CN107481275A - A kind of two step method for registering images based on reference picture compensation - Google Patents
A kind of two step method for registering images based on reference picture compensation Download PDFInfo
- Publication number
- CN107481275A CN107481275A CN201710538321.0A CN201710538321A CN107481275A CN 107481275 A CN107481275 A CN 107481275A CN 201710538321 A CN201710538321 A CN 201710538321A CN 107481275 A CN107481275 A CN 107481275A
- Authority
- CN
- China
- Prior art keywords
- reference picture
- registering
- translation
- image
- target image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/37—Determination of transform parameters for the alignment of images, i.e. image registration using transform domain methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration by the use of more than one image, e.g. averaging, subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
-
- 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
- G06T2207/10004—Still image; Photographic image
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
A kind of two step method for registering images based on reference picture compensation.Present invention assumes that rotation yardstick translation（RST）Transformation model, for the image registration problem of big displacement situation between target image and reference picture be present, using by the thick two-step method strategy to essence, the first step completes rough estimate first by the insensitive method for registering of big displacement, then bit shift compensation is carried out to reference picture according to rough estimate value, the higher method for registering of second step service precision completes high accuracy estimation afterwards, the result finally estimated using the composite formula that residual transform model obtains after bit shift compensation first two steps is completed result and synthesized, and obtains the high-precision sub-pix estimated result under big displacement situation.
Description
Technical field
The present invention relates to a kind of image registration（image registration）Method.
Background technology
Image registration is two width or several different images to same target to be snapped to the process of same coordinate system, its
Middle different image may shoot to obtain from different time, different angle or different cameral.Image registration is many image procossings
And the basis of application, such as, super-resolution image reconstruction（super-resolution image reconstruction）, figure
As splicing（image mosaicing）And image co-registration（image fusion）.In numerous performance indications, registration accuracy is most
To be important, it will directly affect the effect and performance of subsequent treatment, in general, sub-pix（sub-pixel）Precision is to be permitted
The basic demand applied more.Method for registering images can be typically divided into two major classes：Based on region（area-based）Image
Method for registering and feature based（feature-based）Method for registering images；Wherein, the image registration based on region is directed to certain
Individual similarity is directly handled image intensity value, and the image registration of feature based then extracts some features from image
Put and then matched.
When shift value between image subject to registration is bigger, the image registration to succeed becomes difficult, especially, obtains
High-precision sub-pixel precision becomes particularly difficult, now, general from based on by the thick two step method for registering tactful to essence, i.e.,：
The first step is first by the method for registering insensitive to big displacement（Such as method for registering of phase correlation method, feature based etc.）
Rough estimate is completed, bit shift compensation is then carried out according to rough estimate value, afterwards in the higher registration side of second step service precision
Method（Such as the method for registering based on gradient）High accuracy estimation is completed, finally completes result synthesis.
Common transformation model includes translation transformation model, rigid body translation（That is roto-translatory " rotation-
translation”：RT）Model, similarity transformation（That is rotation-yardstick-translation " rotation-scale-translation "：
RST）Model, affine Transform Model, projective transformation model etc..The present invention is based on RST models.In general, any one is converted
Model can be decomposed into the sequential combination of basic transformation model, in RST transformation models, be typically considered as rotation transformation,
The sequential combination of change of scale and translation transformation；Moreover, the result of this combination is typically closely related with combination order.Based on
By slightly into two step method for registering of smart strategy, because the estimated result of the first step is coarse value, bit shift compensation is not inevitable thorough,
The transformation model then obtained after bit shift compensation will be closely related with compensation way, and it actually determines that progress second step is fine
The residual transform model of estimation.
The present invention, based on by the thick two-step method strategy to essence, mends for the image registration of big displacement situation be present in displacement
Reference picture is compensated when repaying, become moreover, this method considers that the estimation fine to second step of reference picture bit shift compensation is remaining
The influence of mold changing type, error compensation is carried out when end product synthesizes, so can further improve registration accuracy.
Bibliography：
B. Zitova and J. Flusser, "Image registration methods: A survey," Image
and Vision Computing, Vol.21, No.11, pp.977-1000 (2003)。
B.D. Lucas and T. Kanade, "An iterative image registration technique
with an application to stereo vision," in Imaging Understanding Workshop,
pp.121-130 (1981)。
D. Keren, S. Peleg, and R. Brada, "Image sequence enhancement using
sub-pixel displacement," in CVPR'88, pp.742-746 (1988)。
X. Li, "Gradient-based registration of rotated, scaled, and
translated images," in SPIE, MIPPR 2013: Pattern Recognition and Computer
Vision , vol. 8919, 2013, p. 891917。
C.D. Kuglin and D.C. Hines, "The phase correlation image alignment
method," in IEEE International Conference on Cybernetics and Society, pp.163-
165 (1975)。
B.S. Reddy and B.N. Chatterji, "An FFT-based technique for
translation, rotation, and scale-invariant image registration," IEEE
Transactions on Image Processing, Vol.5, No.8, pp.1266-1271 (1996)。
The content of the invention
The present invention provides a kind of two step method for registering images based on reference picture compensation, and it is for big displacement situation being present
Image registration problem, to smart two-step method strategy, reference picture is compensated in bit shift compensation using by thick, moreover,
Compared with routine is based on the method for registering images of two-step method, this method considers reference picture bit shift compensation to the fine estimation of second step
The influence of residual transform model, error compensation is carried out when end product synthesizes, can further improve registration accuracy.
The detail of the inventive method refers to " accompanying drawing " part and " embodiment " part.
Brief description of the drawings
Fig. 1 is the general schematic diagram of image registration problem, wherein（11）For reference picture,（12）For target image,
（13）Represent registration process,（14）For（11）With（12）Schematic diagram is superimposed after registration.
Fig. 2 is the schematic diagram of this patent method for registering images, wherein, input two images see reference picture as respectively
（211）And target image（212）, by first stage rough estimate（22）The rough estimate result of first stage is obtained afterwards
（23）, then to reference picture（211）Carry out bit shift compensation（24）, the fine estimation of second stage is carried out afterwards（25）And obtain
The estimated result of second stage（26）, finally carry out result synthesis（27）And obtain final estimated result（28）, in addition, arrow line
（291）Represent incorporated by reference view data（211）To carry out reference picture bit shift compensation（24）, arrow line（292）Represent result
Synthesis（27）Shi Yinyong first stage estimated results（23）, arrow dotted line（293）Represent result synthesis（27）Will be with reference picture
Bit shift compensation（24）It is relevant.
Fig. 3 is the execution step of this patent method for registering images, in comparison with Fig. 2,（31）It is corresponding（211）With（212）,（32）
It is corresponding（22）And（23）,（33）It is corresponding（24）,（34）It is corresponding（25）And（26）,（35）It is corresponding（27）And（28）.
Embodiment
The present invention is described in detail below in conjunction with the accompanying drawings.
As it was previously stated, the method for registering images of the present invention assumes RST transformation models, it includes three classes totally four parameters：It is flat
Move（Translated for horizontal direction,Translated for vertical direction,Operated for matrix transposition）, the anglec of rotationAnd contracting
Put the factor, and assume reference pictureWith target imageBetween relation be, wherein,AndRepresent twiddle operation；So, image registration purpose estimates、AndValue.
This method is made up of following five steps altogether：
Step 1）Read in view data subject to registrationWith。
Step 2）Reference picture is completed to the insensitive method for registering of big displacementWith target imageBetween
Rough estimate, and estimated result is denoted as respectively、And。
Step 3）To reference pictureCompensate to obtain：, it is then right
WithBetween carry out public domain extraction.
Step 4）Completed from the higher method for registering of precisionWithBetween high accuracy estimation, and estimated result
It is denoted as respectively、And。
Step 5）Reference pictureWith target imageBetween final estimated result be denoted as respectively、And, their value obtains in the following way respectively：,, and。
Claims (1)
- A kind of 1. method for registering images, it is characterised in that：Based on rotation-yardstick-translation（rotation-scale- translation：RST）Transformation model；DefinitionFor reference picture,For target image, and the relation for defining them is, wherein,,For horizontal direction coordinate value,For vertical direction coordinate value,For square Battle array transposition operation, translation,Translated for horizontal direction,Translated for vertical direction,For the anglec of rotation,For contracting Put the factor,Represent twiddle operation：, andFor SIN function andFor cosine function；The present invention is made up of following steps：Step 1）Read in view data subject to registration；Step 2）Reference picture is completed using to the insensitive method for registering of big displacementWith target imageBetween it is rough Estimation, estimated result are denoted as respectively、And；Step 3）To reference pictureCompensate to obtain：, it is then rightWithBetween carry out public domain extraction；Step 4）The higher method for registering of service precision is completedWithBetween fine estimation, estimated result is denoted as respectively、And；Step 5）Reference pictureWith target imageBetween final estimated result be denoted as respectively、And, their value obtains in the following way respectively：,, and。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710538321.0A CN107481275A (en) | 2017-07-04 | 2017-07-04 | A kind of two step method for registering images based on reference picture compensation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710538321.0A CN107481275A (en) | 2017-07-04 | 2017-07-04 | A kind of two step method for registering images based on reference picture compensation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107481275A true CN107481275A (en) | 2017-12-15 |
Family
ID=60595430
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710538321.0A Pending CN107481275A (en) | 2017-07-04 | 2017-07-04 | A kind of two step method for registering images based on reference picture compensation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107481275A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111882588A (en) * | 2020-07-29 | 2020-11-03 | Oppo广东移动通信有限公司 | Image block registration method and related product |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103544710A (en) * | 2013-11-08 | 2014-01-29 | 河南工业大学 | Image registration method |
| CN104067312A (en) * | 2011-12-24 | 2014-09-24 | 高等技术学校 | Image registration method and system robust to noise |
| CN104063849A (en) * | 2014-06-20 | 2014-09-24 | 武汉工程大学 | Video super-resolution reconstruction method based on image block self-adaptive registration |
| CN105869141A (en) * | 2015-12-15 | 2016-08-17 | 乐视致新电子科技（天津）有限公司 | Image registration method and apparatus |
| CN106056625A (en) * | 2016-05-25 | 2016-10-26 | 中国民航大学 | Airborne infrared moving target detection method based on geographical homologous point registration |
-
2017
- 2017-07-04 CN CN201710538321.0A patent/CN107481275A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104067312A (en) * | 2011-12-24 | 2014-09-24 | 高等技术学校 | Image registration method and system robust to noise |
| CN103544710A (en) * | 2013-11-08 | 2014-01-29 | 河南工业大学 | Image registration method |
| CN104063849A (en) * | 2014-06-20 | 2014-09-24 | 武汉工程大学 | Video super-resolution reconstruction method based on image block self-adaptive registration |
| CN105869141A (en) * | 2015-12-15 | 2016-08-17 | 乐视致新电子科技（天津）有限公司 | Image registration method and apparatus |
| CN106056625A (en) * | 2016-05-25 | 2016-10-26 | 中国民航大学 | Airborne infrared moving target detection method based on geographical homologous point registration |
Non-Patent Citations (1)
| Title |
|---|
| 李相国等: "《基于相位相关与Keren算法的图像配准》", 《微计算机应用》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111882588A (en) * | 2020-07-29 | 2020-11-03 | Oppo广东移动通信有限公司 | Image block registration method and related product |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP5294343B2 (en) | Image alignment processing device, area expansion processing device, and image quality improvement processing device | |
| Pollefeys | Visual 3D Modeling from Images. | |
| WO2015154601A1 (en) | Non-feature extraction-based dense sfm three-dimensional reconstruction method | |
| US20160217577A1 (en) | Enhanced phase correlation for image registration | |
| WO2012058902A1 (en) | Method and apparatus for combining panoramic image | |
| CN103544710A (en) | Image registration method | |
| Zhu et al. | Image mosaic method based on SIFT features of line segment | |
| Lee et al. | Accurate registration using adaptive block processing for multispectral images | |
| Ben-Ari | A unified approach for registration and depth in depth from defocus | |
| Yan et al. | Image stitching with single-hidden layer feedforward neural networks | |
| Chen et al. | Image stitching algorithm research based on OpenCV | |
| Alyousefi et al. | Multi-camera motion estimation with affine correspondences | |
| CN107481275A (en) | A kind of two step method for registering images based on reference picture compensation | |
| CN108961381B (en) | Method and device for rendering a three-dimensional geometric model of an object | |
| Li | High-accuracy subpixel image registration with large displacements | |
| Shimizu et al. | Multi-parameter simultaneous estimation on area-based matching | |
| Wang et al. | A real-time correction and stitching algorithm for underwater fisheye images | |
| CN107492118A (en) | A kind of two step method for registering images based on target image compensation | |
| Saalfeld | Computational methods for stitching, alignment, and artifact correction of serial section data | |
| CN106570911B (en) | Method for synthesizing facial cartoon based on daisy descriptor | |
| JP2019220032A (en) | Program, apparatus and method that generate display image which is transformed from original image on the basis of target image | |
| Zhang et al. | Image mosaic of bionic compound eye imaging system based on image overlap rate prior | |
| Hu et al. | Digital video stabilization based on multilayer gray projection | |
| Ogawa et al. | SIFT-based multi-frame super resolution for 250 million pixel images | |
| Zhu et al. | Image stitching using simplified SIFT |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20171215 |