding triangle in the first region to be matched
Unidirectional coupling (inversely coupling) and n-th ' the structure of level triangle, finally give reverse corresponding image points group.
Wherein, as it is shown on figure 3, the 9th step specifically includes the most step by step:
Step 2-1, setting preset loop number of times and triangle size threshold value；
Step 2-2, with the 8th step gained first order triangle as constraints, utilize nearest neighbor distance algorithm to carry out by second and treat
Join the first order triangle unidirectional coupling of characteristic point (inversely coupling) to the corresponding triangle in the first region to be matched in region, and lead to
Cross repeatedly circulation RANSAC algorithm and reject error hiding characteristic point, draw the corresponding image points of each first order triangle of correct coupling,
Rank n' of triangle sets 1, and the value of cycle-index m' is set as 0；
Step 2-3, make rank n'=n'+1 of triangle, utilize corresponding image points and the Zi Qu of correct coupling in (n-1)th grade of triangle
The cut-point in territory as three summits of triangle build n-th ' level triangle；
Step 2-4, with n-th ' level triangle as constraints, utilize nearest neighbor distance algorithm to carry out by the of the second region to be matched
N level triangle is to the unidirectional coupling of characteristic point (inversely coupling) of the corresponding triangle in the first region to be matched, and passes through repeatedly to circulate
RANSAC algorithm rejects error hiding characteristic point, draws the corresponding image points of correct coupling, makes cycle-index m'=m'+1；
Step 2-5, judge that cycle-index m' is whether not less than preset loop number of times m'0And n-th ' size T' of level trianglenIt is
No respectively less than triangle size threshold value T'0, when cycle-index m' not less than preset loop number of times and n-th ' the size of level triangle is equal
During less than triangle size threshold value, obtain reverse corresponding image points group, then terminate the 9th step, otherwise return and perform step 2-3.
Because image of the same name can detect the most intensive characteristic point to by SIFT operator, visualization discovery has substantial amounts of same
Name picture point, through subregion and the corresponding image points of triangulation network constrained matching and by RANSAC operator rejecting Mismatching point after,
Still the match is successful to have a lot of corresponding image points not have.Therefore, it can by reverse coupling, i.e. with other the one of image centering of the same name
The reverse coupling of characteristic point is carried out on the basis of sub-picture.
The corresponding image points result matched in tenth step, reservation forward corresponding image points group and reverse corresponding image points group, i.e. obtains final
The corresponding image points result of coupling.
The method mentioned according to the present embodiment, uses the stereoscopic image data of high speed camera shooting under the analog platform of landslide to verify that it can
Row.Using based on gray scale Similarity Match Method when not using the method for the present embodiment, whole slip mass the most correctly mates
Go out 422 pairs of corresponding image points, and the bottom of slip mass only matches 12 corresponding image points.The method using the present embodiment, with son
The triangulation network that region and circulation build is constraints, successively utilizes image of the same name to as reference images, enters according to SIFT algorithm
Row coupling repeatedly circulation RANSAC algorithm reject Mismatching point.Through testing these data, result shows, image centering of the same name
Can relatively evenly match corresponding image points 4552, matching number is 10 times in the past, substantially increases correct coupling and counts.
In sum, under the influence of the factor such as imaging model, external environment, the image quality acquired in high speed camera is relatively
, if directly using SIFT and RANSAC algorithm, there are the results such as few, the skewness of the corresponding image points that successful match goes out in difference；And
It is that constraints repeatedly circulates RANSAC algorithm and carries out the double of image pair of the same name that the present embodiment uses by subregion and the triangulation network
To coupling, obtain corresponding image points data many and be evenly distributed, and reduce Mismatching point, mate at up short stereoscopic image
And the aspect such as three-dimensional stereo model reconstruction has important using value.
The above-mentioned description to embodiment is to be understood that for ease of those skilled in the art and use the present invention.It is familiar with
These embodiments obviously easily can be made various amendment by the personnel of art technology, and should General Principle described herein
Use in other embodiments without through performing creative labour.Therefore, the invention is not restricted to above-described embodiment, art technology
Personnel should be in protection scope of the present invention according to the announcement of the present invention, the improvement made without departing from scope and amendment
Within.
Claims (5)
1. the characteristic point matching method towards up short stereoscopic image data, it is characterised in that: comprise the steps:
(1), use the mode of parallel photography treat coupling scene shoot, select two cameras of synchronization shoot respectively two
Width image, as piece image and the second width image, chooses the first region to be matched in described piece image, described
In two width images and choose the second to be matched region corresponding with described first region to be matched；
(2) SIFT algorithm, is utilized to detect characteristic point and the spy in described second region to be matched in described first region to be matched respectively
Levy a little, respectively described first region to be matched and described second are treated on the basis of the characteristic point in described first region to be matched
Joining region segmentation is multiple subregions one to one；
(3), respectively with each sub regions in described first region to be matched about restraint condition, utilize nearest neighbor distance algorithm carry out by
The subregion in described first region to be matched mates to the characteristic point forward of the corresponding subregion in described second region to be matched, finally
Match the corresponding image points of each sub regions；
(4), with the corresponding image points of each sub regions in described first region to be matched and the characteristic point that is positioned on described subregion edge
Three summits as triangle are split subregions and build triangular network；
(5), with each triangle in step (4) gained triangular network as constraints, nearest neighbor distance algorithm is utilized to carry out
Forward coupling is circulated to the characteristic point of the corresponding triangle in described second region to be matched by the triangle in described first region to be matched,
Obtain forward corresponding image points group；
(6), respectively with each sub regions in described second region to be matched about restraint condition, utilize nearest neighbor distance algorithm carry out by
The subregion in described second region to be matched inversely mates, finally to the characteristic point of the corresponding subregion in described first region to be matched
Match the corresponding image points of each sub regions；
(7), with the corresponding image points of each sub regions in described second region to be matched and the characteristic point that is positioned on described subregion edge
Three summits as triangle are split subregions and build triangular network；
(8), with each triangle in step (7) gained triangular network as constraints, nearest neighbor distance algorithm is utilized to carry out
Reverse coupling is circulated to the characteristic point of the corresponding triangle in described first region to be matched by the triangle in described second region to be matched,
Obtain reverse corresponding image points group；
(9), the corresponding image points result correctly mated according to described forward corresponding image points group and described reverse corresponding image points group.
Characteristic point matching method towards up short stereoscopic image data the most according to claim 1, it is characterised in that:
In step (2), according to described first region to be matched and the described second respective gradient in region to be matched and slope aspect and described the
The one region to be matched characteristic point the most identical with described second region to be matched, as the cut-point of subregion, is treated described first
Matching area and described second region to be matched are intactly divided into the subregion of several correspondence.
Characteristic point matching method towards up short stereoscopic image data the most according to claim 2, it is characterised in that: institute
State and overlap between the first region to be matched or described second respective adjacent subarea territory, region to be matched.
Characteristic point matching method towards up short stereoscopic image data the most according to claim 2, it is characterised in that: step
Suddenly the circulation of the described characteristic point in (5) forward coupling includes:
The first step, setting preset loop number of times and triangle size threshold value；
Second step, using each triangle in step (4) gained triangular network as first order triangle, with the described first order three
Dihedral is constraints, utilizes described nearest neighbor distance algorithm to carry out the first order triangle by described first region to be matched to institute
State the characteristic point forward coupling of the corresponding triangle in the second region to be matched, obtain the corresponding image points of each described first order triangle,
N is set as 1, and m is set as 0；
3rd step, n=n+1, utilize the described cut-point conduct of the corresponding image points of correct coupling and subregion in the n-th 1 grades of trianglees
Three summits of triangle build n-th grade of triangle；
4th step, with described n-th grade of triangle as constraints, utilize described nearest neighbor distance algorithm to carry out by described first and treat
Described n-th grade of triangle of matching area mates to the characteristic point forward of the corresponding triangle in described second region to be matched, and leads to
Cross repeatedly circulation RANSAC algorithm and reject error hiding characteristic point, draw the corresponding image points of correct coupling, m=m+1；
5th step, judge whether whether m be respectively less than described not less than the size of described preset loop number of times and n-th grade of triangle
Triangle size threshold value, when m is respectively less than described three not less than the size of described preset loop number of times and described n-th grade of triangle
During dihedral size threshold value, obtain forward corresponding image points group, otherwise return and perform the 3rd step,
Wherein, n is the rank of triangle, and m is cycle-index.
Characteristic point matching method towards up short stereoscopic image data the most according to claim 2, it is characterised in that: step
Suddenly the reverse coupling of the circulation of the described characteristic point in (8) includes:
The first step, setting preset loop number of times and triangle size threshold value；
Second step, using each triangle in step (7) gained triangular network as first order triangle, with the described first order three
Dihedral is constraints, utilizes described nearest neighbor distance algorithm to carry out the first order triangle by described second region to be matched to institute
The characteristic point of the corresponding triangle stating the first region to be matched is inversely mated, and obtains the corresponding image points of each described first order triangle,
N' is set as 1, and m' is set as 0；
3rd step, n'=n'+1', utilize the corresponding image points of correct coupling in the n-th ' 1 grade triangle to make with the described cut-point of subregion
For three summits of triangle build n-th ' level triangle；
4th step, with described n-th ' level triangle as constraints, utilize described nearest neighbor distance algorithm to carry out by described second and treat
Matching area described n-th ' level triangle inversely mates to the characteristic point of the corresponding triangle in described first region to be matched, and leads to
Cross repeatedly circulation RANSAC algorithm and reject error hiding characteristic point, draw the corresponding image points of correct coupling, m'=m'+1；
5th step, judge m' whether not less than described preset loop number of times and n-th ' whether the size of level triangle be respectively less than described
Triangle size threshold value, when m' not less than described preset loop number of times and described n-th ' the size of level triangle is respectively less than described three
During dihedral size threshold value, obtain reverse corresponding image points group, otherwise return and perform the 3rd step,
Wherein, n' is the rank of triangle, and m' is cycle-index.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410162951.9A CN103927785B (en) | 2014-04-22 | 2014-04-22 | A kind of characteristic point matching method towards up short stereoscopic image data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410162951.9A CN103927785B (en) | 2014-04-22 | 2014-04-22 | A kind of characteristic point matching method towards up short stereoscopic image data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103927785A CN103927785A (en) | 2014-07-16 |
| CN103927785B true CN103927785B (en) | 2016-08-24 |
Family
ID=51145996
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410162951.9A Expired - Fee Related CN103927785B (en) | 2014-04-22 | 2014-04-22 | A kind of characteristic point matching method towards up short stereoscopic image data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103927785B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104574519B (en) * | 2015-01-31 | 2017-05-31 | 华北水利水电大学 | Multi-source resident's terrain feature exempts from the automatic sane matching process of threshold value |
| WO2017091927A1 (en) * | 2015-11-30 | 2017-06-08 | 华为技术有限公司 | Image processing method and dual-camera system |
| CN108108700B (en) * | 2017-12-27 | 2020-02-21 | 江苏大学 | Pig feature region identification method based on chord axis transformation |
| CN108364343B (en) * | 2018-01-11 | 2021-07-09 | 东南大学 | Rapid setting and controlling method for close-range photography |
| CN108765532B (en) * | 2018-05-04 | 2023-08-22 | 卢卡(北京)智能科技有限公司 | Child drawing model building method, reading robot and storage device |
| CN109615646A (en) * | 2018-12-17 | 2019-04-12 | 石家庄爱赛科技有限公司 | Stereo matching system and solid matching method |
| CN110111375B (en) * | 2019-03-29 | 2021-05-04 | 中国地质大学（武汉） | Image matching gross error elimination method and device under Delaunay triangulation network constraint |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101350101A (en) * | 2008-09-09 | 2009-01-21 | 北京航空航天大学 | Method for auto-registration of multi-amplitude deepness image |
| CN101833765A (en) * | 2010-04-30 | 2010-09-15 | 天津大学 | Characteristic matching method based on bilateral matching and trilateral restraining |
| CN103177444A (en) * | 2013-03-08 | 2013-06-26 | 中国电子科技集团公司第十四研究所 | Automatic SAR (synthetic-aperture radar) image rectification method |
-
2014
- 2014-04-22 CN CN201410162951.9A patent/CN103927785B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101350101A (en) * | 2008-09-09 | 2009-01-21 | 北京航空航天大学 | Method for auto-registration of multi-amplitude deepness image |
| CN101833765A (en) * | 2010-04-30 | 2010-09-15 | 天津大学 | Characteristic matching method based on bilateral matching and trilateral restraining |
| CN103177444A (en) * | 2013-03-08 | 2013-06-26 | 中国电子科技集团公司第十四研究所 | Automatic SAR (synthetic-aperture radar) image rectification method |
Non-Patent Citations (3)
| Title |
|---|
| 《A New Remote Sensing Image Registration Approach Based on Retrofitted SIFT Algorithm And a Novel Similarity Measure》;ZhiLi Song;《International Conference on Computational Intelligence & Software Engineering》;20091231;全文 * |
| 《基于控制点约束及区域相关的立体匹配算法》;赵聪等;《山东理工大学学报》;20081130;第22卷(第6期);全文 * |
| 《多重约束下由粗到精的多源图像自适应配准算法》;邢帅等;《光电工程》;20070630;第34卷(第6期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103927785A (en) | 2014-07-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103927785B (en) | A kind of characteristic point matching method towards up short stereoscopic image data | |
| CN104966270B (en) | A kind of more image split-joint methods | |
| CN107833179A (en) | The quick joining method and system of a kind of infrared image | |
| CN108596108B (en) | Aerial remote sensing image change detection method based on triple semantic relation learning | |
| CN104867137B (en) | A kind of method for registering images based on improvement RANSAC algorithms | |
| CN106485690A (en) | Cloud data based on a feature and the autoregistration fusion method of optical image | |
| CN108921864A (en) | A kind of Light stripes center extraction method and device | |
| CN110110793B (en) | Binocular image rapid target detection method based on double-current convolutional neural network | |
| CN104134200A (en) | Mobile scene image splicing method based on improved weighted fusion | |
| CN107563357B (en) | Live-broadcast clothing dressing recommendation method and device based on scene segmentation and computing equipment | |
| CN113610905B (en) | Deep learning remote sensing image registration method based on sub-image matching and application | |
| CN103971366A (en) | Stereoscopic matching method based on double-weight aggregation | |
| CN104700355A (en) | Generation method, device and system for indoor two-dimension plan | |
| CN113095371A (en) | Feature point matching method and system for three-dimensional reconstruction | |
| CN111783834B (en) | Heterogeneous image matching method based on joint graph spectrum feature analysis | |
| Du et al. | Double-channel guided generative adversarial network for image colorization | |
| CN107194960B (en) | A kind of method for registering for high spectrum image | |
| Xiang et al. | PhenoStereo: a high-throughput stereo vision system for field-based plant phenotyping-with an application in sorghum stem diameter estimation | |
| CN113962864A (en) | Image splicing method and device, storage medium and electronic device | |
| CN109087344A (en) | Image-selecting method and device in three-dimensional reconstruction | |
| CN107610216B (en) | Particle swarm optimization-based multi-view three-dimensional point cloud generation method and applied camera | |
| CN104751412B (en) | A kind of image split-joint method based on affine invariants | |
| CN116704111B (en) | Image processing method and apparatus | |
| CN105096309B (en) | A kind of edge detection method and device based on X-ray | |
| CN114998397B (en) | Multi-view satellite image stereopair optimization selection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160824 Termination date: 20190422 |