CN107424118A - Based on the spherical panorama mosaic method for improving Lens Distortion Correction - Google Patents
Based on the spherical panorama mosaic method for improving Lens Distortion Correction Download PDFInfo
- Publication number
- CN107424118A CN107424118A CN201710192145.XA CN201710192145A CN107424118A CN 107424118 A CN107424118 A CN 107424118A CN 201710192145 A CN201710192145 A CN 201710192145A CN 107424118 A CN107424118 A CN 107424118A
- Authority
- CN
- China
- Prior art keywords
- image
- camera
- correction
- formula
- distortion correction
- 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
- 238000012937 correction Methods 0.000 title claims abstract description 73
- 238000000034 method Methods 0.000 title claims abstract description 20
- 230000006735 deficit Effects 0.000 claims abstract description 4
- 238000002156 mixing Methods 0.000 claims abstract description 4
- 230000004927 fusion Effects 0.000 claims description 10
- 230000006872 improvement Effects 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 8
- 238000011161 development Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 5
- 238000003702 image correction Methods 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 4
- 230000003287 optical effect Effects 0.000 claims description 4
- 230000000007 visual effect Effects 0.000 claims description 4
- 230000004075 alteration Effects 0.000 claims description 3
- 238000007654 immersion Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 230000008447 perception Effects 0.000 claims description 3
- 238000012876 topography Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 claims description 2
- 230000017105 transposition Effects 0.000 claims description 2
- 238000012544 monitoring process Methods 0.000 abstract description 10
- 230000007547 defect Effects 0.000 abstract description 4
- 230000007812 deficiency Effects 0.000 abstract description 2
- 230000008030 elimination Effects 0.000 abstract description 2
- 238000003379 elimination reaction Methods 0.000 abstract description 2
- 238000012360 testing method Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 241000251468 Actinopterygii Species 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000002045 lasting effect Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000000717 retained effect Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000003760 hair shine Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 238000009966 trimming Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4038—Scaling the whole image or part thereof for image mosaicing, i.e. plane images composed of plane sub-images
-
- G06T5/80—
Abstract
The invention belongs to computer vision field, for overcome the deficiencies in the prior art, for traditional camera because resolution ratio is low, monitoring range is small, distortion is larger and can not meet the defects of wide area persistently monitors demand, proposes a kind of based on the spherical panoramic mosaic technology for improving Lens Distortion Correction.The technical solution adopted by the present invention is, based on the spherical panorama mosaic method for improving Lens Distortion Correction, image impairment caused by by improving the common distortion correction of Lens Distortion Correction elimination, then panoramic mosaic is completed by spherical stitching algorithm, it is finally based on weighted mean method and light equilibrium blending algorithm eliminates the light difference of repeat region interference and different cameras, realizes high-resolution and wide area shooting.Present invention is mainly applied to computer vision to handle occasion.
Description
Technical field
The invention belongs to computer vision field, and it is a kind of based on improvement radial direction that application demand invention is persistently monitored for wide area
The spherical panoramic mosaic technology of distortion correction.
Background technology
With the rapid development of society, market quickly increases the demand of video monitoring.In video monitoring, wide area is held
Monitoring is important part long, and important function is served in important departments such as army, public security and traffic.Wide area persistently monitors
It is required that camera coverage is big, high resolution.The single resolution ratio of camera head of tradition is limited, and camera wide-angle is smaller, it is difficult to
Monitor large area;Although fisheye type camera lens can realize overall view monitoring, lens distortion is excessive, causes monitored object to be imaged
It is poor, cause very big difficulty for follow-up video analysis.Therefore, single camera and fish eye lens are all difficult to meet that wide area is lasting
Monitoring requirement.
The content of the invention
For overcome the deficiencies in the prior art, for traditional camera because resolution ratio is low, monitoring range is small, distortion is larger and nothing
Method meets the defects of wide area persistently monitors demand, proposes a kind of based on the spherical panoramic mosaic technology for improving Lens Distortion Correction.
The technical solution adopted by the present invention is, radially abnormal by improving based on the spherical panorama mosaic method for improving Lens Distortion Correction
Become correction and eliminate image impairment caused by common distortion correction, panoramic mosaic, last base are then completed by spherical stitching algorithm
The light difference of repeat region interference and different cameras is eliminated in weighted mean method and light equilibrium blending algorithm, realizes high-resolution
Rate and wide area shooting.
Further comprised the concrete steps that in one example,
(1) offline camera calibration step:8 cameras are demarcated respectively using optics black and white lattice camera calibration plate,
Obtain each photocentre position, focal length, pixel length-width ratio, distortion factor and visual field angular dimensions;
(2) video acquisition module:Video acquisition module is based on camera lens and sensor module realizes multichannel high-definition picture
Synchronous acquisition and multichannel high-resolution video real-time decoding;
(3) based on improvement Lens Distortion Correction step:Calibrating parameters of this module based on each camera realize pattern distortion
Correction, shown in conventional radial aberration correction algorithm formula such as formula (1).
U=us+(us-u0)[k1(x2+y2)+k2(x2+y2)2]
V=vs+(vs-v0)[k1(x2+y2)+k2(x2+y2)2] formula (1)
Wherein, (u, v) represents the pixel coordinate after correction distortion, (xs,vs) represent image in the case of actual radial distortion
Pixel coordinate, (u0,v0) represent the principal point for camera coordinate that offline camera calibration obtains, k1, k2The distortion of two ranks is joined before representing camera
Number, consecutive image coordinate when (x, y) represents preferable undistorted, using improvement Lens Distortion Correction module, formula such as formula (2) institute
Show:
U=us+(us-u0*d)[k1(x2+y2)+k2(x2+y2)2]
V=vs+(vs-v0*d)[k1(x2+y2)+k2(x2+y2)2] formula (2)
Embodiment is as follows:
First, new image array is established based on scale factor, newly-established image array removes upper left corner a quarter face
Product is source image pixels value, and other scopes are full null matrix, and new images grow 2 times of a width of original image;Then by figure to be corrected
As moving to new images central area, image Lens Distortion Correction is then carried out, the angle point picture in image is obtained after correction
Plain position, extraction rectangular area are removed as the image after correction, zero pixel region by overlapping region Fusion Module, final real
Image correction function in the case of present non-flanged loss;
(4) panoramic mosaic module:This module by the image after correction by known photographing information, to determine multiple image
Matching connection collection, image after correction is subjected to deformation, projection, during projection, the topography of each video camera will project
On the sphere unified to one so that image has certain depth perception, can partly reflect entirely regarding when viewpoint changes
The space of points, the three-dimensional feeling of immersion more strong to observer, when being spliced using spheric coordinate system, it that is to say each image
Point transformation is a bit on sphere, shown in mapping equation such as formula (3),
Wherein, the plane coordinates of (x, y) representative image, (x ', y ') represent development of a sphere plane coordinates, and f represents that camera is burnt
Away from.Camera is fixed during pan-shot, and it can regard an occurred level rotation, vertical rotation as, not occur to rotate around optical axis, therefore phase
Machine X-axis would generally be maintained at a plane, and the normal direction u of this plane is just used as a preferable major axes orientation, obtained main shaft
Direction u is vertical with all image coordinate system X-axis, therefore obtains formula (4), and i represents camera sequence number,Represent i-th of camera institute
In plane transposed matrix, T represents transposition;
Write as least squares formalism and obtain formula 5:
This is a Linear least squares minimization problem, and optimal solution u meets formula 6:
U is 3 × 3 matrixesCharacteristic vector corresponding to minimal eigenvalue, determine that sphere is sat using u as major axes orientation
Mark system, all images are transformed to according to camera parameter and complete to splice on sphere, obtain panorama sketch;
(5) Fusion Module is spliced:This module is primarily based on zero picture that overlapping region coordinate position interception eliminates correction tape
Plain region, is then based on image weighting average and light balance policy realizes that splicing regions merge, image weighting average formula such as formula
(7) shown in:
G (x+i, y)=af1(x+i,y)+(1-a)f2(x+i, y) formula (7)
Wherein, f1(x, y) and f2(x, y) represents the pixel value of left and right two images respectively, and g (x, y) represents that overlapping region is melted
Pixel value after conjunction, i represents the ith pixel of overlapping region, if the coincidence length is L, then
Because the brightness of image of each camera shooting is inconsistent, so also needing to carry out light equilibrium, cause splicing as far as possible
Each regional luminance difference of image is minimum afterwards, and specific implementation method is:All average gray values for wanting stitching image are obtained first, so
Each width image intensity value contrasts with average gray value afterwards, obtains respective difference delta, will be respective while splicing
Delta is compensated to the gray value of correspondence image so that overall partially dark image can suitably brighten, and overall very bright image can be with
It is appropriate dimmed.
Image mosaic is with merging by the way of tabling look-up, i.e., each picture photo vegetarian refreshments computed in advance and picture after splicing
Pixel corresponding relation, the corresponding relation is stored in file, this document is directly read when subsequently splicing with merging.
The features of the present invention and beneficial effect are:
The present invention realizes that image is abnormal using Lens Distortion Correction algorithm is improved in the case where image edge information does not lose
Become correction, zero pixel region then brought by the overlapping shooting area removal correcting algorithm of adjacent cameras so that adjacent cameras
The image of shooting can realize lossless accurate splicing, and providing technology for panoramic mosaic technology ensures.
Brief description of the drawings:
Fig. 1:Based on the spherical panoramic mosaic technology frame chart for improving Lens Distortion Correction;
Fig. 2:Improve Lens Distortion Correction algorithm pattern.
Embodiment
For the present invention for traditional camera because resolution ratio is low, monitoring range is small, distortion is larger and can not meet that wide area is lasting
The defects of monitoring demand, propose a kind of based on the spherical panoramic mosaic technology for improving Lens Distortion Correction.The skill that the present invention uses
Art scheme is:Image impairment caused by by improving the common distortion correction of Lens Distortion Correction elimination, then passes through spherical splicing
Algorithm completes panoramic mosaic, is finally based on weighted mean method and light equilibrium blending algorithm eliminates repeat region interference and different shootings
The light difference of head, high-resolution and wide area shooting are realized, solves that single resolution ratio of camera head is low, monitoring range is small, fish
The defects of distortion of Eye imaging head is big, realizes that wide area persistently monitors.
Specific steps are further refined as:
To solve the problems, such as edge penalty that General calibration algorithm is brought, the present invention improves Lens Distortion Correction module, improves
Formula is as follows:
U=us+(us-u0*d)[k1(x2+y2)+k2(x2+y2)2]
V=vs+(vs-v0*d)[k1(x2+y2)+k2(x2+y2)2]
The present invention adds scale factor d in original distortion correction formula.First, new figure is established based on scale factor
As matrix, because edge penalty scope is smaller caused by distortion correction, therefore scale factor elects 2 as can meet majority of case
It is required that newly-established image array is source image pixels value except upper left corner a quarter area, other scopes are full null matrix, newly
Image grows 2 times of a width of original image;Then by image translation to be corrected to new images central area, image is then carried out radially
Distortion correction, four corner pixels positions in image are obtained after correction, extraction rectangular area is as the image after correction.This
The complete information of image is all retained after the correction that sample obtains, but can introduce zero pixel region at edge, due to adjacent phase
Machine shooting area has lap, therefore zero pixel region can be removed by overlapping region Fusion Module, final to realize
Image correction function in the case of non-flanged loss.
It is proposed by the present invention as shown in Figure 1 based on the spherical panoramic mosaic technology frame chart for improving Lens Distortion Correction.It is overall
Block diagram is divided into five big modules, and offline camera calibration module, video acquisition module improve Lens Distortion Correction module, and panorama is spelled
Connection module and splicing Fusion Module.Camera calibration module is demarcated using optical camera scaling board to each camera, is obtained
The parameters such as its photocentre position, focal length, pixel length-width ratio, distortion factor;Video acquisition module gathers 8 tunnels using camera module and regarded
Frequency evidence;Distortion correction module is based on improving Lens Distortion Correction algorithm and camera parameter information completes distortion correction;Panorama is spelled
Connection module is based on sphere change, 8 road video-splicings are realized in projection;Fusion Module is based on weighted mean method and light equalization algorithm merges
Repeat region, it is final to obtain output video.Its key step is as follows:
(1) offline camera calibration module:This module is carried out respectively using optics black and white lattice camera calibration plate to 8 cameras
Demarcation, obtains each photocentre position, focal length, pixel length-width ratio, distortion factor and visual field angular dimensions.First, taken the photograph for each
As head, the scaling board photo of 20 frame different angles, different distance, different deviations is shot using camera module, then detects and shines
Angle point (intersection points of black and white lattice) position in piece, the internal reference matrix parameter of camera is calculated using Zhang Zhengyou calibration algorithms, wrapped
Include the transverse and longitudinal coordinate of principal point, the scale factor on transverse axis, y direction, and between world coordinate system and camera coordinates system before
Two rank distortion parameters.By demarcation, the average re-projection error of camera is 0.14 pixel, and demarcation works well.
(2) video acquisition module:Video acquisition module is based on camera lens and sensor module realizes multichannel high-definition picture
Synchronous acquisition and multichannel high-resolution video real-time decoding, the module can realize Real Time Compression and output more than 10M pixels.
(3) based on improvement Lens Distortion Correction module:Calibrating parameters of this module based on each camera realize pattern distortion
Correction, conventional radial aberration correction algorithm formula are as shown in Equation 1.
U=us+(us-u0)[k1(x2+y2)+k2(x2+y2)2]
V=vs+(vs-v0)[k1(x2+y2)+k2(x2+y2)2] formula (1)
Wherein, (u, v) represents the pixel coordinate after correction distortion, (xs,vs) represent image in the case of actual radial distortion
Pixel coordinate, (u0,v0) represent the principal point for camera coordinate that offline camera calibration obtains, k1, k2The distortion of two ranks is joined before representing camera
Number, consecutive image coordinate when (x, y) represents preferable undistorted.Conventional correction algorithm can realize distortion correction, but image
Edge due in trimming process part pixel correct recoil village outside image coordinate scope, therefore be ignored, cause
Edge penalty.The loss at edge can cause the image of No. 8 camera shootings can not realize panoramic mosaic because of loss of learning.To solve
Edge penalty problem, the present invention propose improvement Lens Distortion Correction module, and improvement project is as shown in Fig. 2 formula is as shown in Equation 2.
U=us+(us-u0*d)[k1(x2+y2)+k2(x2+y2)2]
V=vs+(vs-v0*d)[k1(x2+y2)+k2(x2+y2)2] formula (2)
The present invention adds scale factor d in original distortion correction formula, and embodiment is as follows:
First, new image array is established based on scale factor, because edge penalty scope is smaller caused by distortion correction,
Therefore scale factor, which elects 2 as, can meet majority of case requirement, and newly-established image array removes upper left corner a quarter area
For source image pixels value, other scopes are full null matrix, and new images grow 2 times of a width of original image;Then by image to be corrected
New images central area is moved to, then carries out image Lens Distortion Correction, the corner pixels in image are obtained after correction
Position, extraction rectangular area is as the image after correction.The complete information of image is all retained after the correction so obtained, but
It is zero pixel region that can introduce edge, because adjacent cameras shooting area has a lap, therefore zero pixel region can be with
Removed by overlapping region Fusion Module, the final image correction function realized in the case where non-flanged is lost.
(4) panoramic mosaic module:This module by the image after correction by known photographing information, to determine multiple image
Matching connection collection.Then image carries out deformation, projection after correcting.During projection, the topography of each video camera will
Project on a unified sphere so that image has certain depth perception, can partly reflect when viewpoint changes whole
Individual view space, the three-dimensional feeling of immersion more strong to observer.When being spliced using spheric coordinate system, that is to say will be each
Picture point is transformed on sphere a bit, and mapping equation is as shown in Equation 3.
Wherein, the plane coordinates of (x, y) representative image, (x ', y ') represent development of a sphere plane coordinates, and f represents that camera is burnt
Away from.Camera is fixed during pan-shot, and it can regard an occurred level rotation, vertical rotation as, not occur to rotate around optical axis, therefore phase
Machine X-axis would generally be maintained at a plane, and the normal direction u of this plane is just used as a preferable major axes orientation.Obtained main shaft
Direction u is vertical with all image coordinate system X-axis, therefore obtains formula 4 (i represents camera sequence number).
Write as least squares formalism and obtain formula 5:
This is a Linear least squares minimization problem, and optimal solution u meets formula 6:
U is 3 × 3 matrixesCharacteristic vector corresponding to minimal eigenvalue.Determine that sphere is sat using u as major axes orientation
Mark system, all images are transformed to according to camera parameter and complete to splice on sphere, obtain panorama sketch.
(5) Fusion Module is spliced:This module is primarily based on zero picture that overlapping region coordinate position interception eliminates correction tape
Plain region, is then based on image weighting average and light balance policy realizes that splicing regions merge, image weighting average formula such as formula 7
It is shown.
G (x+i, y)=af1(x+i,y)+(1-a)f2(x+i, y) formula (7)
Wherein, f1(x, y) and f2(x, y) represents the pixel value of left and right two images respectively, and g (x, y) represents that overlapping region is melted
Pixel value after conjunction, i represent the ith pixel of overlapping region.If the coincidence length is L, then
Because the brightness of image of each camera shooting is inconsistent, therefore need to carry out light equilibrium, scheme after causing splicing as far as possible
As each regional luminance difference is minimum.Specific implementation method is:All average gray values for wanting stitching image are obtained first, then often
Piece image gray value contrasts with average gray value, obtains respective difference delta, mends respective delta while splicing
Repay to the gray value of correspondence image so that overall partially dark image can suitably brighten, and overall very bright image can suitably become
Secretly.This method realizes that simply computation complexity is low.
In terms of algorithm acceleration, it will not change because each camera position is relative, therefore the position of each picture and coincidence
Region will not also change substantially, so for speed up processing, image mosaic is with merging by the way of tabling look-up.It is i.e. prior
The corresponding relation of each picture photo vegetarian refreshments and the pixel of picture after splicing is calculated, the corresponding relation is stored in file,
This document is directly read when subsequently splicing with merging, real-time calculating process is eliminated, greatly improves processing speed.
Spherical panoramic mosaic technology proposed by the present invention based on improvement Lens Distortion Correction, its test environment and experiment number
According to for:
Test environment:
System environments:Windows7;
Depended software:Visual Studio 2013, Matlab 2015, OpenCV 2.4.9;
Hardware environment:Algorithm development:CPU:I5-4300M@2.60GHz, internal memory:4GB, hard disk:1TB;
Board model:AHB70XXT8-3521;
Camera lens:3.6mm, CMOS, 1/3 inch of sensor.
Experimental data
Our unit data of the test experiments data of the present invention from camera collection module shooting, test result prove this hair
Bright to realize precise image correction and panoramic mosaic, embedded system can gather 8 road videos and real-time exhibition panorama in real time
Splicing result, practicality are good.
Claims (4)
1. it is a kind of based on the spherical panorama mosaic method for improving Lens Distortion Correction, it is characterized in that, by improving radial distortion school
Image impairment caused by just eliminating common distortion correction, panoramic mosaic is then completed by spherical stitching algorithm, be finally based on and add
Weight average method and light equilibrium blending algorithm eliminate the light difference of repeat region interference and different cameras, realize high-resolution with
And wide area shooting.
2. as claimed in claim 1 based on the spherical panorama mosaic method for improving Lens Distortion Correction, it is characterized in that, a reality
Further comprised the concrete steps that in example,
(1) offline camera calibration step:8 cameras are demarcated respectively using optics black and white lattice camera calibration plate, obtained
Each photocentre position, focal length, pixel length-width ratio, distortion factor and visual field angular dimensions;
(2) video acquisition module:Video acquisition module is based on camera lens and sensor module realizes the synchronization of multichannel high-definition picture
Collection and multichannel high-resolution video real-time decoding;
(3) based on improvement Lens Distortion Correction step:Calibrating parameters of this module based on each camera realize image distortion correction,
Shown in conventional radial aberration correction algorithm formula such as formula (1).
U=us+(us-u0)[k1(x2+y2)+k2(x2+y2)2]
V=vs+(vs-v0)[k1(x2+y2)+k2(x2+y2)2] formula (1)
Wherein, (u, v) represents the pixel coordinate after correction distortion, (xs,vs) represent the picture of image in the case of actual radial distortion
Plain coordinate, (u0,v0) represent the principal point for camera coordinate that offline camera calibration obtains, k1, k2Two rank distortion parameters before camera are represented,
Consecutive image coordinate when (x, y) represents preferable undistorted, using Lens Distortion Correction module is improved, formula such as formula (2) is shown:
U=us+(us-u0*d)[k1(x2+y2)+k2(x2+y2)2]
V=vs+(vs-v0*d)[k1(x2+y2)+k2(x2+y2)2] formula (2)
Embodiment is as follows:
First, new image array is established based on scale factor, newly-established image array is except upper left corner a quarter area
Source image pixels value, other scopes are full null matrix, and new images grow 2 times of a width of original image;Then image to be corrected is put down
New images central area is moved to, then carries out image Lens Distortion Correction, the corner pixels position in image is obtained after correction
Put, extraction rectangular area is removed as the image after correction, zero pixel region by overlapping region Fusion Module, final to realize
Image correction function in the case of non-flanged loss;
(4) panoramic mosaic module:This module by the image after correction by known photographing information, to determine the matching of multiple image
Connection collection, image after correction is subjected to deformation, projection, during projection, the topography of each video camera will project to one
On individual unified sphere so that image has certain depth perception, can partly reflect that whole viewpoint is empty when viewpoint changes
Between, the three-dimensional feeling of immersion more strong to observer, when being spliced using spheric coordinate system, it that is to say and become each picture point
A bit is changed on sphere, shown in mapping equation such as formula (3),
<mrow>
<msup>
<mi>x</mi>
<mo>&prime;</mo>
</msup>
<mo>=</mo>
<mi>s</mi>
<mi> </mi>
<msup>
<mi>tan</mi>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msup>
<mfrac>
<mi>x</mi>
<mi>f</mi>
</mfrac>
</mrow>
Wherein, the plane coordinates of (x, y) representative image, (x ', y ') represent development of a sphere plane coordinates, and f represents camera focus.Entirely
Camera is fixed when scape is shot, and it can regard an occurred level rotation, vertical rotation as, not occur to rotate around optical axis, therefore camera X-axis
A plane would generally be maintained at, the normal direction u of this plane is just used as a preferable major axes orientation, obtained major axes orientation u
It is vertical with all image coordinate system X-axis, therefore formula (4) is obtained, i represents camera sequence number,Plane where representing i-th of camera
Transposed matrix, T represent transposition;
Write as least squares formalism and obtain formula 5:
This is a Linear least squares minimization problem, and optimal solution u meets formula 6:
<mrow>
<mo>&dtri;</mo>
<msubsup>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mn>8</mn>
</msubsup>
<mo>|</mo>
<mo>|</mo>
<msubsup>
<mi>X</mi>
<mi>i</mi>
<mi>T</mi>
</msubsup>
<mi>u</mi>
<mo>|</mo>
<mo>|</mo>
<mo>=</mo>
<mn>0</mn>
</mrow>
U is 3 × 3 matrixesCharacteristic vector corresponding to minimal eigenvalue, spherical coordinate is determined using u as major axes orientation
System, all images are transformed to according to camera parameter and complete to splice on sphere, obtain panorama sketch;
(5) Fusion Module is spliced:This module is primarily based on zero pixel region that overlapping region coordinate position interception eliminates correction tape
Domain, is then based on image weighting average and light balance policy realizes that splicing regions merge, image weighting average formula such as formula (7) institute
Show:
G (x+i, y)=af1(x+i,y)+(1-a)f2(x+i, y) formula (7)
Wherein, f1(x, y) and f2After (x, y) represents that the pixel value of left and right two images, g (x, y) represent overlapping region fusion respectively
Pixel value, i represent overlapping region ith pixel, if the coincidence length is L, then0≤i≤L。
3. as claimed in claim 1 based on the spherical panorama mosaic method for improving Lens Distortion Correction, it is characterized in that, due to every
The brightness of image of individual camera shooting is inconsistent, so also needing to carry out light equilibrium, each region of image is bright after causing splicing as far as possible
It is minimum to spend difference, specific implementation method is:All average gray values for wanting stitching image are obtained first, then the ash per piece image
Angle value contrasts with average gray value, obtains respective difference delta, compensates respective delta to corresponding diagram while splicing
The gray value of picture so that overall partially dark image can suitably brighten, and overall very bright image can be suitably dimmed.
4. as claimed in claim 1 based on the spherical panorama mosaic method for improving Lens Distortion Correction, it is characterized in that, image is spelled
Connect with merging by the way of tabling look-up, i.e., each picture photo vegetarian refreshments computed in advance with splice after picture pixel it is corresponding
Relation, the corresponding relation is stored in file, this document is directly read when subsequently splicing with merging.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710192145.XA CN107424118A (en) | 2017-03-28 | 2017-03-28 | Based on the spherical panorama mosaic method for improving Lens Distortion Correction |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710192145.XA CN107424118A (en) | 2017-03-28 | 2017-03-28 | Based on the spherical panorama mosaic method for improving Lens Distortion Correction |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107424118A true CN107424118A (en) | 2017-12-01 |
Family
ID=60423076
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710192145.XA Pending CN107424118A (en) | 2017-03-28 | 2017-03-28 | Based on the spherical panorama mosaic method for improving Lens Distortion Correction |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107424118A (en) |
Cited By (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108122200A (en) * | 2017-12-20 | 2018-06-05 | 宁波视睿迪光电有限公司 | Image split-joint method and device |
CN108734681A (en) * | 2018-05-31 | 2018-11-02 | 天津煋鸟科技有限公司 | One kind being based on Embedded on piece radial distortion bearing calibration |
CN108760767A (en) * | 2018-05-31 | 2018-11-06 | 电子科技大学 | Large-size LCD Screen defect inspection method based on machine vision |
CN109272445A (en) * | 2018-10-29 | 2019-01-25 | 中国航空无线电电子研究所 | Panoramic video joining method based on Sphere Measurement Model |
CN110099220A (en) * | 2019-06-17 | 2019-08-06 | 广东中星微电子有限公司 | A kind of panorama mosaic method and device |
CN110189263A (en) * | 2019-05-05 | 2019-08-30 | 浙江大学 | It is a kind of based on multi-angle sampling big visual field wear display equipment distortion correction method |
CN110580722A (en) * | 2018-06-08 | 2019-12-17 | 上海欧菲智能车联科技有限公司 | Calibration test method and calibration test device for panoramic system and vehicle |
CN110619611A (en) * | 2019-09-09 | 2019-12-27 | 北京迈格威科技有限公司 | Image correction calibration method and device, computer equipment and storage medium |
CN110691203A (en) * | 2019-10-21 | 2020-01-14 | 湖南泽天智航电子技术有限公司 | Multi-path panoramic video splicing display method and system based on texture mapping |
CN110942023A (en) * | 2019-11-25 | 2020-03-31 | 鹰驾科技(深圳)有限公司 | Indication method, device and equipment for vehicle vision blind area and storage medium |
CN111507894A (en) * | 2020-04-17 | 2020-08-07 | 浙江大华技术股份有限公司 | Image splicing processing method and device |
CN112037128A (en) * | 2020-08-21 | 2020-12-04 | 苏州巨能图像检测技术有限公司 | Panoramic video splicing method |
WO2021036353A1 (en) * | 2019-08-23 | 2021-03-04 | 上海亦我信息技术有限公司 | Photographing-based 3d modeling system and method, and automatic 3d modeling apparatus and method |
CN112884689A (en) * | 2021-02-25 | 2021-06-01 | 景德镇陶瓷大学 | Highlight removing method for image on strong reflection surface |
CN113139900A (en) * | 2021-04-01 | 2021-07-20 | 北京科技大学设计研究院有限公司 | Method for acquiring complete surface image of bar |
CN113191974A (en) * | 2021-04-29 | 2021-07-30 | 青岛科技大学 | Method and system for obtaining ship panoramic image based on machine vision |
GB2591857A (en) * | 2019-08-23 | 2021-08-11 | Shang Hai Yiwo Information Tech Co Ltd | Photographing-based 3D modeling system and method, and automatic 3D modeling apparatus and method |
CN113724141A (en) * | 2020-05-26 | 2021-11-30 | 杭州海康威视数字技术股份有限公司 | Image correction method and device and electronic equipment |
CN114143528A (en) * | 2020-09-04 | 2022-03-04 | 北京大视景科技有限公司 | Multi-video stream fusion method, electronic device and storage medium |
CN115471482A (en) * | 2022-09-20 | 2022-12-13 | 重庆理工大学 | Small-caliber container inner wall defect detection method based on computer vision |
CN115500582A (en) * | 2022-08-17 | 2022-12-23 | 上海科技大学 | Foot three-dimensional contour acquisition system |
WO2023272457A1 (en) * | 2021-06-28 | 2023-01-05 | 华为技术有限公司 | Apparatus and system for image splicing, and related method |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100485720C (en) * | 2006-10-13 | 2009-05-06 | 浙江师范大学 | 360 degree around panorama generation method based on serial static image |
CN102984453A (en) * | 2012-11-01 | 2013-03-20 | 深圳大学 | Method and system of real-time generating hemisphere panoramic video images through single camera |
CN103971352A (en) * | 2014-04-18 | 2014-08-06 | 华南理工大学 | Rapid image splicing method based on wide-angle lenses |
CN105957015A (en) * | 2016-06-15 | 2016-09-21 | 武汉理工大学 | Thread bucket interior wall image 360 DEG panorama mosaicing method and system |
CN106303283A (en) * | 2016-08-15 | 2017-01-04 | Tcl集团股份有限公司 | A kind of panoramic image synthesis method based on fish-eye camera and system |
CN105282443B (en) * | 2015-10-13 | 2019-06-14 | 哈尔滨工程大学 | A kind of panorama depth panoramic picture imaging method |
-
2017
- 2017-03-28 CN CN201710192145.XA patent/CN107424118A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100485720C (en) * | 2006-10-13 | 2009-05-06 | 浙江师范大学 | 360 degree around panorama generation method based on serial static image |
CN102984453A (en) * | 2012-11-01 | 2013-03-20 | 深圳大学 | Method and system of real-time generating hemisphere panoramic video images through single camera |
CN103971352A (en) * | 2014-04-18 | 2014-08-06 | 华南理工大学 | Rapid image splicing method based on wide-angle lenses |
CN105282443B (en) * | 2015-10-13 | 2019-06-14 | 哈尔滨工程大学 | A kind of panorama depth panoramic picture imaging method |
CN105957015A (en) * | 2016-06-15 | 2016-09-21 | 武汉理工大学 | Thread bucket interior wall image 360 DEG panorama mosaicing method and system |
CN106303283A (en) * | 2016-08-15 | 2017-01-04 | Tcl集团股份有限公司 | A kind of panoramic image synthesis method based on fish-eye camera and system |
Non-Patent Citations (3)
Title |
---|
WEI LI等: "A panoramic image generating algorithm based on hexahedral and spherical model", 《2011 2ND INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE, MANAGEMENT SCIENCE AND ELECTRONIC COMMERCE (AIMSEC)》 * |
刘祥峰: "基于鱼眼相机的全景环视系统关键技术研究", 《中国优秀硕士论文全文数据库信息科技辑》 * |
李艳丽等: "稳健的球面全景图全自动生成算法", 《计算机辅助设计与图形学学报》 * |
Cited By (33)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108122200A (en) * | 2017-12-20 | 2018-06-05 | 宁波视睿迪光电有限公司 | Image split-joint method and device |
CN108760767B (en) * | 2018-05-31 | 2020-07-17 | 电子科技大学 | Large-size liquid crystal display defect detection method based on machine vision |
CN108734681A (en) * | 2018-05-31 | 2018-11-02 | 天津煋鸟科技有限公司 | One kind being based on Embedded on piece radial distortion bearing calibration |
CN108760767A (en) * | 2018-05-31 | 2018-11-06 | 电子科技大学 | Large-size LCD Screen defect inspection method based on machine vision |
CN110580722A (en) * | 2018-06-08 | 2019-12-17 | 上海欧菲智能车联科技有限公司 | Calibration test method and calibration test device for panoramic system and vehicle |
CN109272445A (en) * | 2018-10-29 | 2019-01-25 | 中国航空无线电电子研究所 | Panoramic video joining method based on Sphere Measurement Model |
CN109272445B (en) * | 2018-10-29 | 2022-11-04 | 中国航空无线电电子研究所 | Panoramic video stitching method based on spherical model |
CN110189263A (en) * | 2019-05-05 | 2019-08-30 | 浙江大学 | It is a kind of based on multi-angle sampling big visual field wear display equipment distortion correction method |
CN110189263B (en) * | 2019-05-05 | 2021-05-07 | 浙江大学 | Large-view-field head-mounted display equipment distortion correction method based on multi-angle sampling |
CN110099220A (en) * | 2019-06-17 | 2019-08-06 | 广东中星微电子有限公司 | A kind of panorama mosaic method and device |
GB2591857A (en) * | 2019-08-23 | 2021-08-11 | Shang Hai Yiwo Information Tech Co Ltd | Photographing-based 3D modeling system and method, and automatic 3D modeling apparatus and method |
WO2021036353A1 (en) * | 2019-08-23 | 2021-03-04 | 上海亦我信息技术有限公司 | Photographing-based 3d modeling system and method, and automatic 3d modeling apparatus and method |
GB2591857B (en) * | 2019-08-23 | 2023-12-06 | Shang Hai Yiwo Information Tech Co Ltd | Photography-based 3D modeling system and method, and automatic 3D modeling apparatus and method |
CN110619611A (en) * | 2019-09-09 | 2019-12-27 | 北京迈格威科技有限公司 | Image correction calibration method and device, computer equipment and storage medium |
CN110691203A (en) * | 2019-10-21 | 2020-01-14 | 湖南泽天智航电子技术有限公司 | Multi-path panoramic video splicing display method and system based on texture mapping |
CN110691203B (en) * | 2019-10-21 | 2021-11-16 | 湖南泽天智航电子技术有限公司 | Multi-path panoramic video splicing display method and system based on texture mapping |
CN110942023A (en) * | 2019-11-25 | 2020-03-31 | 鹰驾科技(深圳)有限公司 | Indication method, device and equipment for vehicle vision blind area and storage medium |
CN111507894B (en) * | 2020-04-17 | 2023-06-13 | 浙江大华技术股份有限公司 | Image stitching processing method and device |
CN111507894A (en) * | 2020-04-17 | 2020-08-07 | 浙江大华技术股份有限公司 | Image splicing processing method and device |
CN113724141A (en) * | 2020-05-26 | 2021-11-30 | 杭州海康威视数字技术股份有限公司 | Image correction method and device and electronic equipment |
CN113724141B (en) * | 2020-05-26 | 2023-09-05 | 杭州海康威视数字技术股份有限公司 | Image correction method and device and electronic equipment |
CN112037128A (en) * | 2020-08-21 | 2020-12-04 | 苏州巨能图像检测技术有限公司 | Panoramic video splicing method |
CN112037128B (en) * | 2020-08-21 | 2023-11-03 | 苏州巨能图像检测技术有限公司 | Panoramic video stitching method |
CN114143528A (en) * | 2020-09-04 | 2022-03-04 | 北京大视景科技有限公司 | Multi-video stream fusion method, electronic device and storage medium |
CN112884689A (en) * | 2021-02-25 | 2021-06-01 | 景德镇陶瓷大学 | Highlight removing method for image on strong reflection surface |
CN112884689B (en) * | 2021-02-25 | 2023-11-17 | 景德镇陶瓷大学 | Method for removing high light of strong reflection surface image |
CN113139900A (en) * | 2021-04-01 | 2021-07-20 | 北京科技大学设计研究院有限公司 | Method for acquiring complete surface image of bar |
CN113139900B (en) * | 2021-04-01 | 2023-09-01 | 北京科技大学设计研究院有限公司 | Method for acquiring complete surface image of bar |
CN113191974A (en) * | 2021-04-29 | 2021-07-30 | 青岛科技大学 | Method and system for obtaining ship panoramic image based on machine vision |
WO2023272457A1 (en) * | 2021-06-28 | 2023-01-05 | 华为技术有限公司 | Apparatus and system for image splicing, and related method |
CN115500582A (en) * | 2022-08-17 | 2022-12-23 | 上海科技大学 | Foot three-dimensional contour acquisition system |
CN115500582B (en) * | 2022-08-17 | 2024-03-22 | 上海科技大学 | Foot three-dimensional contour acquisition system |
CN115471482A (en) * | 2022-09-20 | 2022-12-13 | 重庆理工大学 | Small-caliber container inner wall defect detection method based on computer vision |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107424118A (en) | Based on the spherical panorama mosaic method for improving Lens Distortion Correction | |
CN105118055B (en) | Camera position amendment scaling method and system | |
TWI555378B (en) | An image calibration, composing and depth rebuilding method of a panoramic fish-eye camera and a system thereof | |
CN110197466B (en) | Wide-angle fisheye image correction method | |
CN110211043B (en) | Registration method based on grid optimization for panoramic image stitching | |
CN111047510B (en) | Large-field-angle image real-time splicing method based on calibration | |
CN110782394A (en) | Panoramic video rapid splicing method and system | |
CN103973989B (en) | Obtain the method and system of high-dynamics image | |
JP5739584B2 (en) | 3D image synthesizing apparatus and method for visualizing vehicle periphery | |
CN103826033B (en) | Image processing method, image processing equipment, image pick up equipment and storage medium | |
CN108805801A (en) | A kind of panoramic picture bearing calibration and system | |
CN107925751A (en) | For multiple views noise reduction and the system and method for high dynamic range | |
CN107358577B (en) | Rapid splicing method of cubic panoramic image | |
CN101916455B (en) | Method and device for reconstructing three-dimensional model of high dynamic range texture | |
CN105979238A (en) | Method for controlling global imaging consistency of multiple cameras | |
CN111199528A (en) | Fisheye image distortion correction method | |
CN105488766B (en) | Fisheye image bearing calibration and device | |
KR20090078463A (en) | Distorted image correction apparatus and method | |
CN111461963B (en) | Fisheye image stitching method and device | |
CN114697623B (en) | Projection plane selection and projection image correction method, device, projector and medium | |
CN110322485A (en) | A kind of fast image registration method of isomery polyphaser imaging system | |
CN109166076B (en) | Multi-camera splicing brightness adjusting method and device and portable terminal | |
CN111009030A (en) | Multi-view high-resolution texture image and binocular three-dimensional point cloud mapping method | |
TW201839716A (en) | Panoramic Image Stitching Method and System Thereof | |
CN109889799B (en) | Monocular structure light depth perception method and device based on RGBIR camera |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
AD01 | Patent right deemed abandoned |
Effective date of abandoning: 20201127 |
|
AD01 | Patent right deemed abandoned |