CN101286232A - High precision subpixel image registration method - Google Patents
High precision subpixel image registration method Download PDFInfo
- Publication number
- CN101286232A CN101286232A CNA2008100383272A CN200810038327A CN101286232A CN 101286232 A CN101286232 A CN 101286232A CN A2008100383272 A CNA2008100383272 A CN A2008100383272A CN 200810038327 A CN200810038327 A CN 200810038327A CN 101286232 A CN101286232 A CN 101286232A
- Authority
- CN
- China
- Prior art keywords
- image
- sub
- pixel
- registration
- width
- 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
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a high-precision sub-pixel image registration method, which comprises that: an image space registration degree solving method which utilizes the least square as an evaluation function is increased according to the spatial features of an original image of a sub-pixel detection system before the image processing procedure of the ideal sub-pixel detection system, and the space registration inaccurate coefficient is transferred to the image processing procedure of the sub-pixel detection system to improve the space registration degree of the original image. The registration method of the invention has the advantages that the specific algorithm is utilized to solve the space registration inaccurate coefficient of the original image brought by the precision of a device, the self-adaptive compensation is formed by the coefficient transfer, and the image processing procedure of the sub-pixel detection system takes the registration inaccurate degree of the original image as the reference, thus eliminating the reduction of the sub-pixel detection effect caused by the precision problem. The sub-pixel detection system can be closer to the engineering practical degree.
Description
Technical field
The present invention relates to image processing techniques, specifically refer to a kind of method for registering of high precision subpixel image, it is used for the subpixel image registration, improves the registration accuracy of image.
Background technology
Utilizing inferior pixel technology to carry out super-resolution image reconstruction is the effective means that improves image spatial resolution, it also is important branch in the image processing field, recent years, domestic a lot of scholar is also in the inferior pixel detecting technology of research, but research mainly concentrates on the algorithm research, most of research all is Utopian Image Acquisition subsystem of hypothesis, then desirable original image is obtained final high-definition picture by Processing Algorithm.Shown in Figure 1 is desirable inferior pixel detecting process model figure, at first suppose with X and Y direction respectively displacement 50% pixel obtain several raw images, then original image is carried out the pixel coupling according to 50% pixel displacement---in desirable processing procedure, even original image is not because reasons of error has the desirable space displacement of 50% pixel displacement, also can be assumed to be 50% pixel displacement to original image---utilize the super-resolution image blending algorithm at last, carry out image co-registration as convex set projection (POCS) method commonly used and handle.The research correlativity of this type of research and actual instrumentation is smaller, and significant limitation is arranged, and directive significance is limited, can't solve the inaccurate problem of original image coupling that actual instrumentation is brought by the mechanical precision problem in processing and operational process.
Summary of the invention
The method for registering that the purpose of this invention is to provide a kind of high precision subpixel image solves the inaccurate problem of inferior pixel detecting system's original image coupling that immesurable mechanical precision brought by actual instrumentation.
The invention discloses and a kind ofly improve inferior pixel detecting system original image matching precision, finally improve the method for reconstructed image quality by adaptive equalization.Specifically, be before at the inferior pixel detecting of ideal system image processing program---many image co-registration---, space characteristics according to inferior pixel detecting system original image: be exactly that the displacement meeting is controlled in 50% pixel, and rotation is normally non-existent, or exist but under the negligible condition, in order to increase accuracy in computation and rapidity, utilize the gradation of image least square to calculate the inaccurate degree of original image coupling for the image space registration degree method for solving of evaluation function, and inaccurate coefficient is joined in the space pass to inferior pixel detecting system image processing program and compensate, by the ideal position of 50% pixel displacement in the actual space displacement alternative image processing procedure of original image, thus the picture displacement error problem of preventing and bringing by the instrument mechanical precision.
Self-adapting compensation method of the present invention is shown in Fig. 2 FB(flow block), at first the figure a of need registration and the gray-scale value of each pixel of figure b are carried out 2n cube interpolation doubly, n span 3～5, it is low that n gets low value images match precision, n gets high value images match precision height, but can increase the weight of the COMPUTER CALCULATION load, reduce Flame Image Process efficient; Adopt least square method then, find the solution the space displacement of raw image reality, obtain image and join inaccurate coefficient, scheme the ideal space displaced position of a relatively as starting point to scheme b during calculating, carry out four direction up and down, slide in the position of each n sub-neighborhood of pixels, slide into a position at every turn, just calculate the correlativity of two width of cloth images: with the pixel grey scale least square of same spatial location as evaluation function, operator is for carrying out n * n slip, after calculating and the comparison result of calculation, choose the image space position of evaluation function value minimum, obtain the real space displacement of original image; Utilize convex set projection (POCS) method that figure a and figure b are carried out fusion treatment at last, the actual space displacement with original image during processing replaces 50% theoretical space displacement to obtain target image.
The invention has the advantages that:
Regulate flow process by increasing self-adaptation, solved the immesurable original image space displacement error of bringing by precision problem, improved inferior pixel detecting technology improving the effect of instrument spatial resolution.
Description of drawings
Fig. 1: the subpixel image fusion treatment block diagram of 50% pixel displacement.
Fig. 2: adaptive equalization subpixel image fusion treatment block diagram.
Fig. 3: image cube interpolation synoptic diagram.
Fig. 4: subpixel image registration synoptic diagram.
Fig. 5: actual routine registering images, wherein: need the image of registration when figure a, figure b, figure c is the image of handling with the subpixel image fusion method of 50% pixel displacement, and figure d is the image that adopts after the inventive method is handled.
Embodiment
Below in conjunction with accompanying drawing 3～Fig. 5 the embodiment that adaptive equalization of the present invention improves the method for inferior pixel detecting technique effect is elaborated:
As shown in Figure 3, after original graph image pattern a in obtaining Fig. 5 and the figure b, directly original image is not carried out registration according to 50% space displacement, but at first carry out 6 times cube interpolation, because the pattern space registration accuracy that 6 duplication of the cube interpolation can calculate is controlled at below 8%, and according to our analysis, image registration accuracy is controlled at 8% pixel when following, influential effect to many image co-registration is just smaller, so select 6 times of space interpolations, be to combine counting yield and calculate the selection that the double requirements of effect is done.
As Fig. 4, after original graph image pattern a and figure b done 6 times of space interpolations, scheme the ideal space displaced position of a relatively as starting point to scheme b, carry out four direction up and down, slide in the position of each 3 sub-neighborhood of pixels, slide into a position at every turn, just calculate the correlativity of two width of cloth images: with the pixel grey scale least square of same spatial location as evaluation function, operator is for carrying out 36 slips, after calculating and the comparison result of calculation, choose the image space position of evaluation function value minimum, obtain the real space displacement of original image, and transmit two locus coefficients.
As Fig. 5, convex set projection (POCS) method is according to transmitting the image real space position parameter of coming, calculate the real space displacement of original image, when carrying out image co-registration, replace theoretical value 50% pixel displacement (actual area that same direction striped covers on the pre-estimation image among the figure b) that the pre-estimation image is carried out projection to real image with original image real space displacement (actual area that same direction striped covers on the pre-estimation image among the figure a), so that the pixel relevant real space merges, obtain the output image of more realistic scenery half-tone information, figure c carries out result calculated according to the method that does not have error to the image that the real space error is arranged, figure d is the algorithm that has added the image space registration, the image that the real space error is arranged is carried out result calculated, and the effect of discovery figure d is better than figure c.
Because the existence of error, the inaccurate image of coupling carries out fusion treatment, not only can't improve the spatial resolution of target image, also can in the image co-registration process, bring the deterioration of image effect, improve inferior pixel detecting system image registration accuracy and increased adaptive equalization, the influence of the error of having prevented will effectively address this problem.
Claims (2)
1. the method for registering of a subpixel image, it adopts the convex set sciagraphy to carry out image co-registration and handles, it is characterized in that: adopting the convex set sciagraphy to carry out before image co-registration handles, it also has a step of finding the solution the space displacement of raw image reality to obtain the required image of convex set sciagraphy to join inaccurate coefficient.
2. the method for registering of a kind of subpixel image according to claim 1 is characterized in that: said image is joined inaccurate coefficient and is tried to achieve according to the following steps:
A. will need each pixel segmentation of two width of cloth images of registration to become 2n * 2n sub-pixel, the gray-scale value of each sub-pixel adopts a 2n cube interpolation doubly to try to achieve n span 3～5;
B. with the ideal space displaced position of above-mentioned two width of cloth images that need registration as starting point, adopt least square method to calculate evaluation to the gray-scale value of corresponding each sub-pixel in two width of cloth images;
C. beginning carry out up and down from the ideal space displaced position above-mentioned two width of cloth images that need registration, four direction carries out the slippage of 1 to n sub-pixel, after each slippage, the gray-scale value employing least square method of corresponding each sub-pixel in two width of cloth images is calculated evaluation;
D. comparison step B, C n * n the calculated value of trying to achieve, the pairing sub-pixel slip value of reckling is exactly the optimum value that image is joined inaccurate coefficient.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008100383272A CN101286232A (en) | 2008-05-30 | 2008-05-30 | High precision subpixel image registration method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008100383272A CN101286232A (en) | 2008-05-30 | 2008-05-30 | High precision subpixel image registration method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101286232A true CN101286232A (en) | 2008-10-15 |
Family
ID=40058423
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2008100383272A Pending CN101286232A (en) | 2008-05-30 | 2008-05-30 | High precision subpixel image registration method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101286232A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101980291A (en) * | 2010-11-03 | 2011-02-23 | 天津大学 | Random micro-displacement-based super-resolution image reconstruction method |
| CN101980289A (en) * | 2010-10-25 | 2011-02-23 | 上海大学 | Frequency domain registration and convex set projection-based multi-frame image super-resolution reconstruction method |
| CN101706961B (en) * | 2009-11-10 | 2012-12-12 | 北京航空航天大学 | Image registration method and image registration device |
| CN103262121A (en) * | 2010-12-20 | 2013-08-21 | 国际商业机器公司 | Detection and tracking of moving objects |
| CN105205812A (en) * | 2015-09-01 | 2015-12-30 | 哈尔滨工业大学 | Multiframe image reconstruction method based on microsatellite constellation |
| CN105651939A (en) * | 2015-12-29 | 2016-06-08 | 重庆大学 | Concentration detection precision correction method based on projection onto convex set in electron nose system |
-
2008
- 2008-05-30 CN CNA2008100383272A patent/CN101286232A/en active Pending
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101706961B (en) * | 2009-11-10 | 2012-12-12 | 北京航空航天大学 | Image registration method and image registration device |
| CN101980289A (en) * | 2010-10-25 | 2011-02-23 | 上海大学 | Frequency domain registration and convex set projection-based multi-frame image super-resolution reconstruction method |
| CN101980289B (en) * | 2010-10-25 | 2012-06-27 | 上海大学 | Frequency domain registration and convex set projection-based multi-frame image super-resolution reconstruction method |
| CN101980291A (en) * | 2010-11-03 | 2011-02-23 | 天津大学 | Random micro-displacement-based super-resolution image reconstruction method |
| CN101980291B (en) * | 2010-11-03 | 2012-01-18 | 天津大学 | Random micro-displacement-based super-resolution image reconstruction method |
| CN103262121A (en) * | 2010-12-20 | 2013-08-21 | 国际商业机器公司 | Detection and tracking of moving objects |
| CN105205812A (en) * | 2015-09-01 | 2015-12-30 | 哈尔滨工业大学 | Multiframe image reconstruction method based on microsatellite constellation |
| CN105651939A (en) * | 2015-12-29 | 2016-06-08 | 重庆大学 | Concentration detection precision correction method based on projection onto convex set in electron nose system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104935909B (en) | Multi-image super-resolution method based on depth information | |
| CN101286232A (en) | High precision subpixel image registration method | |
| JP6553826B1 (en) | Image processing apparatus, image processing method, and image processing program | |
| CN102273208A (en) | Image processing device, image processing method, and program | |
| JP2009170970A5 (en) | ||
| CN102113330A (en) | Adaptive interpolation with artifact reduction of images | |
| CN104685513A (en) | Feature based high resolution motion estimation from low resolution images captured using an array source | |
| US9578264B2 (en) | Image processing device, imaging device, and image processing method | |
| US20090268984A1 (en) | Subpixel Registration | |
| JP2011254321A (en) | Image processing system, image processing method, and program | |
| CN111105460B (en) | RGB-D camera pose estimation method for three-dimensional reconstruction of indoor scene | |
| CN104052979A (en) | Apparatus and techniques for image processing | |
| CN102184542A (en) | Stereo matching method for stereo binocular vision measurement | |
| CN104635188A (en) | K-space reconstruction method and magnetic resonance imaging method | |
| CN102265628B (en) | Method for detecting and correcting bad pixels in image sensor | |
| CN112652020A (en) | Visual SLAM method based on AdaLAM algorithm | |
| Roussel et al. | Monocular depth estimation in new environments with absolute scale | |
| US10134136B2 (en) | Image processing apparatus and image processing method | |
| CN108076341B (en) | In-orbit real-time digital image stabilization method and system is imaged in a kind of video satellite | |
| CN102521837A (en) | Image registration method based on global information | |
| CN106920213B (en) | Method and system for acquiring high-resolution image | |
| CN103248838A (en) | CMOS camera image bias adjusting method based on digital field TDI (time delayed and integration) algorithm | |
| JP2015154324A (en) | Imaging system, imaging apparatus, and image processing system | |
| CN103997611A (en) | Method for suppressing image fixed-pattern noise based on noise template | |
| US8035698B2 (en) | Joint automatic demosaicking and white balancing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20081015 |