CN107146242A - A kind of high precision image method for registering that kernel estimates are obscured for imaging system - Google Patents
A kind of high precision image method for registering that kernel estimates are obscured for imaging system Download PDFInfo
- Publication number
- CN107146242A CN107146242A CN201710175300.7A CN201710175300A CN107146242A CN 107146242 A CN107146242 A CN 107146242A CN 201710175300 A CN201710175300 A CN 201710175300A CN 107146242 A CN107146242 A CN 107146242A
- Authority
- CN
- China
- Prior art keywords
- image
- checkerboard
- original
- chessboard
- coordinate
- 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
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000003384 imaging method Methods 0.000 title claims abstract description 18
- 239000011159 matrix material Substances 0.000 claims description 12
- 238000013507 mapping Methods 0.000 claims description 11
- 238000001514 detection method Methods 0.000 claims description 7
- 239000004072 C09CA03 - Valsartan Substances 0.000 claims description 3
- 235000005121 Sorbus torminalis Nutrition 0.000 claims description 3
- 244000152100 Sorbus torminalis Species 0.000 claims description 3
- 230000009286 beneficial effect Effects 0.000 description 1
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
- G06T7/32—Determination of transform parameters for the alignment of images, i.e. image registration using correlation-based methods
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of high precision image method for registering that kernel estimates are obscured for imaging system.Original checkerboard image is obtained first and checkerboard image is shot, and by detecting original gridiron pattern and shooting tessellated angular coordinate, the coordinate pair derived therebetween answers relational expression, is mapped according to the relational expression, so as to obtain the image of accuracy registration.Obtain that chessboard table images can be substituted for required image after corresponding relation formula according to actual needs, facilitate successive image to handle.
Description
Technical Field
The invention mainly relates to the field of digital image processing, in particular to high-precision image registration for fuzzy kernel estimation of an imaging system
A method.
Background
In The field of image processing such as computational photography and image restoration, a blur kernel of an imaging system needs to be estimated in many cases, wherein a common method Is to use a checkerboard calibration plate shot and printed by The imaging system to obtain a corresponding blur image and a clear image, and then use a Non-blind convolution image restoration algorithm to estimate The blur kernel (The Non-spatial Sub-pixel local Point Spread Function Estimation Is a Well past distributed map algorithm, 2012). The method has the main problems that the registration accuracy between the shot blurred image and the clear image is not high, because the blurred image and the clear image are shot separately twice, and the camera parameters are different during shooting, the external conditions of the two shots are difficult to ensure to be completely consistent, and the blurred image and the clear image have deviations which directly influence the estimation accuracy of a blur kernel. Therefore, after the blurred image and the sharp image are obtained by shooting, the registration operation is carried out, and even a relatively high registration algorithm still has difficulty in eliminating errors caused during shooting. This is also one of the main factors affecting the estimation accuracy of the blur kernel. Besides estimating an estimation blur kernel in image restoration, the image processing field such as image stitching also faces the problem of low registration accuracy, so how to further improve the image registration accuracy is also an urgent need to be solved in the image processing field.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: aiming at the problem that the image registration precision needs to be improved at present, the invention provides a high-precision image registration method for imaging system fuzzy kernel estimation. The method is not to shoot the printed calibration plate, but to directly shoot the calibration plate image generated on the computer. The method comprises the steps of shooting a calibration board image on a computer to obtain a fuzzy image, deducing a coordinate corresponding relation between an original checkerboard and a shooting checkerboard by detecting corner point coordinates of the original checkerboard and the shooting checkerboard, mapping according to the relation, and taking the mapped image as a clear image corresponding to the fuzzy image. This avoids the need to separately capture a sharp image, which is accurately registered with the blurred image by the mapping method. And according to the actual image processing requirement, the chessboard grids can be replaced by the required specific image after the corresponding relation is obtained, so that the subsequent image processing is facilitated.
In order to solve the technical problems, the technical scheme provided by the invention is as follows:
a high-precision image registration method for imaging system fuzzy kernel estimation is characterized in that:
the method comprises the following steps: an original checkerboard image is generated on a computer, wherein the checkerboard image can be directly generated on the computer by matlab software and is a black and white checkerboard image.
Step two: the checkerboard image on the computer is shot by using an imaging device to obtain a blurred image, wherein the imaging device can be an imaging device which is actually needed by a mobile phone or a camera and the like.
Step three: carrying out corner detection on the checkerboard image pair in the first step and the second step to obtain a corresponding original checkerboard image and a corner coordinate matrix corresponding to the shot checkerboard imageMat1 andMat2, the corner point detection method can adopt a commonly used corner point detection method to obtain a corner point coordinate matrixMat1 andMat2 is 2 xRow*ColWhereinRowRepresenting the number of transverse chequers in the checkerboard image used,Colrepresenting the number of columns in the checkerboard image used.
Step four: and calculating the coordinate corresponding relation between the original checkerboard image and the shot checkerboard image, and mapping the original checkerboard image to the corresponding area of the shot checkerboard image according to the corresponding coordinate relation, so that the precise and matched clear and actual shot checkerboard image can be obtained.
The coordinate correspondence between the original checkerboard image and the shot checkerboard image is derived as follows:
aiming at the angular point coordinate matrix obtained in the third stepMat1 andMat2, 1iThe coordinate indices of the four corner points of the checkerboard, up, down, left, and right, can be expressed as:
(1)
wherein,c1P、c2P、c3Pandc4Prespectively representiCoordinate indexes of four corner points of the upper part, the lower part, the left part and the right part of the checkerboard are obtained;floorthe operation symbol of the nearest lower integer is taken;Rowrepresenting the number of transverse chessboards in the checkerboard image.
The index value obtained by the formula (1) can be respectively obtained from the corner point coordinate matrix obtained in the third stepMat1 andMat2 obtaining original chessboard image and shooting the second in the chessboard imageiThe coordinates corresponding to each checkerboard are shown in formula (2) and formula (3):
(2)
(3)
wherein,c1、c2、c3 andc4 respectively representing the original chessboard patternsiCoordinates of four corner points of upper, lower, left and right corresponding to each checkerboard,cc1、cc2、cc3 andcc4 respectively representing the chessboard picturesiCoordinates of four corner points, upper, lower, left and right, corresponding to each checkerboard.
From the first in the original chessboard imageiThe coordinates corresponding to each checkerboard may result in parameters α and β:
(4)
wherein,c1(2) representing coordinate points of original chessboard imagecOrdinate of 1, β denotes coordinate points for capturing checkerboard imagesc3, abscissa.
Based on the above parameters, the corresponding relation between the original chessboard image and the photographed chessboard image can be found:
(5)
wherein,cc1、cc2、cc3 andcc4 respectively representing the chessboard picturesiThe coordinates of the four corner points of the upper, lower, left and right sides corresponding to each checkerboardx,y]Representing the corresponding coordinates mapped from a point on the shot checkerboard to the original checkerboard image.
The original checkerboard images can be mapped to the corresponding positions of the shot checkerboard images one by one according to the corresponding relation (5), so that the precisely registered checkerboard images are obtained.
The invention has the beneficial effects that:
compared with the method of respectively shooting the fuzzy image and the clear image and then carrying out registration mentioned in the reference document [1], the method only needs to shoot a calibration plate on a computer to obtain the fuzzy image, then obtains the clear image corresponding to the fuzzy image through deducing a corresponding mapping formula and a mapping method, avoids the subsequent registration process, and can ensure that the error is within one pixel because the obtained clear image can be accurately matched with the previously shot fuzzy image. The registration method can well avoid errors caused by the shooting process, so that the subsequent image processing precision is finally improved.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is an original checkerboard image;
FIG. 3 is a checkerboard image taken;
fig. 4 is a checkerboard image that matches exactly after mapping.
Detailed Description
The invention is described in detail below with reference to fig. 1.
The embodiment provides a high-precision image registration method for imaging system fuzzy kernel estimation, which comprises the following steps: the method comprises the following steps: an original checkerboard image is generated on a computer, as shown in fig. 2, the checkerboard image can be directly generated on the computer by matlab software, and is a black and white checkerboard image.
Step two: the checkerboard image on the computer is shot by using an imaging device to obtain a blurred image, wherein the imaging device can be an imaging device which is actually needed by a mobile phone or a camera and the like. In a specific embodiment, a checkerboard image on a computer is captured with a camera.
Step three: carrying out corner detection on the checkerboard image pair in the first step and the second step to obtain a corresponding original checkerboard image and a corner coordinate matrix corresponding to the shot checkerboard imageMat1 andMat2, the corner point detection method can adopt a commonly used corner point detection method to obtain a corner point coordinate matrixMat1 andMat2 is 2 xRow*ColWhereinRowRepresenting the number of transverse chequers in the checkerboard image used,Colrepresenting the number of columns in the checkerboard image used. In an embodiment, selectingRow=18 andCol=30。
step four: and calculating the coordinate corresponding relation between the original checkerboard image and the shot checkerboard image, and mapping the original checkerboard image to the corresponding area of the shot checkerboard image according to the corresponding coordinate relation, so that the precise and matched clear and actual shot checkerboard image can be obtained.
The coordinate correspondence between the original checkerboard image and the shot checkerboard image is derived as follows:
aiming at the angular point coordinate matrix obtained in the third stepMat1 andMat2, 1iThe coordinate indices of the four corner points of the checkerboard, up, down, left, and right, can be expressed as:
(1)
wherein,c1P、c2P、c3Pandc4Prespectively representiCoordinate indexes of four corner points of the upper part, the lower part, the left part and the right part of the checkerboard are obtained;floorthe operation symbol of the nearest lower integer is taken;Rowrepresenting the number of transverse chessboards in the checkerboard image.
The index value obtained by the formula (1) can be respectively obtained from the corner point coordinate matrix obtained in the third stepMat1 andMat2 obtaining original chessboard image and shooting the second in the chessboard imageiThe coordinates corresponding to each checkerboard are shown in formula (2) and formula (3):
(2)
(3)
wherein,c1、c2、c3 andc4 respectively representing the original chessboard patternsiCoordinates of four corner points of upper, lower, left and right corresponding to each checkerboard,cc1、cc2、cc3 andcc4 respectively represent beatsTake a picture of chessboardiCoordinates of four corner points, upper, lower, left and right, corresponding to each checkerboard.
From the first in the original chessboard imageiThe coordinates corresponding to each checkerboard may result in parameters α and β:
(4)
wherein,c1(2) representing coordinate points of original chessboard imagecOrdinate of 1, β denotes coordinate points for capturing checkerboard imagesc3, abscissa.
Based on the above parameters, the corresponding relation between the original chessboard image and the photographed chessboard image can be found:
(5)
wherein,cc1、cc2、cc3 andcc4 respectively representing the chessboard picturesiThe coordinates of the four corner points of the upper, lower, left and right sides corresponding to each checkerboardx,y]Representing the corresponding coordinates mapped from a point on the shot checkerboard to the original checkerboard image.
The original checkerboard images can be mapped to the corresponding positions of the shot checkerboard images one by one according to the corresponding relation (5), so that the precisely registered checkerboard images are obtained. In a specific embodiment, there are 18 × 30=540 checkerboards in total, and the mapping needs to be performed sequentially according to the above correspondence.
As described above, the present invention provides a high-precision image registration method for blur kernel estimation of an imaging system, aiming at the problem that the precision of image registration needs to be improved at present. The method avoids shooting the printed calibration board image twice, directly shoots the calibration board image generated on a computer as a fuzzy image, deduces a coordinate corresponding relation between an original checkerboard and a shot checkerboard by detecting the angular point coordinates of the two checkerboard, carries out mapping according to the relation, obtains a clear image corresponding to the fuzzy image by mapping, and thus can obtain an accurately registered image. After the corresponding relation is obtained according to actual needs, the checkerboard image can be replaced by a required image, and subsequent image processing is facilitated. The method well avoids errors generated in the shooting process and has very important significance in image processing.
Claims (1)
1. A high-precision image registration method for blur kernel estimation in an imaging system, comprising the steps of:
the method comprises the following steps: generating an original checkerboard image A on a computer;
step two: shooting a checkerboard image A on a computer screen by using imaging equipment to obtain a shot checkerboard image B;
step three: carrying out angular point detection on the checkerboard images A and B to obtain a coordinate matrix of the same angular points of the image A and the image BMat1 andMat2; angular point coordinate matrixMat1 andMat2 is 2 xRow*ColWhereinRowRepresenting the number of transverse chequers in the checkerboard image used,Colrepresenting the number of columns in the checkerboard image;
step four: calculating and solving the coordinate corresponding relation between the original checkerboard image and the shot checkerboard image, and mapping the original checkerboard image to the corresponding area of the shot checkerboard image according to the corresponding coordinate relation, so as to obtain the precise and accurate matched clear and actual shot checkerboard image;
aiming at the angular point coordinate matrix obtained in the third stepMat1 andMat2, 1iThe coordinate indices of the four corner points of the checkerboard, up, down, left, and right, can be expressed as:
(1)
wherein,c1P、c2P、c3Pandc4Prespectively representiCoordinate indexes of four corner points of the upper part, the lower part, the left part and the right part of the checkerboard are obtained;floorthe operation symbol of the nearest lower integer is taken;Rowrepresenting the number of transverse chessboards in the chessboard image;
the index value obtained by the formula (1) can be respectively obtained from the corner point coordinate matrix obtained in the third stepMat1AndMat2obtaining the original chessboard image and shooting the second in the chessboard imageiThe coordinates corresponding to each checkerboard are shown in formula (2) and formula (3):
(2)
(3)
wherein,c1、c2、c3 andc4 respectively representing the original chessboard patternsiCoordinates of four corner points of upper, lower, left and right corresponding to each checkerboard,cc1、cc2、cc3 andcc4 respectively representing the chessboard picturesiCoordinates of four corner points, namely, an upper corner point, a lower corner point, a left corner point, a right corner point and a right corner point, which correspond to the checkerboards;
from the first in the original chessboard imageiThe coordinates corresponding to each checkerboard may result in parameters α and β:
(4)
wherein,c1(2) representing coordinate points of original chessboard imagecOrdinate of 1, β denotes coordinate points for capturing checkerboard imagesc3, the abscissa;
based on the above parameters, the corresponding relation between the original chessboard image and the photographed chessboard image can be found:
(5)
wherein,cc1、cc2、cc3 andcc4 respectively representing the chessboard picturesiThe coordinates of the four corner points of the upper, lower, left and right sides corresponding to each checkerboardx,y]Representing the corresponding coordinates mapped to the original checkerboard image by one point on the shooting checkerboard;
the original checkerboard images can be mapped to the corresponding positions of the shot checkerboard images one by one according to the corresponding relation (5), so that the precisely registered checkerboard images are obtained.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710175300.7A CN107146242A (en) | 2017-03-22 | 2017-03-22 | A kind of high precision image method for registering that kernel estimates are obscured for imaging system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710175300.7A CN107146242A (en) | 2017-03-22 | 2017-03-22 | A kind of high precision image method for registering that kernel estimates are obscured for imaging system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107146242A true CN107146242A (en) | 2017-09-08 |
Family
ID=59784011
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710175300.7A Pending CN107146242A (en) | 2017-03-22 | 2017-03-22 | A kind of high precision image method for registering that kernel estimates are obscured for imaging system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107146242A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107730469A (en) * | 2017-10-17 | 2018-02-23 | 长沙全度影像科技有限公司 | A kind of three unzoned lens image recovery methods based on convolutional neural networks CNN |
CN107833186A (en) * | 2017-10-26 | 2018-03-23 | 长沙全度影像科技有限公司 | A kind of simple lens spatial variations image recovery method based on Encoder Decoder deep learning models |
CN107833193A (en) * | 2017-11-20 | 2018-03-23 | 长沙全度影像科技有限公司 | A kind of simple lens global image restored method based on refinement network deep learning models |
CN107823883A (en) * | 2017-11-21 | 2018-03-23 | 河南黄烨科技有限公司 | Aiming point screen coordinate acquisition methods based on image recognition and laser positioning |
CN114612580A (en) * | 2022-03-15 | 2022-06-10 | 中国人民解放军国防科技大学 | High-definition imaging method for low-quality camera |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101321303A (en) * | 2008-07-17 | 2008-12-10 | 上海交通大学 | Geometric and optical correction method for non-plane multi-projection display |
CN102769771A (en) * | 2011-05-05 | 2012-11-07 | 友达光电股份有限公司 | Testing system and testing method for testing photographic equipment |
CN103019643A (en) * | 2012-12-30 | 2013-04-03 | 中国海洋大学 | Method for automatic correction and tiled display of plug-and-play large screen projections |
CN105303574A (en) * | 2015-07-30 | 2016-02-03 | 四川大学 | Integrated imaging camera array calibration method based on homography transformation |
CN105959669A (en) * | 2016-06-06 | 2016-09-21 | 四川大学 | Remapping-based integral imaging micro-image array rapid generation method |
-
2017
- 2017-03-22 CN CN201710175300.7A patent/CN107146242A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101321303A (en) * | 2008-07-17 | 2008-12-10 | 上海交通大学 | Geometric and optical correction method for non-plane multi-projection display |
CN102769771A (en) * | 2011-05-05 | 2012-11-07 | 友达光电股份有限公司 | Testing system and testing method for testing photographic equipment |
CN103019643A (en) * | 2012-12-30 | 2013-04-03 | 中国海洋大学 | Method for automatic correction and tiled display of plug-and-play large screen projections |
CN105303574A (en) * | 2015-07-30 | 2016-02-03 | 四川大学 | Integrated imaging camera array calibration method based on homography transformation |
CN105959669A (en) * | 2016-06-06 | 2016-09-21 | 四川大学 | Remapping-based integral imaging micro-image array rapid generation method |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107730469A (en) * | 2017-10-17 | 2018-02-23 | 长沙全度影像科技有限公司 | A kind of three unzoned lens image recovery methods based on convolutional neural networks CNN |
CN107833186A (en) * | 2017-10-26 | 2018-03-23 | 长沙全度影像科技有限公司 | A kind of simple lens spatial variations image recovery method based on Encoder Decoder deep learning models |
CN107833193A (en) * | 2017-11-20 | 2018-03-23 | 长沙全度影像科技有限公司 | A kind of simple lens global image restored method based on refinement network deep learning models |
CN107823883A (en) * | 2017-11-21 | 2018-03-23 | 河南黄烨科技有限公司 | Aiming point screen coordinate acquisition methods based on image recognition and laser positioning |
CN114612580A (en) * | 2022-03-15 | 2022-06-10 | 中国人民解放军国防科技大学 | High-definition imaging method for low-quality camera |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107146242A (en) | A kind of high precision image method for registering that kernel estimates are obscured for imaging system | |
CN110717942B (en) | Image processing method and device, electronic equipment and computer readable storage medium | |
WO2018209968A1 (en) | Camera calibration method and system | |
CN107886547B (en) | Fisheye camera calibration method and system | |
CN106887023A (en) | For scaling board and its scaling method and calibration system that binocular camera is demarcated | |
CN110660107A (en) | Plane calibration plate, calibration data acquisition method and system | |
WO2021136386A1 (en) | Data processing method, terminal, and server | |
CN109920003B (en) | Camera calibration detection method, device and equipment | |
CN113012234B (en) | High-precision camera calibration method based on plane transformation | |
CN102194223B (en) | A kind of distortion factor scaling method of zoom lens and system | |
JP2013113600A (en) | Stereo three-dimensional measuring apparatus | |
CN112396663B (en) | Visual calibration method, device, equipment and medium for multi-depth camera | |
CN109724537B (en) | Binocular three-dimensional imaging method and system | |
CN111681186A (en) | Image processing method and device, electronic equipment and readable storage medium | |
CN110766615A (en) | Picture correction method, device, terminal and computer readable storage medium | |
CN113538590B (en) | Calibration method and device of zoom camera, terminal equipment and storage medium | |
CN112381887A (en) | Multi-depth camera calibration method, device, equipment and medium | |
CN113838151B (en) | Camera calibration method, device, equipment and medium | |
KR102023087B1 (en) | Method for camera calibration | |
CN117196955A (en) | Panoramic image stitching method and terminal | |
CN112014408A (en) | Detection method for reconstructing pcb (printed circuit board) based on structured light principle | |
CN115457142B (en) | Calibration method and system of MR hybrid photographic camera | |
CN111833441A (en) | Face three-dimensional reconstruction method and device based on multi-camera system | |
CN116071562A (en) | Plant seed identification method and device, electronic equipment and storage medium | |
CN115564845A (en) | Regional binocular camera calibration method |
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 |
Application publication date: 20170908 |
|
WD01 | Invention patent application deemed withdrawn after publication |