CN105654479A - Multispectral image registering method and multispectral image registering device - Google Patents
Multispectral image registering method and multispectral image registering device Download PDFInfo
- Publication number
- CN105654479A CN105654479A CN201511000876.7A CN201511000876A CN105654479A CN 105654479 A CN105654479 A CN 105654479A CN 201511000876 A CN201511000876 A CN 201511000876A CN 105654479 A CN105654479 A CN 105654479A
- Authority
- CN
- China
- Prior art keywords
- registration
- reference picture
- unique point
- image subject
- point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Image Analysis (AREA)
Abstract
The invention provides a multispectral image registering method and a multispectral image registering device, wherein the multispectral image registering method comprises the steps of respectively calculating an EOH descriptor which corresponds with each characteristic point according to an acquired reference image and the characteristic point of a to-be-registered image; calculating an Euclidean distance between each characteristic point of the reference image and each characteristic point of the to-be-registered image according to the EOH descriptor; performing matching between each characteristic point in the reference image and all characteristic points in the to-be-registered image and matching between each characteristic point in the to-be-registered image and all characteristic points in the reference image according to the Euclidean distance, thereby obtaining characteristic point matching pairs between the reference image and the to-be-registered image; and performing image registering on the reference image and the to-be-registered image according to the characteristic point matching pairs. According to the multispectral image registering method of the invention, image registering accuracy is improved through bidirectional matching of characteristic points in the reference image and the to-be-registered image.
Description
Technical field
The present invention relates to image processing field, particularly relate to a kind of multi-spectrum image registration method and device.
Background technology
Two width that (weather, illumination, shooting position and angle etc.) under different time, different sensors (imaging equipment) or different condition are obtained by image registration exactly or several images carry out mating, the process of superposition. Wherein, the registration of multispectral image obtains to be applied more and more widely, the registration of so-called multispectral image, refers to that several the multispectral images to generating by the sensor (imaging equipment) of different frequency range carry out registration, such as: visible images and infrared light image are carried out registration.
At present, the registration Algorithm of distinguished point based is method the most common in image registration, and algorithm flow is as follows: first, several images carries out feature extraction and obtains unique point; Then, by similarity measurement, find the feature point pairs of coupling in several images; Finally, image registration is carried out by the feature point pairs of coupling. Wherein, for the coupling of unique point, usually unidirectional coupling is adopted, concrete, define two width images and it is respectively reference picture and image subject to registration, calculating the Europe formula distance between any two of all unique points in two width images, for each the unique point T in image subject to registration, Europe formula is S apart from minimum unique point in reference picture and between unique point T1, formula distance in Europe is D1, the unique point that formula distance time in Europe is little in reference picture and between unique point T is S2, formula distance in Europe is D2If, D1<����D2, 0 < �� < 1, then unique point S1Being and the unique point that unique point T-phase is mated, constitutive characteristic point is to (S1, T).
But, in multi-mode image, owing to imaging equipment is different, between multi-mode image, its texture structure difference and pixel difference are all very big, and the unique point detected out is also unstable, by the unidirectional coupling of unique point, and the unique point S in reference picture1For the Optimum Matching point of unique point T in image subject to registration, but conversely, the unique point T in image subject to registration might not be unique point S in reference picture1Optimum Matching point, the Feature Points Matching obtained like this is to (S1, T) and will be inaccurate, and then, based on the lower Feature Points Matching of accuracy to carrying out image registration, reduce the accuracy of image registration.
Summary of the invention
The present invention provides a kind of multi-spectrum image registration method and device, by reference to the bi-directional matching of unique point in image and image subject to registration, it is to increase the accuracy of Feature Points Matching pair, and then improves the accuracy of image registration.
Multi-spectrum image registration method provided by the invention, comprising:
Determine reference picture and the unique point of image subject to registration respectively;
The edge orientation histogram EOH descriptor of each Feature point correspondence in described reference picture and described image subject to registration is calculated respectively according to the unique point of described reference picture and described image subject to registration;
Europe formula distance between each unique point in each unique point and described image subject to registration in reference picture according to described EOH descriptor computation;
Carry out mating and the mating of each unique point and all unique points in described reference picture in described image subject to registration of each unique point and all unique points in described image subject to registration in described reference picture according to described Europe formula distance, obtain the Feature Points Matching pair between described reference picture and described image subject to registration;
Image registration is carried out to by described reference picture and described image subject to registration according to described Feature Points Matching.
Multi-spectral image registration device provided by the invention, comprising:
Unique point determination module, for determining reference picture and the unique point of image subject to registration respectively;
Descriptor generation module, for calculating the edge orientation histogram EOH descriptor of each Feature point correspondence in described reference picture and described image subject to registration respectively according to the unique point of described reference picture and described image subject to registration;
Calculate module, for each unique point in reference picture according to described EOH descriptor computation and Europe formula distance between each unique point in described image subject to registration;
First matching module, for carrying out mating and the mating of each unique point and all unique points in described reference picture in described image subject to registration of each unique point and all unique points in described image subject to registration in described reference picture according to described Europe formula distance, obtain the Feature Points Matching pair between described reference picture and described image subject to registration;
2nd matching module, for carrying out image registration according to described Feature Points Matching to by described reference picture and described image subject to registration.
The present invention provides a kind of multi-spectrum image registration method and device, wherein, multi-spectrum image registration method comprises: determine reference picture and the unique point of image subject to registration respectively, respectively according to the EOH descriptor of each Feature point correspondence in the unique point computing reference image of reference picture and image subject to registration and image subject to registration, according to each unique point in EOH descriptor computation reference picture and Europe formula distance between each unique point in image subject to registration, mating and the mating of each unique point and unique points all in reference picture in image subject to registration of each unique point and all unique points in image subject to registration in reference picture is carried out according to Europe formula distance, obtain the Feature Points Matching pair between reference picture and image subject to registration, image registration is carried out to reference to image and image subject to registration according to Feature Points Matching. multi-spectrum image registration method provided by the invention, Feature Points Matching pair is obtained by reference to bi-directional matching between unique point in image and image subject to registration, ensure that for the matching relationship reference picture and image subject to registration being all qualified optimum, improve the accuracy of Feature Points Matching pair, and then improve the accuracy of image registration.
Accompanying drawing explanation
In order to be illustrated more clearly in the embodiment of the present invention or technical scheme of the prior art, it is briefly described to the accompanying drawing used required in embodiment or description of the prior art below, apparently, accompanying drawing in the following describes is some embodiments of the present invention, for those of ordinary skill in the art, under the prerequisite not paying creative work, it is also possible to obtain other accompanying drawing according to these accompanying drawings.
The schema of the multi-spectrum image registration method that Fig. 1 provides for the embodiment of the present invention one;
The schema of the multi-spectrum image registration method that Fig. 2 provides for the embodiment of the present invention two;
The structural representation of the multi-spectral image registration device that Fig. 3 provides for the embodiment of the present invention one.
Embodiment
For making the object of the embodiment of the present invention, technical scheme and advantage clearly, below in conjunction with the accompanying drawing in the embodiment of the present invention, technical scheme in the embodiment of the present invention is clearly and completely described, obviously, described embodiment is the present invention's part embodiment, instead of whole embodiments. Based on the embodiment in the present invention, those of ordinary skill in the art, not making other embodiments all obtained under creative work prerequisite, belong to the scope of protection of the invention.
The schema of the multi-spectrum image registration method that Fig. 1 provides for the embodiment of the present invention one. As shown in Figure 1, multi-spectrum image registration method provided by the invention, it is possible to comprising:
Step 101, the unique point determining reference picture and image subject to registration respectively.
So-called unique point, refers to the pixel with certain pattern feature or regional area that not easily change in image, also referred to as point of interest, significant point or key point. Unique point is the description that image content is the most abstract, and image noise, grey scale change, image deformation and blocking etc. is had good adaptive faculty.
The unique point obtaining image can have multiple mode, and the present embodiment is not limited.
Optionally, a kind of implementation of this step can be:
Scale invariant feature conversion (Scale-invariantfeaturetransform is called for short SIFT) algorithm is adopted to obtain reference picture and the unique point of image subject to registration respectively.
The unique point obtained by SIFT algorithm is the unique point based on local appearance, and the tolerance changed in scaling, rotation, brightness flop, noise, visual angle is quite high, and unique point highly significant and easily obtaining, is conducive to the stability of image registration.
Optionally, another kind of implementation of this step can be:
Determine reference picture and the Eigenvector of image subject to registration respectively.
Extend the Eigenvector of reference picture and image subject to registration respectively, with reference to the intersection point of extended line of Eigenvector in image as the unique point with reference to image, and using the unique point of the intersection point of the extended line of Eigenvector in image subject to registration as image subject to registration.
Owing to, in multi-mode image, pixel and texture structure between image differ greatly, and Eigenvector will be more stable compared to independent unique point, so, the unique point obtained by Eigenvector will be more accurate, be conducive to the accuracy of image registration.
Step 103, respectively according to the EOH descriptor of each Feature point correspondence in the unique point computing reference image of reference picture and image subject to registration and image subject to registration.
So-called descriptor, refers to the feature description that unique point is set up, also referred to as descriptor. Desirable descriptor to be met yardstick, rotation, the conversion such as even affine had certain unchangeability, and descriptor dependency corresponding to different characteristics point is little, could effectively distinguish different unique points like this.
In this step, edge orientation histogram (EdgeOrientationHistogram has been calculated by the unique point of image, it is called for short EOH) descriptor, for multispectral image, relative to the overall region in image, region, edge in image is more stable for multispectral image, and EOH descriptor have employed region, edge comparatively stable in image, is conducive to the accuracy of multi-spectral image registration.
Step 105, according to each unique point in EOH descriptor computation reference picture and Europe formula distance between each unique point in image subject to registration.
In this step, it is necessary to travel through unique points all in reference picture and image subject to registration, the Europe formula distance of a unique point in computing reference image and a unique point in image subject to registration.
Such as, having 3 unique points in reference picture, image subject to registration has 2 unique points, then by the calculating of this step, obtain the Europe formula distance between 3 �� 2=6 unique point altogether.
Step 107, mating and the mating of each unique point and unique points all in reference picture in image subject to registration of carrying out each unique point and all unique points in image subject to registration in reference picture according to Europe formula distance, obtain the Feature Points Matching pair between reference picture and image subject to registration.
In this step, the bi-directional matching of unique point in reference picture and image subject to registration will be carried out by the Europe formula distance between unique point, namely, carrying out in reference picture in each unique point and image subject to registration while Feature Points Matching, also carrying out mating of each unique point and unique point in reference picture in image subject to registration.
In the prior art, obtain Feature Points Matching pair by the unidirectional coupling of unique point, by the unidirectional coupling of unique point, the unique point S in reference picture1For the Optimum Matching point of unique point T in image subject to registration, but conversely, the unique point T in image subject to registration might not be unique point S in reference picture1Optimum Matching point. And in this step, by the bi-directional matching between unique point, take into account unique point in reference picture to mate with the mutual of unique point in image subject to registration simultaneously, can ensure that the Feature Points Matching obtained after mating is concerning the matching relationship being all qualified optimum for reference picture and image both sides subject to registration, improve the accuracy of Feature Points Matching pair, and then improve the accuracy of image registration.
Step 109, carry out image registration according to Feature Points Matching to reference to image and image subject to registration.
Owing to obtaining Feature Points Matching pair accurately by the bi-directional matching of unique point, so, according to this accurately Feature Points Matching Root of Nepal Rattlesnake Plantain examine image and image subject to registration mates, it is possible to promote the accuracy of registration between reference picture and image subject to registration.
Present embodiments provide a kind of multi-spectrum image registration method, comprise: determine reference picture and the unique point of image subject to registration respectively, respectively according to the EOH descriptor of each Feature point correspondence in the unique point computing reference image of reference picture and image subject to registration and image subject to registration, according to each unique point in EOH descriptor computation reference picture and Europe formula distance between each unique point in image subject to registration, mating and the mating of each unique point and unique points all in reference picture in image subject to registration of each unique point and all unique points in image subject to registration in reference picture is carried out according to Europe formula distance, obtain the Feature Points Matching pair between reference picture and image subject to registration, image registration is carried out to reference to image and image subject to registration according to Feature Points Matching.The multi-spectrum image registration method that the present embodiment provides, Feature Points Matching pair is obtained by reference to bi-directional matching between unique point in image and image subject to registration, ensure that for the matching relationship reference picture and image subject to registration being all qualified optimum, improve the accuracy of Feature Points Matching pair, and then improve the accuracy of image registration.
The schema of the multi-spectrum image registration method that Fig. 2 provides for the embodiment of the present invention two, the present embodiment is on the basis of embodiment one, it provides a kind of specific implementation of step 107 in embodiment one. As shown in Figure 2, step 107 carries out mating and the mating of each unique point and unique points all in reference picture in image subject to registration of each unique point and all unique points in image subject to registration in reference picture according to Europe formula distance, obtain the Feature Points Matching pair between reference picture and image subject to registration, it is possible to comprising:
Step 201, for each the unique point S in reference picturei, by unique point T all in image subject to registrationjWith unique point SiBetween Europe formula distance d (Si,Tj) sort from small to large, by feature point pairs (S corresponding for front k Europe formula distancei,Tj) as fisrt feature point to collection.
Wherein, SiFor the unique point in reference picture, TjFor the unique point in image subject to registration, 1��k��min (m, n), m is the number of unique point in reference picture, and n is the number of unique point in image subject to registration.
For each the unique point S in reference picturei, 1��i��m, in itself and image subject to registration, the Europe formula distance of all unique points is Di={ d (Si,T1),d(Si,T2),��,d(Si,Tn), to DiSort from small to large, the set C of the feature point pairs composition of k Europe formula distance correspondence before selectingi={ (Si,Tj_1),(Si,Tj_2),��,(Si,Tj_k) be possible with unique point SiThe feature point set mated mutually, wherein, 1��j_k��k, all unique point S in reference pictureiCorresponding feature point set CiConstitute fisrt feature point to collection.
By this step, obtain all with unique point S in reference pictureiThe Feature Points Matching pair mated mutually, in fact completes the characteristic of unidirectional Point matching of reference picture side.
Step 203, for each the unique point T in image subject to registrationj, with reference to unique point S all in imageiWith unique point TjBetween Europe formula distance d (Si,Tj) carry out sequence from small to large, by feature point pairs (S corresponding for front k Europe formula distancei,Tj) as second feature point to collection.
For each the unique point T in image subject to registrationj, 1��j��n, in itself and image subject to registration, the Europe formula distance of all unique points is Dj={ d (S1,Tj),d(S2,Tj),��,d(Sm,Tj), to DjSort from small to large, the set C of the feature point pairs composition of k Europe formula distance correspondence before selectingj={ (Si_1,Tj),(Si_2,Tj),��,(Si_k,Tj) be possible with unique point TjThe feature point set mated mutually, wherein, 1��i_k��k, all unique point T in image subject to registrationjCorresponding feature point set CjConstitute second feature point to collection. ,
By this step, obtain unique point T in all and subject to registration imagesjThe Feature Points Matching pair mated mutually, in fact completes the characteristic of unidirectional Point matching of image side subject to registration.
It should be noted that, step 203 can also before step 201, and the present embodiment is not limited for the order of step 201 and step 203.
Step 205, using fisrt feature point to collection and second feature point to the common factor of collection as with reference to the Feature Points Matching pair between image and image subject to registration.
By step 201 and step 203, the fisrt feature point obtaining unidirectional coupling to collection and second feature point to collection, collection is got common factor by collection and second feature point by fisrt feature point, then take into account unique point in reference picture to mate with the mutual of unique point in image subject to registration simultaneously, can ensure that the Feature Points Matching obtained after mating is concerning the matching relationship being all qualified optimum for reference picture and image both sides subject to registration, improve the accuracy of Feature Points Matching pair, and then improve the accuracy of image registration.
Preferably, as k=1, it is possible to obtain Feature Points Matching pair one to one.
Present embodiments provide a kind of multi-spectrum image registration method, comprising: for each the unique point S in reference picturei, by unique point T all in image subject to registrationjWith unique point SiBetween Europe formula distance d (Si,Tj) sort from small to large, by feature point pairs (S corresponding for front k Europe formula distancei,Tj) as fisrt feature point to collection, for each the unique point T in image subject to registrationj, with reference to unique point S all in imageiWith unique point TjBetween Europe formula distance d (Si,Tj) carry out sequence from small to large, by feature point pairs (S corresponding for front k Europe formula distancei,Tj) as second feature point to collection, using fisrt feature point to collection and second feature point to the common factor of collection as with reference to the Feature Points Matching pair between image and image subject to registration. The multi-spectrum image registration method that the present embodiment provides, Feature Points Matching pair is obtained by reference to bi-directional matching between unique point in image and image subject to registration, ensure that for the matching relationship reference picture and image subject to registration being all qualified optimum, improve the accuracy of Feature Points Matching pair, and then improve the accuracy of image registration.
The structural representation of the multi-spectral image registration device that Fig. 3 provides for the embodiment of the present invention one, the multi-spectral image registration device that the present embodiment provides, in order to perform the multi-spectrum image registration method of Fig. 1��offer embodiment illustrated in fig. 2. The multi-spectral image registration device that the present embodiment provides, it is possible to comprising:
Unique point determination module 11, for determining reference picture and the unique point of image subject to registration respectively.
Descriptor generation module 13, for respectively according to the EOH descriptor of each Feature point correspondence in the unique point computing reference image of reference picture and image subject to registration and image subject to registration.
Calculate module 15, for according to each unique point in EOH descriptor computation reference picture and Europe formula distance between each unique point in image subject to registration.
First matching module 17, for carrying out mating and the mating of each unique point and unique points all in reference picture in image subject to registration of each unique point and all unique points in image subject to registration in reference picture according to Europe formula distance, obtain the Feature Points Matching pair between reference picture and image subject to registration.
2nd matching module 19, for carrying out image registration according to Feature Points Matching to reference to image and image subject to registration.
Present embodiments provide a kind of multi-spectral image registration device, comprise: unique point determination module, for determining reference picture and the unique point of image subject to registration respectively, descriptor generation module, for respectively according to the EOH descriptor of each Feature point correspondence in the unique point computing reference image of reference picture and image subject to registration and image subject to registration, calculate module, for according to each unique point in EOH descriptor computation reference picture and Europe formula distance between each unique point in image subject to registration, first matching module, for carrying out mating and the mating of each unique point and unique points all in reference picture in image subject to registration of each unique point and all unique points in image subject to registration in reference picture according to Europe formula distance, obtain the Feature Points Matching pair between reference picture and image subject to registration, 2nd matching module, for carrying out image registration according to Feature Points Matching to reference to image and image subject to registration. the multi-spectral image registration device that the present embodiment provides, Feature Points Matching pair is obtained by reference to bi-directional matching between unique point in image and image subject to registration, ensure that for the matching relationship reference picture and image subject to registration being all qualified optimum, improve the accuracy of Feature Points Matching pair, and then improve the accuracy of image registration.
Optionally, unique point determination module 11, specifically for:
SIFT algorithm is adopted to obtain reference picture and the unique point of image subject to registration respectively.
Optionally, unique point determination module 11, specifically for:
Determine reference picture and the Eigenvector of image subject to registration respectively;
Extend the Eigenvector of reference picture and image subject to registration respectively, with reference to the intersection point of image and the Eigenvector extended line of image subject to registration as the unique point with reference to image and image subject to registration.
Optionally, the first matching module 17, specifically for:
For each the unique point S in reference picturei, by unique point T all in image subject to registrationjWith unique point SiBetween Europe formula distance d (Si,Tj) carry out sequence from small to large, by feature point pairs (S corresponding for front k Europe formula distancei,Tj) as fisrt feature point to collection.
For each the unique point T in image subject to registrationj, with reference to unique point S all in imageiWith unique point TjBetween Europe formula distance d (Si,Tj) carry out sequence from small to large, by feature point pairs (S corresponding for front k Europe formula distancei,Tj) as second feature point to collection.
Using fisrt feature point to collection and second feature point to the common factor of collection as with reference to the Feature Points Matching pair between image and image subject to registration.
Wherein, SiFor the unique point in reference picture, TjFor the unique point in image subject to registration, 1��k��min (m, n), m is the number of unique point in reference picture, and n is the number of unique point in image subject to registration.
Last it is noted that above each embodiment is only in order to illustrate the technical scheme of the present invention, it is not intended to limit; Although with reference to foregoing embodiments to invention has been detailed description, it will be understood by those within the art that: the technical scheme described in foregoing embodiments still can be modified by it, or wherein some or all of technology feature is carried out equivalent replacement; And these amendments or replacement, do not make the scope of the essence disengaging various embodiments of the present invention technical scheme of appropriate technical solution.
Claims (8)
1. a multi-spectrum image registration method, it is characterised in that, comprising:
Determine reference picture and the unique point of image subject to registration respectively;
The edge orientation histogram EOH descriptor of each Feature point correspondence in described reference picture and described image subject to registration is calculated respectively according to the unique point of described reference picture and described image subject to registration;
Europe formula distance between each unique point in each unique point and described image subject to registration in reference picture according to described EOH descriptor computation;
Carry out mating and the mating of each unique point and all unique points in described reference picture in described image subject to registration of each unique point and all unique points in described image subject to registration in described reference picture according to described Europe formula distance, obtain the Feature Points Matching pair between described reference picture and described image subject to registration;
Image registration is carried out to by described reference picture and described image subject to registration according to described Feature Points Matching.
2. multi-spectrum image registration method according to claim 1, it is characterized in that, described carry out mating and the mating of each unique point and all unique points in described reference picture in described image subject to registration of each unique point and all unique points in described image subject to registration in described reference picture according to described Europe formula distance, obtain the Feature Points Matching pair between described reference picture and described image subject to registration, comprising:
For each the unique point S in described reference picturei, by all unique point T in described image subject to registrationjWith described unique point SiBetween Europe formula distance d (Si,Tj) sort from small to large, by feature point pairs (S corresponding for front k Europe formula distancei,Tj) as fisrt feature point to collection;
For each the unique point T in described image subject to registrationj, by unique point S all in described reference pictureiWith described unique point TjBetween Europe formula distance d (Si,Tj) sort from small to large, by feature point pairs (S corresponding for front k Europe formula distancei,Tj) as second feature point to collection;
Using fisrt feature point to collecting and second feature point occuring simultaneously as the Feature Points Matching pair between described reference picture and described image subject to registration to collection;
Wherein, SiFor the unique point in reference picture, TjFor the unique point in image subject to registration, 1��k��min (m, n), m is the number of unique point in reference picture, and n is the number of unique point in image subject to registration.
3. multi-spectrum image registration method according to claim 1 and 2, it is characterised in that, described determine reference picture and the unique point of image subject to registration respectively, comprising:
Scale invariant feature conversion SIFT algorithm is adopted to obtain described reference picture and the unique point of described image subject to registration respectively.
4. multi-spectrum image registration method according to claim 1 and 2, it is characterised in that, described determine reference picture and the unique point of image subject to registration respectively, comprising:
Determine described reference picture and the Eigenvector of described image subject to registration respectively;
Extend described reference picture and the Eigenvector of described image subject to registration respectively, using the unique point of the intersection point of the extended line of Eigenvector in described reference picture as described reference picture, and using the unique point of the intersection point of the extended line of Eigenvector in described image subject to registration as described image subject to registration.
5. a multi-spectral image registration device, it is characterised in that, comprising:
Unique point determination module, for determining reference picture and the unique point of image subject to registration respectively;
Descriptor generation module, for calculating the edge orientation histogram EOH descriptor of each Feature point correspondence in described reference picture and described image subject to registration respectively according to the unique point of described reference picture and described image subject to registration;
Calculate module, for each unique point in reference picture according to described EOH descriptor computation and Europe formula distance between each unique point in described image subject to registration;
First matching module, for carrying out mating and the mating of each unique point and all unique points in described reference picture in described image subject to registration of each unique point and all unique points in described image subject to registration in described reference picture according to described Europe formula distance, obtain the Feature Points Matching pair between described reference picture and described image subject to registration;
2nd matching module, for carrying out image registration according to described Feature Points Matching to by described reference picture and described image subject to registration.
6. multi-spectral image registration device according to claim 5, it is characterised in that, described first matching module specifically for:
For each the unique point S in described reference picturei, by all unique point T in described image subject to registrationjWith described unique point SiBetween Europe formula distance d (Si,Tj) sort from small to large, by feature point pairs (S corresponding for front k Europe formula distancei,Tj) as fisrt feature point to collection;
For each the unique point T in described image subject to registrationj, by unique point S all in described reference pictureiWith described unique point TjBetween Europe formula distance d (Si,Tj) sort from small to large, by feature point pairs (S corresponding for front k Europe formula distancei,Tj) as second feature point to collection;
Using fisrt feature point to collecting and second feature point occuring simultaneously as the Feature Points Matching pair between described reference picture and described image subject to registration to collection;
Wherein, SiFor the unique point in reference picture, TjFor the unique point in image subject to registration, 1��k��min (m, n), m is the number of unique point in reference picture, and n is the number of unique point in image subject to registration.
7. multi-spectral image registration device according to claim 5 or 6, it is characterised in that, described unique point determination module specifically for:
Scale invariant feature conversion SIFT algorithm is adopted to obtain described reference picture and the unique point of described image subject to registration respectively.
8. multi-spectral image registration device according to claim 5 or 6, it is characterised in that, described unique point determination module specifically for:
Determine described reference picture and the Eigenvector of described image subject to registration respectively;
Extend described reference picture and the Eigenvector of described image subject to registration respectively, using the unique point of the intersection point of the extended line of Eigenvector in described reference picture as described reference picture, and using the unique point of the intersection point of the extended line of Eigenvector in described image subject to registration as described image subject to registration.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201511000876.7A CN105654479A (en) | 2015-12-28 | 2015-12-28 | Multispectral image registering method and multispectral image registering device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201511000876.7A CN105654479A (en) | 2015-12-28 | 2015-12-28 | Multispectral image registering method and multispectral image registering device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105654479A true CN105654479A (en) | 2016-06-08 |
Family
ID=56477759
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201511000876.7A Pending CN105654479A (en) | 2015-12-28 | 2015-12-28 | Multispectral image registering method and multispectral image registering device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105654479A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106952223A (en) * | 2017-03-17 | 2017-07-14 | 北京邮电大学 | Method for registering images and device |
| CN107680112A (en) * | 2017-10-16 | 2018-02-09 | 北京邮电大学 | Method for registering images |
| CN108921212A (en) * | 2018-06-27 | 2018-11-30 | 努比亚技术有限公司 | A kind of image matching method, mobile terminal and computer readable storage medium |
| CN109389628A (en) * | 2018-09-07 | 2019-02-26 | 北京邮电大学 | Method for registering images, equipment and storage medium |
| CN111504981A (en) * | 2020-04-26 | 2020-08-07 | 上海交通大学 | Method for determining chemical components and moisture content in powder material |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103383700A (en) * | 2013-06-26 | 2013-11-06 | 西安电子科技大学 | Image retrieval method based on margin directional error histogram |
| CN104978742A (en) * | 2015-06-12 | 2015-10-14 | 北京邮电大学 | Image registration method and image registration device based on cascade structure |
| CN104992400A (en) * | 2015-06-19 | 2015-10-21 | 北京邮电大学 | Method and device for multispectral image registration |
-
2015
- 2015-12-28 CN CN201511000876.7A patent/CN105654479A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103383700A (en) * | 2013-06-26 | 2013-11-06 | 西安电子科技大学 | Image retrieval method based on margin directional error histogram |
| CN104978742A (en) * | 2015-06-12 | 2015-10-14 | 北京邮电大学 | Image registration method and image registration device based on cascade structure |
| CN104992400A (en) * | 2015-06-19 | 2015-10-21 | 北京邮电大学 | Method and device for multispectral image registration |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106952223A (en) * | 2017-03-17 | 2017-07-14 | 北京邮电大学 | Method for registering images and device |
| CN107680112A (en) * | 2017-10-16 | 2018-02-09 | 北京邮电大学 | Method for registering images |
| CN107680112B (en) * | 2017-10-16 | 2021-01-26 | 北京邮电大学 | Image registration method |
| CN108921212A (en) * | 2018-06-27 | 2018-11-30 | 努比亚技术有限公司 | A kind of image matching method, mobile terminal and computer readable storage medium |
| CN108921212B (en) * | 2018-06-27 | 2021-11-19 | 努比亚技术有限公司 | Image matching method, mobile terminal and computer readable storage medium |
| CN109389628A (en) * | 2018-09-07 | 2019-02-26 | 北京邮电大学 | Method for registering images, equipment and storage medium |
| CN109389628B (en) * | 2018-09-07 | 2021-03-23 | 北京邮电大学 | Image registration method, apparatus and storage medium |
| CN111504981A (en) * | 2020-04-26 | 2020-08-07 | 上海交通大学 | Method for determining chemical components and moisture content in powder material |
| CN111504981B (en) * | 2020-04-26 | 2021-10-22 | 上海交通大学 | Method for determining chemical components and moisture content in powder material |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110443836B (en) | Point cloud data automatic registration method and device based on plane features | |
| CN104574421B (en) | Large-breadth small-overlapping-area high-precision multispectral image registration method and device | |
| CN107993258B (en) | Image registration method and device | |
| CN108921895B (en) | Sensor relative pose estimation method | |
| CN104751465A (en) | ORB (oriented brief) image feature registration method based on LK (Lucas-Kanade) optical flow constraint | |
| CN104200461B (en) | The remote sensing image registration method of block and sift features is selected based on mutual information image | |
| CN103337077B (en) | A kind of based on multi-scale division and the visible ray of SIFT and the method for infrared image registration | |
| CN105654479A (en) | Multispectral image registering method and multispectral image registering device | |
| CN103093191A (en) | Object recognition method with three-dimensional point cloud data and digital image data combined | |
| CN106991695A (en) | A kind of method for registering images and device | |
| CN108830888B (en) | Coarse matching method based on improved multi-scale covariance matrix characteristic descriptor | |
| CN103606170B (en) | Streetscape image feature based on colored Scale invariant detects and matching process | |
| CN108550166B (en) | Spatial target image matching method | |
| CN103679714A (en) | Method for automatic registration of optical image and SAR image based on gradient cross-correlation | |
| CN103886611A (en) | Image matching method suitable for automatically detecting flight quality of aerial photography | |
| CN110223355B (en) | Feature mark point matching method based on dual epipolar constraint | |
| CN110120013B (en) | Point cloud splicing method and device | |
| CN103136525A (en) | High-precision positioning method for special-shaped extended target by utilizing generalized Hough transformation | |
| CN104866851A (en) | Scale-invariant feature transform (SIFT) algorithm for image matching | |
| CN108182705A (en) | A kind of three-dimensional coordinate localization method based on machine vision | |
| CN105513094A (en) | Stereo vision tracking method and stereo vision tracking system based on 3D Delaunay triangulation | |
| CN103077528A (en) | Rapid image matching method based on DCCD (Digital Current Coupling)-Laplace and SIFT (Scale Invariant Feature Transform) descriptors | |
| CN105869168A (en) | Multi-source remote sensing image shape registering method based on polynomial fitting | |
| CN104077769A (en) | Error matching point pair removing algorithm in image registration | |
| CN103308000A (en) | Method for measuring curve object on basis of binocular vision |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160608 |
|
| RJ01 | Rejection of invention patent application after publication |