CN104992433B - The method and device of multi-spectral image registration based on line match - Google Patents
The method and device of multi-spectral image registration based on line match Download PDFInfo
- Publication number
- CN104992433B CN104992433B CN201510347566.6A CN201510347566A CN104992433B CN 104992433 B CN104992433 B CN 104992433B CN 201510347566 A CN201510347566 A CN 201510347566A CN 104992433 B CN104992433 B CN 104992433B
- Authority
- CN
- China
- Prior art keywords
- intersection point
- point
- line segment
- intersection
- mtd
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
- G06T2207/10036—Multispectral image; Hyperspectral image
Abstract
The present invention provides a kind of method and device of the multi-spectral image registration based on line match.Method the present invention is based on the multi-spectral image registration of line match includes：The first intersection point of multigroup intersecting line segment pair in multispectral image subject to registration is obtained, and obtains the second intersection point with reference to multigroup intersecting line segment pair in multispectral image；It determines with reference to multiple second intersection points with the first intersection point pre-matching on multispectral image；Feature Points Matching is carried out to the second intersection point and the first intersection point, determine with reference on multispectral image with matched second intersection point of the first intersection point；According to the first intersection point and with matched second intersection point of the first intersection point, determine coordinate conversion parameter；Registration is carried out to multispectral image subject to registration and with reference to multispectral image according to coordinate conversion parameter.The present invention improves the accuracy of multi-spectral image registration.
Description
Technical field
The present invention relates to field of communication technology more particularly to a kind of methods of the multi-spectral image registration based on line match
And device.
Background technology
Image registration (Image registration) refers to determine the conversion ginseng between image according to some similarity measurements
Number makes two width or multiple image of the Same Scene obtained from different detectors, different visual angles, different time, transforms to same
Under coordinate system, the process of best match is obtained on pixel layer, it is image co-registration, target identification, object variations detection, calculating
One important prior step of the processes such as machine vision, has a wide range of applications in fields such as military affairs, remote sensing, medicine.
A kind of existing method for registering images is to convert (Scale-invariant feature using scale invariant feature
Transform, SIFT) method, it is as follows：Graphical rule space is firstly generated, then detects the office in metric space
Portion's extreme point, then by rejecting the point of the low contrast in Local Extremum and skirt response point, obtain characteristic point；Feature is clicked through
During row description, gradient orientation histogram statistics is carried out to the region centered on characteristic point, obtains the direction distribution of characteristic point, and
Generate Feature Descriptor；Finally by matched characteristic point pair between Feature Descriptor searching image, then pass through matched spy
Sign point finally carries out image registration to obtaining image space coordinate conversion parameter according to coordinate conversion parameter.
But the above method is when being used for multi-spectral image registration, since the content of multispectral image is more complicated, and picture
There is no correspondence between plain value, texture information has very big difference.Thus, the characteristic point determined according to the above method is not
Stablize, without representativeness, by characteristic point match somebody with somebody will definitely cause the accuracy of registration relatively low.
The content of the invention
The present invention provides a kind of method and device of the multi-spectral image registration based on line match, to solve the prior art
The accuracy of middle registration it is relatively low the problem of.
In a first aspect, the embodiment of the present invention provides a kind of method of the multi-spectral image registration based on line match, including：
The first intersection point of multigroup intersecting line segment pair in multispectral image subject to registration is obtained, and is obtained with reference to multispectral image
In multigroup intersecting line segment pair the second intersection point；
Determine multiple second intersection points with reference to the first intersection point pre-matching on multispectral image；
Feature Points Matching is carried out to second intersection point and first intersection point, determine it is described with reference on multispectral image with
Matched second intersection point of first intersection point；
According to first intersection point and with matched second intersection point of first intersection point, determine coordinate conversion parameter；
The multispectral image subject to registration and the multispectral image that refers to are matched somebody with somebody according to the coordinate conversion parameter
It is accurate.
Optionally, it is described to obtain in multispectral image subject to registration before first intersection point of multigroup intersecting line segment pair, including：
Determine the marginal point of the multispectral image subject to registration；
The marginal point for being spaced pre-determined distance is connected to form line segment；
The angle of adjacent segments is calculated, if the angle is less than default angle threshold, retains the adjacent segments；If
More than or equal to the default angle threshold, then the intersection point of the adjacent segments is deleted, connects the another of the adjacent segments
A endpoint forms new line segment；
Multigroup intersecting line segment pair is obtained according to the line segment.
Optionally, before second intersection point obtained with reference to multigroup intersecting line segment pair in multispectral image, including：
Determine the marginal point with reference to multispectral image；
The marginal point for being spaced pre-determined distance is connected to form line segment；
The angle of adjacent segments is calculated, if the angle is less than default angle threshold, retains the adjacent segments；If
More than or equal to the default angle threshold, then the intersection point of the adjacent segments is deleted, connects the another of the adjacent segments
A endpoint forms new line segment；
Multigroup intersecting line segment pair is obtained according to the line segment.
Optionally, it is described to determine multiple second friendships with reference to the first intersection point pre-matching on multispectral image
Point, including：
Respectively to first intersection point and the second intersection point calculation edge orientation histogram EOH descriptors；
Calculate the Euclidean distance between the EOH descriptors of first intersection point and the EOH descriptors of second intersection point；
If the Euclidean distance is less than default first distance threshold, it is determined that second intersection point is to be handed over described first
Second intersection point of point pre-matching.
Optionally, it is described Feature Points Matching is carried out to second intersection point and first intersection point before, including：
On the line segment where second intersection point, it is spaced pre-determined distance successively in the both sides of second intersection point and chooses the
One characteristic point；
On the line segment where first intersection point, the pre-determined distance choosing is spaced successively in the both sides of first intersection point
Take second feature point；Line segment where second intersection point is consistent with the line segment direction where first intersection point.
Optionally, the pre-determined distance that is spaced successively in the both sides of second intersection point chooses fisrt feature point, specific to wrap
It includes：
It is the fisrt feature point that the intersection point on line segment is chosen in the both sides of second intersection point；
It is described to be spaced the pre-determined distance selection second feature point successively in the both sides of first intersection point, it specifically includes：
It is the second feature point that the intersection point on line segment is chosen in the both sides of first intersection point.
Optionally, Feature Points Matching is carried out to second intersection point and first intersection point, determined described with reference to multispectral
On image with matched second intersection point of first intersection point, including：
EOH descriptors are calculated to the fisrt feature point and second feature point respectively；
Calculate Euclidean between the EOH descriptors of the fisrt feature point and the EOH descriptors of second feature point away from
From；The fisrt feature point is corresponding with the position of the second feature point；
If the Euclidean distance for having more than default amount threshold is less than default second distance threshold value, it is determined that described the
Two intersection points be and matched second intersection point of first intersection point；The second distance threshold value is less than first distance threshold.
Optionally, it is described according to first intersection point and with matched second intersection point of first intersection point, determine coordinate
Transformation parameter, including：
Transformation matrix of coordinates H is determined according to formula below (1), is joined the transformation matrix of coordinates H as coordinate transform
Number；
Wherein,For the coordinate of second intersection point；For the coordinate of first intersection point；Transformation matrix of coordinates
Optionally, it is described multispectral to the multispectral image subject to registration and the reference according to the coordinate conversion parameter
Image carries out registration, including：
The coordinate of pixel in the multispectral image subject to registration is multiplied with the transformation matrix of coordinates so that described
Multispectral image subject to registration transform to under the coordinate system identical with reference to multispectral image.
Second aspect, the embodiment of the present invention provide a kind of device of the multi-spectral image registration based on line match, including：
Acquisition module for obtaining the first intersection point of multigroup intersecting line segment pair in multispectral image subject to registration, and obtains
With reference to the second intersection point of multigroup intersecting line segment pair in multispectral image；
Determining module, for determining multiple second friendships with reference to the first intersection point pre-matching on multispectral image
Point；
The determining module is additionally operable to carry out Feature Points Matching to second intersection point and first intersection point, determines institute
State with reference on multispectral image with matched second intersection point of first intersection point；Line segment where second intersection point and described the
Line segment direction where one intersection point is consistent；
The determining module, be additionally operable to according to first intersection point and with matched second intersection point of first intersection point,
Determine coordinate conversion parameter；
Processing module, for according to the coordinate conversion parameter to the multispectral image subject to registration and described with reference to mostly light
Spectrogram picture carries out registration.
The method and device of multi-spectral image registration provided by the invention based on line match is subject to registration more by obtaining
Spectrum picture and the intersection point with reference to multigroup intersecting line segment pair in multispectral image carry out pre- using the intersection point of line segment as characteristic point
Matching, and the second intersection point that pre-matching goes out and the first intersection point are further matched, finally definite matched intersection point pair, and root
According to the matched intersection point determined to determining coordinate conversion parameter, finally according to the coordinate conversion parameter to described subject to registration more
Spectrum picture and the reference multispectral image carry out registration, compared with prior art, since line segment feature is in multispectral image
In there is better stability, therefore the intersection point of line segment select as characteristic point match and improve multi-spectral image registration
Accuracy.
Description of the drawings
It, below will be to embodiment or the prior art in order to illustrate more clearly of technical solution of the invention or of the prior art
Attached drawing is briefly described needed in description, it should be apparent that, the accompanying drawings in the following description is the one of the present invention
A little embodiments, for those of ordinary skill in the art, without having to pay creative labor, can also be according to this
A little attached drawings obtain other attached drawings.
Fig. 1 is the flow chart of one embodiment of method of the multi-spectral image registration the present invention is based on line match；
Fig. 2 is the structure diagram of one embodiment of device of the multi-spectral image registration the present invention is based on line match.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with the attached drawing in the present invention, to this
Technical solution in invention is clearly and completely described, it is clear that and described embodiment is 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 are not making creative labor
All other embodiments obtained under the premise of dynamic, belong to the scope of protection of the invention.
Fig. 1 is the flow chart of one embodiment of method of the multi-spectral image registration the present invention is based on line match, such as Fig. 1 institutes
Show, the method for the present embodiment can include：
Step 101, the first intersection point for obtaining multigroup intersecting line segment pair in multispectral image subject to registration, and obtain with reference to more
Second intersection point of multigroup intersecting line segment pair in spectrum picture；
Step 102 determines multiple second intersection points with reference to the first intersection point pre-matching on multispectral image；
Step 103 carries out Feature Points Matching to second intersection point and first intersection point, determines described with reference to multispectral
On image with matched second intersection point of first intersection point；
Step 104, according to first intersection point and with matched second intersection point of first intersection point, determine coordinate transform
Parameter；
Step 105, according to the coordinate conversion parameter to the multispectral image subject to registration and it is described refer to multispectral figure
As carrying out registration.
Specifically, at least two width multispectral images are selected when carrying out multi-spectral image registration, a width mostly light as reference
Spectrogram picture, an at least width in addition is as multispectral image subject to registration；The line segment at least two width multispectral images is obtained, then
Multigroup intersecting line segment pair and intersection point are obtained according to the equation of line segment；Intersecting line segment pair in multispectral image subject to registration
Intersection point as the first intersection point, with reference to the intersecting line segment pair in multispectral image intersection point as the second intersection point.
Determine that second intersection point with reference to the first intersection point pre-matching on multispectral image, such as the present invention are implemented
It can be realized in example by the Euclidean distance calculated between the EOH descriptors of the first intersection point and the EOH descriptors of the second intersection point, on
Definite and matched second intersection point of the first intersection point is stated, it need to be to be traveled through with reference to all second intersection points on multispectral image, in advance
Matched second intersection point can have multiple.
Feature Points Matching is carried out to second intersection point and first intersection point, Feature Points Matching mentioned here can be
It is further matched in the line segment where the second intersection point with being taken respectively on the line segment where first intersection point, described second
Line segment where intersection point is consistent with the line segment direction where first intersection point, and direction said herein unanimously refers to that line segment is all
No.1 line segment is all No. two line segments；Wherein, the setting rule of No.1 line segment and No. two line segments is as follows：An if line segment inverse time
Pin overlaps after rotating a sharp angle with another line segment, then the line segment is just denoted as No.1 line segment, and another line segment is just
It is No. two line segments；It can be that multiple points are taken on line segment to carry out line match, and multiple points are matched respectively, finally determine institute
State with reference on multispectral image with matched second intersection point of first intersection point.
According to first intersection point and with matched second intersection point of first intersection point, that is, determine the matching of multigroup intersection point
It is right, the coordinate conversion parameter of multispectral image subject to registration is determined according to the coordinate relation of multigroup intersection point matching between；According to
The coordinate conversion parameter carries out registration to the multispectral image subject to registration and the reference multispectral image, i.e., according to
The multispectral image subject to registration is carried out coordinate transform by coordinate conversion parameter, realizes the multispectral image subject to registration and described
With reference to multi-spectral image registration.
Method for registering images provided in this embodiment based on line match, by obtaining multispectral image subject to registration and ginseng
The intersection point of multigroup intersecting line segment pair in multispectral image is examined, pre-matching is carried out using the intersection point of line segment as characteristic point, and will be pre-
The second intersection point and the first intersection point matched further matched, and finally determines matched intersection point pair, and according to determining
Matched intersection point to determining coordinate conversion parameter, finally according to the coordinate conversion parameter to the multispectral image subject to registration and
It is described to carry out registration with reference to multispectral image, compared with prior art, since line segment feature has more preferably in multispectral image
Stability, therefore the intersection point of line segment is selected to carry out the accuracy that matching improves multi-spectral image registration as characteristic point.
Below using several specific embodiments, the technical solution of embodiment of the method shown in Fig. 1 is described in detail.
On the basis of embodiment shown in Fig. 1, further, obtain multigroup intersecting in multispectral image subject to registration
Before second intersection point of line segment pair, it is necessary first to get intersecting line segment, and in practical applications, obtain intersecting line segment
Mode can there are many, it is optionally, described to obtain multigroup phase in multispectral image subject to registration as a kind of enforceable mode
Before first intersection point of the line segment pair of friendship, including：
Determine the marginal point of the multispectral image subject to registration；
The marginal point for being spaced pre-determined distance is connected to form line segment；
The angle of adjacent segments is calculated, if the angle is less than default angle threshold, retains the adjacent segments；If
More than or equal to the default angle threshold, then the intersection point of the adjacent segments is deleted, connects the another of the adjacent segments
A endpoint forms new line segment；
Multigroup intersecting line segment pair is obtained according to the line segment.
Specifically, the marginal point of the multispectral image subject to registration can be determined according to canny operators；Interval is default
The marginal point of distance d1 connects to form line segment；The angle of adjacent segments is calculated, if the angle is less than default angle threshold
Value, then retain the adjacent segments；If more than or equal to the default angle threshold, then the friendship of the adjacent segments is deleted
Point, another endpoint for connecting the adjacent segments respectively form new line segment；Further these line segments can also be carried out
Screening, i.e., carry out fitting a straight line on every line segment of formation, if error is less than set threshold value and the line segment length of the fitting is big
In a certain value, such as 10 pixels, then the line segment matching line segment obtained as needs abandons the line segment if being unsatisfactory for；May be used also
Further to be screened to these line segments, the line segment of all acquisitions is summarized two-by-two, if on same straight line
Merge this two lines section as a line segment, completed until merging, i.e., finally get required line segment；According to the line segment
Equation can get multigroup intersecting line segment pair.
On the basis of embodiment shown in Fig. 1, further, obtain with reference to multigroup intersecting line in multispectral image
Before second intersection point of section pair, it is necessary first to get line segment, and in practical applications, obtain the mode of line segment can have it is more
Kind, it is optionally, described to obtain second with reference to multigroup intersecting line segment pair in multispectral image as a kind of enforceable mode
Before intersection point, including：
Determine the marginal point with reference to multispectral image；
The marginal point for being spaced pre-determined distance is connected to form line segment；
The angle of adjacent segments is calculated, if the angle is less than default angle threshold, retains the adjacent segments；If
More than or equal to the default angle threshold, then the intersection point of the adjacent segments is deleted, connects the another of the adjacent segments
A endpoint forms new line segment；
Multigroup intersecting line segment pair is obtained according to the line segment.
Specifically, obtain with reference to multigroup intersecting line segment in multispectral image to above-mentioned identical method may be employed,
Details are not described herein again.
On the basis of above-described embodiment, further, in practical applications, determine that the mode of the intersection point of pre-matching can be with
There are many, optionally, as a kind of enforceable mode, it is described determine it is described with reference on multispectral image with first intersection point
Multiple second intersection points of pre-matching, including：
Respectively to first intersection point and the second intersection point calculation edge orientation histogram EOH descriptors；
Calculate the Euclidean distance between the EOH descriptors of first intersection point and the EOH descriptors of second intersection point；
If the Euclidean distance is less than default first distance threshold, it is determined that second intersection point is to be handed over described first
Second intersection point of point pre-matching.
Specifically, due to it needs to be determined that described hand over reference on multispectral image with the second of the first intersection point pre-matching
Point can determine the first intersection point of pre-matching in the embodiment of the present invention according to the Euclidean distance between the EOH descriptors of intersection point
With the second intersection point, therefore the EOH descriptors of the first intersection point and the second intersection point need to be calculated before pre-matching.
Calculating EOH descriptors can be achieved by the steps of：
The first intersection point is used below for example, the pixel region of the N*N neighborhoods of the first intersection point is divided into 16 pieces first,
Calculate each piece of edge orientation histogram.Each edge orientation histogram records corresponding blocks inside gradient directional spreding information.
Each edge orientation histogram is divided into 5 sub-regions (bins), wherein each representing different directions there are four bins, (this four
The number of the point of the Grad maximum in a direction is more than the number of the point of the Grad maximum in remaining direction), a remaining bin
Represent no direction.The numerical value of each bin is exactly the number of the point of the Grad maximum of the direction in corresponding blocks, is not had directive
The numerical value of that bin is exactly the number of the unconspicuous point in corresponding blocks inside gradient direction (i.e. except remaining side of aforementioned four direction
To point).In this way, each first intersection point just can use the vector description of one 80 dimension, which is exactly the EOH descriptions of the first intersection point
Symbol, wherein N are integer more than 0, such as 80 or 100.
After calculating EOH descriptors, calculate between the EOH descriptors of the first intersection point and the EOH descriptors of the second intersection point
Euclidean distance；If judge the Euclidean distance be less than default first distance threshold, it is determined that second intersection point be with
Second intersection point of the first intersection point pre-matching, if less than the first distance threshold the second intersection point have it is multiple, can select appoint
One or the second intersection point of selection Euclidean distance minimum.
On the basis of above-described embodiment, further, characteristic point is carried out to second intersection point and first intersection point
Before matching, including：
On the line segment where second intersection point, it is spaced pre-determined distance successively in the both sides of second intersection point and chooses the
One characteristic point；
On the line segment where first intersection point, the pre-determined distance choosing is spaced successively in the both sides of first intersection point
Take second feature point；Line segment where second intersection point is consistent with the line segment direction where first intersection point.
Specifically, on the line segment where second intersection point, and it is spaced successively in the both sides of second intersection point
Pre-determined distance chooses fisrt feature point, and is similarly operated on the line segment where first intersection point.
On the basis of above-described embodiment, further, in practical applications, determine that the mode of the intersection point of pre-matching can be with
There are many, it is optionally, described to be spaced pre-determined distance choosing successively in the both sides of second intersection point as a kind of enforceable mode
Fisrt feature point is taken, is specifically included：
It is the fisrt feature point that the intersection point on line segment is chosen in the both sides of second intersection point；
It is described to be spaced the pre-determined distance selection second feature point successively in the both sides of first intersection point, it specifically includes：
It is the second feature point that the intersection point on line segment is chosen in the both sides of first intersection point.
Specifically, the intersection point of the above-mentioned point taken on the line segment where intersection point and the line segment and other line segments.
On the basis of above-described embodiment, further, in practical applications, determine that the mode of the intersection point of pre-matching can be with
There are many, optionally, as a kind of enforceable mode, characteristic point is carried out to second intersection point and first intersection point
Match somebody with somebody, determine it is described with reference on multispectral image with matched second intersection point of first intersection point, including：
EOH descriptors are calculated to the fisrt feature point and second feature point respectively；
Calculate Euclidean between the EOH descriptors of the fisrt feature point and the EOH descriptors of second feature point away from
From；The fisrt feature point is corresponding with the position of the second feature point；
If the Euclidean distance for having more than default amount threshold is less than default second distance threshold value, it is determined that described the
Two intersection points be and matched second intersection point of first intersection point；The second distance threshold value is less than first distance threshold.
Specifically, for the fisrt feature point of the line segment where each second intersection point, fisrt feature is all calculated respectively
Euclidean distance between the EOH descriptors of the second feature point of point and correspondence position；
If for the line segment where a certain second intersection point, the Euclidean distance for having more than default amount threshold is less than in advance
If second distance threshold value, it is determined that second intersection point on the line segment is and matched second intersection point of first intersection point.
If there is the line segment where multiple second intersection points, have more than default amount threshold the Euclidean distance be less than default second away from
From threshold value, then the line segment that Euclidean distance is most less than the quantity of default second distance threshold value can be selected, corresponding second hands over
Point is used as matched second intersection point.
In above-mentioned specific embodiment, as where thus according to the line segment where multiple second intersection points and first intersection point
Line segment matched, when calculating, make use of the descriptor compared with multiple spot, and the matching result obtained when being matched is more accurate.
On the basis of above-described embodiment, further, in practical applications, determine that the mode of coordinate conversion parameter can be with
There are many, optionally, as a kind of enforceable mode, according to first intersection point and with first intersection point matched
Two intersection points determine coordinate conversion parameter, following manner specifically may be employed：
Transformation matrix of coordinates H is determined according to formula below (1), using transformation matrix of coordinates H as coordinate conversion parameter；
Wherein,For the coordinate of second intersection point；For the coordinate of first intersection point；Transformation matrix of coordinates
Specifically, three groups of matched first intersection points and the second intersection point may finally at least be determined, coordinate is substituted into
The parameter that transformation matrix of coordinates H is calculated in formula (1) is stated, determines the transformation matrix of coordinates.
On the basis of above-described embodiment, further, in practical applications, as a kind of enforceable mode, according to
The coordinate conversion parameter carries out registration to the multispectral image subject to registration and the reference multispectral image, can specifically adopt
Use following manner：
The coordinate of pixel in the multispectral image subject to registration is multiplied with the transformation matrix of coordinates so that described
Multispectral image subject to registration transform to under the coordinate system identical with reference to multispectral image.
Specifically, by the coordinate of the pixel in the multispectral image subject to registration and the transformation matrix of coordinates phase
Multiply, the multispectral image subject to registration after being converted, the multispectral image subject to registration after conversion is with reference multispectral image in phase
Under same coordinate system.
In above-mentioned specific embodiment, according to first intersection point and with matched second intersection point of first intersection point,
It determines transformation matrix of coordinates, finally realizes and the multispectral image subject to registration and the multispectral image that refers to are matched somebody with somebody
It is accurate.
Fig. 2 is the structure diagram of one embodiment of device of the multi-spectral image registration the present invention is based on line match, such as
Shown in Fig. 2, the device of the present embodiment can include：Acquisition module 201, determining module 202 and processing module 203；
Wherein, acquisition module 201, for obtaining first of multigroup intersecting line segment pair in multispectral image subject to registration the friendship
Point, and obtain the second intersection point with reference to multigroup intersecting line segment pair in multispectral image；
Determining module 202, for determine it is described with reference on multispectral image with the first intersection point pre-matching multiple
Two intersection points；
The determining module 202 is additionally operable to carry out Feature Points Matching to second intersection point and first intersection point, determine
It is described with reference on multispectral image with matched second intersection point of first intersection point；
The determining module 202 is additionally operable to be handed over according to first intersection point and matched with first intersection point second
Point, determines coordinate conversion parameter；
Processing module 203, for according to the coordinate conversion parameter to the multispectral image subject to registration and the reference
Multispectral image carries out registration.
Optionally, acquisition module 201 are specifically used for：
Determine the marginal point of the multispectral image subject to registration；
The marginal point for being spaced pre-determined distance is connected to form line segment；
The angle of adjacent segments is calculated, if the angle is less than default angle threshold, retains the adjacent segments；If
More than or equal to the default angle threshold, then the intersection point of the adjacent segments is deleted, connects the another of the adjacent segments
A endpoint forms new line segment；
Multigroup intersecting line segment pair is obtained according to the line segment.
Optionally, acquisition module 201 are specifically used for：
Determine the marginal point with reference to multispectral image；
The marginal point for being spaced pre-determined distance is connected to form line segment；
The angle of adjacent segments is calculated, if the angle is less than default angle threshold, retains the adjacent segments；If
More than or equal to the default angle threshold, then the intersection point of the adjacent segments is deleted, connects the another of the adjacent segments
A endpoint forms new line segment；
Multigroup intersecting line segment pair is obtained according to the line segment.
Optionally, the determining module 202, is specifically used for：
Respectively to first intersection point and the second intersection point calculation edge orientation histogram EOH descriptors；
Calculate the Euclidean distance between the EOH descriptors of first intersection point and the EOH descriptors of second intersection point；
If the Euclidean distance is less than default first distance threshold, it is determined that second intersection point is to be handed over described first
Second intersection point of point pre-matching.
Optionally, the processing module 203, is additionally operable to：
On the line segment where second intersection point, it is spaced pre-determined distance successively in the both sides of second intersection point and chooses the
One characteristic point；
On the line segment where first intersection point, the pre-determined distance choosing is spaced successively in the both sides of first intersection point
Take second feature point；Line segment where second intersection point is consistent with the line segment direction where first intersection point.
Optionally, the processing module 203, also particularly useful for：
It is the fisrt feature point that the intersection point on line segment is chosen in the both sides of second intersection point；
It is described to be spaced the pre-determined distance selection second feature point successively in the both sides of first intersection point, it specifically includes：
It is the second feature point that the intersection point on line segment is chosen in the both sides of first intersection point.
Optionally, the determining module 202, is specifically used for：
EOH descriptors are calculated to the fisrt feature point and second feature point respectively；
Calculate Euclidean between the EOH descriptors of the fisrt feature point and the EOH descriptors of second feature point away from
From；The fisrt feature point is corresponding with the position of the second feature point；
If the Euclidean distance for having more than default amount threshold is less than default second distance threshold value, it is determined that described the
Two intersection points be and matched second intersection point of first intersection point；The second distance threshold value is less than first distance threshold.
Optionally, the determining module 202, is specifically used for：
Transformation matrix of coordinates H is determined according to formula below (1), is joined the transformation matrix of coordinates H as coordinate transform
Number；
Wherein,For the coordinate of second intersection point；For the coordinate of first intersection point；Transformation matrix of coordinates
Optionally, the processing module 203, is specifically used for：
The coordinate of pixel in the multispectral image subject to registration is multiplied with the transformation matrix of coordinates so that described
Multispectral image subject to registration transform to under the coordinate system identical with reference to multispectral image.
The device of the present embodiment can be used for the technical solution for performing embodiment of the method shown in Fig. 1, realization principle class
Seemingly, details are not described herein again.
One of ordinary skill in the art will appreciate that：Realizing all or part of step of above-mentioned each method embodiment can lead to
The relevant hardware of program instruction is crossed to complete.Foregoing program can be stored in a computer read/write memory medium.The journey
Sequence upon execution, execution the step of including above-mentioned each method embodiment；And foregoing storage medium includes：ROM, RAM, magnetic disc or
The various media that can store program code such as person's CD.
Finally it should be noted that：The above embodiments are only used to illustrate the technical solution of the present invention., rather than its limitations；To the greatest extent
Pipe is described in detail the present invention with reference to foregoing embodiments, it will be understood by those of ordinary skill in the art that：Its according to
Can so modify to the technical solution recorded in foregoing embodiments either to which part or all technical characteristic into
Row equivalent substitution；And these modifications or replacement, the essence of appropriate technical solution is not made to depart from various embodiments of the present invention technology
The scope of scheme.
Claims (6)
- A kind of 1. method of the multi-spectral image registration based on line match, which is characterized in that including：The first intersection point of multigroup intersecting line segment pair in multispectral image subject to registration is obtained, and is obtained with reference to more in multispectral image Second intersection point of the intersecting line segment pair of group；Determine multiple second intersection points with reference to the first intersection point pre-matching on multispectral image；Feature Points Matching is carried out to multiple second intersection points and first intersection point of the pre-matching, is determined described with reference to multispectral On image with matched second intersection point of first intersection point；According to first intersection point and with first intersection point matched Two intersection points, determine coordinate conversion parameter；Registration is carried out to the multispectral image subject to registration and the reference multispectral image according to the coordinate conversion parameter；It is described to determine multiple second intersection points with reference to the first intersection point pre-matching on multispectral image, including：Respectively to first intersection point and the second intersection point calculation edge orientation histogram EOH descriptors；Calculate the Euclidean distance between the EOH descriptors of first intersection point and the EOH descriptors of second intersection point；If the Euclidean distance is less than default first distance threshold, it is determined that second intersection point is pre- with first intersection point Matched second intersection point；Before multiple second intersection points to the pre-matching carry out Feature Points Matching with first intersection point, including：On the line segment where second intersection point, it is spaced pre-determined distance successively in the both sides of second intersection point and chooses the first spy Sign point；On the line segment where first intersection point, it is spaced the pre-determined distance successively in the both sides of first intersection point and chooses the Two characteristic points；Line segment where second intersection point is consistent with the line segment direction where first intersection point；The pre-determined distance that is spaced successively in the both sides of second intersection point chooses fisrt feature point, specifically includes：It is the fisrt feature point that the intersection point on line segment is chosen in the both sides of second intersection point；It is described to be spaced the pre-determined distance selection second feature point successively in the both sides of first intersection point, it specifically includes：It is the second feature point that the intersection point on line segment is chosen in the both sides of first intersection point；Feature Points Matching is carried out to multiple second intersection points and first intersection point of the pre-matching, is determined described with reference to multispectral On image with matched second intersection point of first intersection point, including：EOH descriptors are calculated to the fisrt feature point and second feature point respectively；Calculate the Euclidean distance between the EOH descriptors of the fisrt feature point and the EOH descriptors of second feature point；Institute The position that fisrt feature point is stated with the second feature point is corresponding；If the Euclidean distance for having more than default amount threshold is less than default second distance threshold value, it is determined that described second hands over Point be and matched second intersection point of first intersection point；The second distance threshold value is less than first distance threshold.
- It is 2. according to the method described in claim 1, it is characterized in that, multigroup intersecting in the acquisition multispectral image subject to registration Before first intersection point of line segment pair, including：Determine the marginal point of the multispectral image subject to registration；The marginal point for being spaced pre-determined distance is connected to form line segment；The angle of adjacent segments is calculated, if the angle is less than default angle threshold, retains the adjacent segments；If more than Or equal to the default angle threshold, then the intersection point of the adjacent segments is deleted, connect another end of the adjacent segments Point forms new line segment；Multigroup intersecting line segment pair is obtained according to the line segment.
- 3. according to the method described in claim 1, it is characterized in that, described obtain with reference to multigroup intersecting line in multispectral image Before second intersection point of section pair, including：Determine the marginal point with reference to multispectral image；The marginal point for being spaced pre-determined distance is connected to form line segment；The angle of adjacent segments is calculated, if the angle is less than default angle threshold, retains the adjacent segments；If more than Or equal to the default angle threshold, then the intersection point of the adjacent segments is deleted, connect another end of the adjacent segments Point forms new line segment；Multigroup intersecting line segment pair is obtained according to the line segment.
- 4. according to claim 1-3 any one of them methods, which is characterized in that it is described according to first intersection point and with institute Matched second intersection point of the first intersection point is stated, determines coordinate conversion parameter, including：Transformation matrix of coordinates H is determined according to formula below (1), using the transformation matrix of coordinates H as coordinate conversion parameter；<mrow> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <msup> <mi>x</mi> <mo>&prime;</mo> </msup> </mtd> </mtr> <mtr> <mtd> <msup> <mi>y</mi> <mo>&prime;</mo> </msup> </mtd> </mtr> <mtr> <mtd> <mn>1</mn> </mtd> </mtr> </mtable> </mfenced> <mo>=</mo> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <msub> <mi>a</mi> <mn>11</mn> </msub> </mtd> <mtd> <msub> <mi>a</mi> <mn>12</mn> </msub> </mtd> <mtd> <msub> <mi>a</mi> <mn>13</mn> </msub> </mtd> </mtr> <mtr> <mtd> <msub> <mi>a</mi> <mn>21</mn> </msub> </mtd> <mtd> <msub> <mi>a</mi> <mn>22</mn> </msub> </mtd> <mtd> <msub> <mi>a</mi> <mn>23</mn> </msub> </mtd> </mtr> <mtr> <mtd> <mn>0</mn> </mtd> <mtd> <mn>0</mn> </mtd> <mtd> <mn>1</mn> </mtd> </mtr> </mtable> </mfenced> <mo>&times;</mo> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <mi>x</mi> </mtd> </mtr> <mtr> <mtd> <mi>y</mi> </mtd> </mtr> <mtr> <mtd> <mn>1</mn> </mtd> </mtr> </mtable> </mfenced> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow>Wherein,For the coordinate of second intersection point；For the coordinate of first intersection point；Transformation matrix of coordinates
- 5. according to the method described in claim 4, it is characterized in that, it is described according to the coordinate conversion parameter to described subject to registration Multispectral image and the reference multispectral image carry out registration, including：The coordinate of pixel in the multispectral image subject to registration is multiplied with the transformation matrix of coordinates so that described to wait to match somebody with somebody Quasi- multispectral image transform to under the coordinate system identical with reference to multispectral image.
- 6. a kind of device of the multi-spectral image registration based on line match, which is characterized in that including：Acquisition module for obtaining the first intersection point of multigroup intersecting line segment pair in multispectral image subject to registration, and obtains reference Second intersection point of multigroup intersecting line segment pair in multispectral image；Determining module, for determining multiple second intersection points with reference to the first intersection point pre-matching on multispectral image；The determining module is additionally operable to carry out characteristic point with first intersection point to multiple second intersection points of the pre-matching Match somebody with somebody, determine it is described with reference on multispectral image with matched second intersection point of first intersection point；The determining module, be additionally operable to according to first intersection point and with matched second intersection point of first intersection point, determine Coordinate conversion parameter；Processing module, for according to the coordinate conversion parameter to the multispectral image subject to registration and it is described refer to multispectral figure As carrying out registration；It is described to determine multiple second intersection points with reference to the first intersection point pre-matching on multispectral image, including：Respectively to first intersection point and the second intersection point calculation edge orientation histogram EOH descriptors；Calculate the Euclidean distance between the EOH descriptors of first intersection point and the EOH descriptors of second intersection point；If the Euclidean distance is less than default first distance threshold, it is determined that second intersection point is pre- with first intersection point Matched second intersection point；The processing module, is additionally operable to：Before multiple second intersection points of the pre-matching carry out Feature Points Matching with first intersection point,On the line segment where second intersection point, it is spaced pre-determined distance successively in the both sides of second intersection point and chooses the first spy Sign point；On the line segment where first intersection point, it is spaced the pre-determined distance successively in the both sides of first intersection point and chooses the Two characteristic points；Line segment where second intersection point is consistent with the line segment direction where first intersection point；The pre-determined distance that is spaced successively in the both sides of second intersection point chooses fisrt feature point, specifically includes：It is the fisrt feature point that the intersection point on line segment is chosen in the both sides of second intersection point；It is described to be spaced the pre-determined distance selection second feature point successively in the both sides of first intersection point, it specifically includes：It is the second feature point that the intersection point on line segment is chosen in the both sides of first intersection point；Multiple second intersection points to the pre-matching carry out Feature Points Matching with first intersection point, determine described with reference to more With matched second intersection point of first intersection point on spectrum picture, specifically include：EOH descriptors are calculated to the fisrt feature point and second feature point respectively；Calculate the Euclidean distance between the EOH descriptors of the fisrt feature point and the EOH descriptors of second feature point；Institute The position that fisrt feature point is stated with the second feature point is corresponding；If the Euclidean distance for having more than default amount threshold is less than default second distance threshold value, it is determined that described second hands over Point be and matched second intersection point of first intersection point；The second distance threshold value is less than first distance threshold.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510347566.6A CN104992433B (en) | 2015-06-19 | 2015-06-19 | The method and device of multi-spectral image registration based on line match |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510347566.6A CN104992433B (en) | 2015-06-19 | 2015-06-19 | The method and device of multi-spectral image registration based on line match |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104992433A CN104992433A (en) | 2015-10-21 |
| CN104992433B true CN104992433B (en) | 2018-05-25 |
Family
ID=54304241
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510347566.6A Active CN104992433B (en) | 2015-06-19 | 2015-06-19 | The method and device of multi-spectral image registration based on line match |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104992433B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109389628B (en) * | 2018-09-07 | 2021-03-23 | 北京邮电大学 | Image registration method, apparatus and storage medium |
| CN112204615B (en) * | 2019-01-31 | 2023-05-02 | 深圳华大生命科学研究院 | Fluorescence image registration method, gene sequencer, system and storage medium |
| CN110084236B (en) * | 2019-04-29 | 2021-05-28 | 北京朗镜科技有限责任公司 | Image correction method and device |
| CN113496505B (en) * | 2020-04-03 | 2022-11-08 | 广州极飞科技股份有限公司 | Image registration method and device, multispectral camera, unmanned equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6665440B1 (en) * | 2000-05-30 | 2003-12-16 | Microsoft Corporation | System and method for performing corner guided curve matching of multiple images representing a scene |
| CN101916445A (en) * | 2010-08-25 | 2010-12-15 | 天津大学 | Affine parameter estimation-based image registration method |
| CN103514606A (en) * | 2013-10-14 | 2014-01-15 | 武汉大学 | Heterology remote sensing image registration method |
-
2015
- 2015-06-19 CN CN201510347566.6A patent/CN104992433B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6665440B1 (en) * | 2000-05-30 | 2003-12-16 | Microsoft Corporation | System and method for performing corner guided curve matching of multiple images representing a scene |
| CN101916445A (en) * | 2010-08-25 | 2010-12-15 | 天津大学 | Affine parameter estimation-based image registration method |
| CN103514606A (en) * | 2013-10-14 | 2014-01-15 | 武汉大学 | Heterology remote sensing image registration method |
Non-Patent Citations (1)
| Title |
|---|
| 胡伏原 等.印刷品缺陷检测系统的快速配准方法研究.《苏州科技学院学报(工程技术版)》.2013,第26卷(第3期),76-80. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104992433A (en) | 2015-10-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hu et al. | Revisiting single image depth estimation: Toward higher resolution maps with accurate object boundaries | |
| CN105957007B (en) | Image split-joint method based on characteristic point plane similarity | |
| EP2491529B1 (en) | Providing a descriptor for at least one feature of an image | |
| WO2016062159A1 (en) | Image matching method and platform for testing of mobile phone applications | |
| CN104599258B (en) | A kind of image split-joint method based on anisotropic character descriptor | |
| CN104992433B (en) | The method and device of multi-spectral image registration based on line match | |
| CN105427298A (en) | Remote sensing image registration method based on anisotropic gradient dimension space | |
| CN104992431B (en) | The method and device of multi-spectral image registration | |
| US20020181780A1 (en) | Geometic hashing method for model-based recognition of an object | |
| CN104616247B (en) | A kind of method for map splicing of being taken photo by plane based on super-pixel SIFT | |
| CN109308715A (en) | A kind of optical imagery method for registering combined based on point feature and line feature | |
| CN108122280A (en) | The method for reconstructing and device of a kind of three-dimensional point cloud | |
| CN104050675A (en) | Feature point matching method based on triangle description | |
| CN108759788A (en) | Unmanned plane image positioning and orientation method and unmanned plane | |
| JP2017003525A (en) | Three-dimensional measuring device | |
| CN107909018B (en) | Stable multi-mode remote sensing image matching method and system | |
| Wang et al. | Robust line feature matching based on pair-wise geometric constraints and matching redundancy | |
| CN103995860A (en) | Image retrieval system and method based on color feature extraction | |
| CN107133260A (en) | The matching and recognition method and device of a kind of landmark image | |
| Chen et al. | An intelligent vision recognition method based on deep learning for pointer meters | |
| CN109711375A (en) | A kind of recognition methods of signal lamp and device | |
| CN110175975A (en) | Method for checking object, device, computer readable storage medium and computer equipment | |
| CN109087344A (en) | Image-selecting method and device in three-dimensional reconstruction | |
| CN108447092A (en) | The method and device of vision positioning marker | |
| Aktar et al. | Robust mosaicking of maize fields from aerial imagery |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |