CN107833179A - The quick joining method and system of a kind of infrared image - Google Patents
The quick joining method and system of a kind of infrared image Download PDFInfo
- Publication number
- CN107833179A CN107833179A CN201710792862.6A CN201710792862A CN107833179A CN 107833179 A CN107833179 A CN 107833179A CN 201710792862 A CN201710792862 A CN 201710792862A CN 107833179 A CN107833179 A CN 107833179A
- Authority
- CN
- China
- Prior art keywords
- point
- image
- algorithms
- carried out
- matching
- 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 24
- 238000001514 detection method Methods 0.000 claims abstract description 17
- 238000005259 measurement Methods 0.000 claims abstract description 16
- 238000013459 approach Methods 0.000 claims abstract description 8
- 238000005070 sampling Methods 0.000 claims description 11
- 238000012360 testing method Methods 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 6
- 230000000007 visual effect Effects 0.000 description 9
- 230000004438 eyesight Effects 0.000 description 6
- 230000007704 transition Effects 0.000 description 4
- 230000007812 deficiency Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 206010047571 Visual impairment Diseases 0.000 description 2
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 1
- 206010044565 Tremor Diseases 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000000686 essence Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000003331 infrared imaging Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000004297 night vision Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000013519 translation 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 transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4038—Image mosaicing, e.g. composing plane images from plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- 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
- G06T2207/10048—Infrared image
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses the quick joining method and system of a kind of infrared image.The inventive method includes step:N images of a scene are obtained, and are overlapped between every adjacent two images;Feature point detection and characteristic point are carried out to image using SIFT algorithms just to match, in Feature Points Matching, using manhatton distance as similarity measurement;Error hiding is removed, exterior point and estimation model stability are rejected using RANSAC algorithms, in RANSAC algorithms, matching double points with a high credibility stood out;Cylindrical surface projecting is carried out to image, each pixel of every pictures is projected into the same coordinate system;Image co-registration is carried out to the picture matched using with weigthed sums approach, finally obtains the panoramic picture of whole scene.Present system is corresponding with the above method.The present invention is improved to the SIFT algorithms and RANSAC algorithms of classics, in the case where not influenceing splicing effect, improves splicing speed, and it is higher to splice accuracy.
Description
Technical field
The invention belongs to infrared image processing technology field, more particularly to a kind of quick joining method of infrared image and it is
System.
Background technology
Image mosaic is the research field of an increased popularity, and it has become photograph cartography, computer vision, image
Focus in processing and computer graphics study.Image mosaic refers to the image by a series of space overlaps that align, and forms
One seamless, high-resolution image, make spliced image that there is higher resolution ratio and the bigger visual field than single image.
Based drive Panorama Mosaic model was proposed by Richard Szeliski in 1996, was used
Levenberg-Marquardt iterative nonlinears minimize method, and figure is carried out by obtaining the geometric transform relation between image
As registering, because the method effect is preferable, fast convergence rate, and can handle and be treated with a variety of conversion such as translation, rotation, affine
Stitching image, therefore also turn into the classic algorithm in image mosaic field, most stitching algorithm all thus algorithm improvements.
In terms of the night vision imaging of military field, night-viewing twilight or infrared imaging device all can be due to apparatuss for making a video recording
Limitation and wide-field picture can not be shot.Using image mosaic technology, in the feelings according to capture apparatus and surrounding scenes
After condition is analyzed, it is possible to the multiple image of 360 degree scenery around will be covered by rotating shooting equipment to shoot, then general
They are stitched together, so as to obtain the panoramic picture at the even 360 degree angles in super large visual angle in real time.This is played in infrared early warning
Very big effect.
In microminiature caterpillar mobile robot project, monocular vision can not meet robotic vision navigation needs, and
And the field range of monocular vision robot is significantly less than the visual field of binocular vision robot.Utilize image mosaic technology, splicing
The image of robot binocular collection, the visual field of robot can be increased, to robotic vision navigation provider just.
In terms of Medical Image Processing, the visual field of microscope or ultrasonic wave is smaller, and doctor can not be carried out by piece image
Examination, is also required to incomplete image mosaic be an entirety simultaneously for the DATA REASONING of big target image.So phase
It is the key link for realizing teledata measurement and remote medical consultation with specialists that adjacent each width image mosaic, which is got up,.
, can be to from the same area using the image registration techniques in image mosaic technology in remote sensing technology field
Two width or multiple image are compared, and can also lose genuine surface map by what remote sensing satellite photographed using image mosaic technology
As being spliced into more accurately complete image, as the foundation further studied.
The content of the invention
It is an object of the invention to provide the quick joining method and system of a kind of infrared image, it is intended to solves above-mentioned background
The deficiencies in the prior art in technology.
The present invention is achieved in that a kind of quick joining method of infrared image, and this method comprises the following steps:
S1, n images for obtaining a scene, and overlapped between every adjacent two images;
S2, the just matching of feature point detection and characteristic point is carried out to image using SIFT algorithms, will be graceful in Feature Points Matching
Hatton's distance is used as similarity measurement;
S3, error hiding is removed, exterior point and estimation model stability are rejected using RANSAC algorithms, in RANSAC algorithms
In, matching double points with a high credibility are stood out;
S4, cylindrical surface projecting is carried out to image, each pixel of every pictures is projected into the same coordinate system;
S5, using with weigthed sums approach image co-registration is carried out to the picture matched, finally obtain the panorama of whole scene
Image.
Preferably, in step s 2, the SIFT algorithms specifically include following steps:
DoG graphical rules space is generated, DoG extreme values is asked for and forms key point, key point is accurately positioned, is rejected
The relatively low unstable extreme point of contrast;
Generation includes key point position, pixel value, the sub- container of the description in direction;
Establish KD-tree;
It is manhatton distance by feature similarity measurement selection, sets initial distance threshold value, feature is carried out in KD-Tree
Point matching.
Preferably, in step s3, the removal error hiding includes:
Reject one-to-many point, many-to-one point, and wherein still existing partial error match point;
The match point of apparent error is rejected by Slope Constraint;
The distant matching double points of fraction mistake are deleted using epipolar-line constraint.
Preferably, in step s3, the RANSAC algorithms comprise the following steps:Sign point is divided to set s,
When producing new model parameter every time, tested first in the high data acquisition system of the quality for carrying out this sampling, it is assumed that be
{S1, S2, if interior ratio of the model, which has exceeded, pre-sets point ratio B in a set1With set Euclidean distance threshold value
T1, then the model be used for full data detection.
The present invention further discloses a kind of quick splicing system of infrared image, the system includes:
Image collection module, for obtaining n images of a scene, and there is part weight between every adjacent two images
It is folded;
Feature Points Matching module, just matched for carrying out feature point detection and characteristic point to image using SIFT algorithms,
During Feature Points Matching, using manhatton distance as similarity measurement;
Error hiding removes module, and for removing error hiding, exterior point and estimation model stability are rejected using RANSAC algorithms
Property, in RANSAC algorithms, matching double points with a high credibility are stood out;
Cylindrical surface projecting module, for carrying out cylindrical surface projecting to image, each pixel of every pictures is projected to same
Coordinate system;
Image co-registration module, for using image co-registration is carried out to the picture matched with weigthed sums approach, finally obtaining
The panoramic picture of whole scene.
Preferably, in Feature Points Matching module, the SIFT algorithms specifically include following steps:
DoG graphical rules space is generated, DoG extreme values is asked for and forms key point, key point is accurately positioned, is rejected
The relatively low unstable extreme point of contrast;
Generation includes key point position, pixel value, the sub- container of the description in direction;
Establish KD-tree;
It is manhatton distance by feature similarity measurement selection, sets initial distance threshold value, feature is carried out in KD-Tree
Point matching.
Preferably, in error hiding removes module, the removal error hiding includes:
Reject one-to-many point, many-to-one point, and wherein still existing partial error match point;
The match point of apparent error is rejected by Slope Constraint;
The distant matching double points of fraction mistake are deleted using epipolar-line constraint.
Preferably, in error hiding removes module, the RANSAC algorithms comprise the following steps:Sign point is entered to set s
Row division, when producing new model parameter every time, examined first in the high data acquisition system of the quality for carrying out this sampling
Test, it is assumed that be { S1, S2, if interior ratio of the model, which has exceeded, pre-sets point ratio B in a set1It is European with gathering
Distance threshold T1, then the model be used for full data detection.
Overcome the deficiencies in the prior art of the present invention, there is provided the quick joining method and system of a kind of infrared image.In this hair
In bright, n images of a scene are obtained, this n opens in images, overlaps per adjacent two, entered first with SIFT algorithms
Row feature point detection and characteristic point just match, in Feature Points Matching, using manhatton distance as similarity measurement, Ran Houyong
RANSAC algorithms remove error hiding, matching double points with a high credibility are come before, i.e., raw data set is ranked up, then from
The high data of quality start to extract, and reduce frequency in sampling, cylindrical surface projecting, image are carried out to the picture matched with weigthed sums approach
Fusion, finally obtain the panoramic picture of whole scene.
The present invention has considered the accuracy and speed of image mosaic, on the basis of traditional SIFT algorithms, using graceful Kazakhstan
Threshold value of pausing instead of distance threshold, on the basis of traditional RANSAC algorithms, by characteristic point to being arranged from high in the end by quality
Sequence, full data detection is avoided, so as on the premise of splicing effect is not influenceed, improve splicing speed.
The shortcomings that compared to prior art and deficiency, the invention has the advantages that:SIFT of the present invention to classics
Algorithm and RANSAC algorithms are improved, and in the case where not influenceing splicing effect, improve splicing speed, and splice accuracy
It is higher.
Brief description of the drawings
Fig. 1 is the step flow chart of the quick joining method of infrared image of the present invention;
Fig. 2 is two artwork to be spliced in the embodiment of the present invention;
Fig. 3 is that two artworks to be spliced pass through the spliced splicing effect figure of the present invention in Fig. 2;
Fig. 4 is the structural representation of the quick splicing system of infrared image of the present invention.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
As shown in figure 1, the invention discloses a kind of quick joining method of infrared image, this method comprises the following steps:
S1, n images for obtaining a scene, and overlapped between every adjacent two images;
S2, the just matching of feature point detection and characteristic point is carried out to image using SIFT algorithms, will be graceful in Feature Points Matching
Hatton's distance is used as similarity measurement;
S3, error hiding is removed, exterior point and estimation model stability are rejected using RANSAC algorithms, in RANSAC algorithms
In, matching double points with a high credibility are stood out;
S4, cylindrical surface projecting is carried out to image, each pixel of every pictures is projected into the same coordinate system;
S5, using with weigthed sums approach image co-registration is carried out to the picture matched, finally obtain the panorama of whole scene
Image.
As described in step S1, n images of a scene are obtained, this n is opened in images, there are 40% weights per adjacent two
It is folded.
As described in step S2, feature point detection and characteristic point is carried out with SIFT algorithms just to match, will in Feature Points Matching
Manhatton distance is as similarity measurement.In SIFT algorithms of the present invention, more specifically comprise the following steps:Firstly generate DoG figures
As metric space, ask for DoG extreme values and form key point, key point is accurately positioned, reject the relatively low shakiness of contrast
Fixed extreme point;Generation includes key point position, pixel value, the sub- container of the description in direction.Establish KD-tree;Feature is similar
Property metric sebection be manhatton distance, set initial distance threshold value, Feature Points Matching is carried out in KD-Tree.
As described in step S3, after SIFT feature slightly matches, substantial amounts of matching double points have been obtained.Eliminate it is one-to-many and
Partial error match point wherein still be present after many-to-one point, the match point of apparent error can be picked after Slope Constraint
Remove, the distant matching double points of fraction mistake can be deleted using epipolar-line constraint.So far, matching double points have aligned
Really, part exterior point but still be present, need further to carry out rejecting exterior point by RANSAC algorithms, and the algorithm can be passed through
Some model is carried out to stablize estimation.
The RANSAC algorithms are first random to choose minimum sampling collection among whole data set, and passes through these sampling collection meters
The initial value of correlation model parameters, then the model by calculating are calculated to find other interior points in data set, and will be outer
Point rejects and farthest eliminates influence of the exterior point to overall estimation with this.In embodiments of the present invention, the RANSAC algorithms
More specifically comprise the following steps:Division of the point to set s will be levied, can entered first when producing new model parameter every time
Tested in the high data acquisition system of the capable quality that this is sampled, it is assumed that be { S1, S2, if interior ratio of this model surpasses
Cross and pre-set point ratio B in a set1(it is interior points with sampling set in data count ratios) and gather it is European away from
From threshold value T1, then this model can participate in full data detection;Otherwise this model is abandoned, is sampled into next time.In original number
In the case of huge, model data Check-Out Time can be greatly reduced.
As described in step S4, each pixel of every pictures is projected in a coordinate system.Image sequence is entity
Two-dimensional projection of the scenery under different coordinates, visual consistency can not be met by directly carrying out splicing to shooting image, so needing
Image to be spliced is projected to respectively under the coordinate system of a standard, then carry out the splicing of image again, can so obtained
Obtain preferable visual effect.
As described in step S5, image co-registration refers to after images match is completed, and image is sutured, and to the side of suture
Boundary is smoothed, and is allowed and is sutured nature transition.Because any two width adjacent image is impossible to accomplish on acquisition condition
It is exactly the same, therefore, for some should identical picture characteristics, the light characteristics of such as image, in two images just not
What can be showed is just the same.Image mosaic gap is exactly the image district that another piece image is transitioned into from the image-region of piece image
During domain, because some correlation properties in image are there occurs transition and caused.Image co-registration is exactly to allow spelling between image
Seam gap unobvious, splicing are more natural.
Identical weights are directly taken to the pixel value of original image, are then weighted the pixel for averagely obtaining fused images
Value, two images A, B will such as be merged by illustrating, and the pixel value of their fused image is exactly A*50%+B*50%.
In embodiments of the present invention, with two artworks to be spliced shown in Fig. 2, by above method splicing of the present invention
Afterwards, spliced map as shown in Figure 3 is obtained.As seen from the figure, image split-joint method of the invention, accuracy are very high.
As shown in figure 4, the present invention further discloses a kind of quick splicing system of infrared image, the system includes:
Image collection module 1, for obtaining n images of a scene, and there is part weight between every adjacent two images
It is folded;
Feature Points Matching module 2, just matched for carrying out feature point detection and characteristic point to image using SIFT algorithms,
During Feature Points Matching, using manhatton distance as similarity measurement;
Error hiding removes module 3, and for removing error hiding, exterior point and estimation model stability are rejected using RANSAC algorithms
Property, in RANSAC algorithms, matching double points with a high credibility are stood out;
Cylindrical surface projecting module 4, for carrying out cylindrical surface projecting to image, each pixel of every pictures is projected to same
Coordinate system;
Image co-registration module 5, for using image co-registration is carried out to the picture matched with weigthed sums approach, finally obtaining
The panoramic picture of whole scene.
As described in image collection module 1, n images of a scene are obtained, this n opens in images, had per adjacent two
40% it is overlapping.
As described in Feature Points Matching module 2, carry out feature point detection and characteristic point with SIFT algorithms and just match, in characteristic point
During matching, using manhatton distance as similarity measurement.In SIFT algorithms of the present invention, more specifically comprise the following steps:First
Generate DoG graphical rules spaces, ask for DoG extreme values and form key point, key point is accurately positioned, reject contrast compared with
Low unstable extreme point;Generation includes key point position, pixel value, the sub- container of the description in direction.Establish KD-tree;Will
Feature similarity measurement selection is manhatton distance, sets initial distance threshold value, Feature Points Matching is carried out in KD-Tree.
As described in error hiding removes module 3, after SIFT feature slightly matches, substantial amounts of matching double points have been obtained.Reject
Partial error match point wherein still be present after one-to-many and many-to-one point, after Slope Constraint can be by apparent error
Match point reject, can delete the distant matching double points of fraction mistake using epipolar-line constraint.So far, matching double points have been
Through relatively accurate, but part exterior point still be present, need further to carry out rejecting exterior point by RANSAC algorithms, and can lead to
The algorithm is crossed some model is carried out to stablize estimation.
The RANSAC algorithms are first random to choose minimum sampling collection among whole data set, and passes through these sampling collection meters
The initial value of correlation model parameters, then the model by calculating are calculated to find other interior points in data set, and will be outer
Point rejects and farthest eliminates influence of the exterior point to overall estimation with this.In embodiments of the present invention, the RANSAC algorithms
More specifically comprise the following steps:Division of the point to set s will be levied, can entered first when producing new model parameter every time
Tested in the high data acquisition system of the capable quality that this is sampled, it is assumed that be { S1, S2, if interior ratio of this model surpasses
Cross and pre-set point ratio B in a set1(it is interior points with sampling set in data count ratios) and gather it is European away from
From threshold value T1, then this model can participate in full data detection;Otherwise this model is abandoned, is sampled into next time.In original number
In the case of huge, model data Check-Out Time can be greatly reduced.
As described in cylindrical surface projecting module 4, each pixel of every pictures is projected in a coordinate system.Image sequence
It is two-dimensional projection of the entity scenery under different coordinates, visual consistency can not be met by directly carrying out splicing to shooting image,
So needing to project to image to be spliced respectively under the coordinate system of one standard, the splicing of image is then carried out again, so
Preferable visual effect can be obtained.
As described in image co-registration module 5, image co-registration refers to after images match is completed, and image is sutured, and right
The border of suture is smoothed, and is allowed and is sutured nature transition.Because any two width adjacent image all can not on acquisition condition
Can accomplish it is identical, therefore, for some should identical picture characteristics, the light characteristics of such as image, in two width figures
It would not be showed as in just the same.Image mosaic gap is exactly to be transitioned into another piece image from the image-region of piece image
Image-region when, because some correlation properties in image are there occurs transition and caused.Image co-registration is exactly to allow figure
Splicing gap unobvious as between, splicing are more natural.
Identical weights are directly taken to the pixel value of original image, are then weighted the pixel for averagely obtaining fused images
Value, two images A, B will such as be merged by illustrating, and the pixel value of their fused image is exactly A*50%+B*50%.
In embodiments of the present invention, with two artworks to be spliced shown in Fig. 2, by said system splicing of the present invention
Afterwards, spliced map as shown in Figure 3 is obtained.As seen from the figure, image split-joint method of the invention, accuracy are very high.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
All any modification, equivalent and improvement made within refreshing and principle etc., should be included in the scope of the protection.
Claims (8)
1. the quick joining method of a kind of infrared image, it is characterised in that this method comprises the following steps:
S1, n images for obtaining a scene, and overlapped between every adjacent two images;
S2, the just matching of feature point detection and characteristic point is carried out to image using SIFT algorithms, in Feature Points Matching, by Manhattan
Distance is used as similarity measurement;
S3, error hiding is removed, exterior point and estimation model stability are rejected using RANSAC algorithms, will in RANSAC algorithms
Matching double points with a high credibility are stood out;
S4, cylindrical surface projecting is carried out to image, each pixel of every pictures is projected into the same coordinate system;
S5, using with weigthed sums approach image co-registration is carried out to the picture matched, finally obtain the panoramic picture of whole scene.
2. the quick joining method of infrared image as claimed in claim 1, it is characterised in that in step s 2, the SIFT
Algorithm specifically includes following steps:
DoG graphical rules space is generated, DoG extreme values is asked for and forms key point, key point is accurately positioned, rejects contrast
Spend relatively low unstable extreme point;
Generation includes key point position, pixel value, the sub- container of the description in direction;
Establish KD-tree;
It is manhatton distance by feature similarity measurement selection, sets initial distance threshold value, characteristic point is carried out in KD-Tree
Match somebody with somebody.
3. the quick joining method of infrared image as claimed in claim 1, it is characterised in that in step s3, the removal
Error hiding includes:
Reject one-to-many point, many-to-one point, and wherein still existing partial error match point;
The match point of apparent error is rejected by Slope Constraint;
The distant matching double points of fraction mistake are deleted using epipolar-line constraint.
4. the quick joining method of infrared image as claimed in claim 1, it is characterised in that in step s3, described
RANSAC algorithms comprise the following steps:Sign point is divided to set s, when producing new model parameter every time, existed first
Test in the high data acquisition system of quality of this sampling, it is assumed that be { S1, S2, if interior ratio of the model exceedes
Pre-set point ratio B in a set1With set Euclidean distance threshold value T1, then the model be used for full data detection.
5. the quick splicing system of a kind of infrared image, it is characterised in that the system includes:
Image collection module, for obtaining n images of a scene, and overlapped between every adjacent two images;
Feature Points Matching module, just matched for carrying out feature point detection and characteristic point to image using SIFT algorithms, in feature
During Point matching, using manhatton distance as similarity measurement;
Error hiding removes module, and for removing error hiding, exterior point and estimation model stability are rejected using RANSAC algorithms,
In RANSAC algorithms, matching double points with a high credibility are stood out;
Cylindrical surface projecting module, for carrying out cylindrical surface projecting to image, each pixel of every pictures is projected into same coordinate
System;
Image co-registration module, for using image co-registration is carried out to the picture matched with weigthed sums approach, finally obtaining whole
The panoramic picture of scene.
6. the quick splicing system of infrared image as claimed in claim 5, it is characterised in that in Feature Points Matching module,
The SIFT algorithms specifically include following steps:
DoG graphical rules space is generated, DoG extreme values is asked for and forms key point, key point is accurately positioned, rejects contrast
Spend relatively low unstable extreme point;
Generation includes key point position, pixel value, the sub- container of the description in direction;
Establish KD-tree;
It is manhatton distance by feature similarity measurement selection, sets initial distance threshold value, characteristic point is carried out in KD-Tree
Match somebody with somebody.
7. the quick splicing system of infrared image as claimed in claim 6, it is characterised in that in error hiding removes module,
The removal error hiding includes:
Reject one-to-many point, many-to-one point, and wherein still existing partial error match point;
The match point of apparent error is rejected by Slope Constraint;
The distant matching double points of fraction mistake are deleted using epipolar-line constraint.
8. the quick splicing system of infrared image as claimed in claim 7, it is characterised in that in error hiding removes module,
The RANSAC algorithms comprise the following steps:Sign point is divided to set s, it is first when producing new model parameter every time
First tested in the high data acquisition system of the quality for carrying out this sampling, it is assumed that be { S1, S2, if interior ratio of the model
Exceed and pre-set point ratio B in a set1With set Euclidean distance threshold value T1, then the model be used for full data detection.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710792862.6A CN107833179A (en) | 2017-09-05 | 2017-09-05 | The quick joining method and system of a kind of infrared image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710792862.6A CN107833179A (en) | 2017-09-05 | 2017-09-05 | The quick joining method and system of a kind of infrared image |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107833179A true CN107833179A (en) | 2018-03-23 |
Family
ID=61643222
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710792862.6A Pending CN107833179A (en) | 2017-09-05 | 2017-09-05 | The quick joining method and system of a kind of infrared image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107833179A (en) |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108537833A (en) * | 2018-04-18 | 2018-09-14 | 昆明物理研究所 | A kind of quick joining method of infrared image |
CN108648146A (en) * | 2018-05-08 | 2018-10-12 | 南京齿贝犀科技有限公司 | Tooth tongue palate side Panorama Mosaic method based on Local Optimization Algorithm |
CN108805799A (en) * | 2018-04-20 | 2018-11-13 | 平安科技(深圳)有限公司 | Panoramic picture synthesizer, method and computer readable storage medium |
CN108961322A (en) * | 2018-05-18 | 2018-12-07 | 辽宁工程技术大学 | A kind of error hiding elimination method suitable for the sequential images that land |
CN109493278A (en) * | 2018-10-24 | 2019-03-19 | 北京工业大学 | A kind of large scene image mosaic system based on SIFT feature |
CN109712070A (en) * | 2018-12-04 | 2019-05-03 | 天津津航技术物理研究所 | A kind of infrared panoramic image split-joint method based on graph cut |
CN110084754A (en) * | 2019-06-25 | 2019-08-02 | 江苏德劭信息科技有限公司 | A kind of image superimposing method based on improvement SIFT feature point matching algorithm |
CN110443838A (en) * | 2019-07-09 | 2019-11-12 | 中山大学 | A kind of associated diagram building method for stereo-picture splicing |
CN110689485A (en) * | 2019-10-14 | 2020-01-14 | 中国空气动力研究与发展中心超高速空气动力研究所 | SIFT image splicing method applied to infrared nondestructive testing of large pressure container |
CN110807732A (en) * | 2019-10-11 | 2020-02-18 | 武汉兰丁医学高科技有限公司 | Panoramic stitching system and method for microscopic images |
CN111179176A (en) * | 2019-12-30 | 2020-05-19 | 北京东宇宏达科技有限公司 | Automatic splicing calibration method for infrared panoramic imaging |
CN111680703A (en) * | 2020-06-01 | 2020-09-18 | 中国电建集团昆明勘测设计研究院有限公司 | 360-degree construction panorama linkage positioning method based on image feature point detection and matching |
CN112162107A (en) * | 2020-10-10 | 2021-01-01 | 广西信路威科技发展有限公司 | Vehicle running speed measuring method and system based on spliced image |
CN112986329A (en) * | 2021-02-07 | 2021-06-18 | 电子科技大学 | Infrared thermal imaging detection method for ultra-high speed impact damage of large-size non-planar test piece |
CN113807451A (en) * | 2021-05-25 | 2021-12-17 | 中德(珠海)人工智能研究院有限公司 | Panoramic image feature point matching model training method and device and server |
CN108510446B (en) * | 2018-04-10 | 2023-03-14 | 四川和生视界医药技术开发有限公司 | Method and device for superimposing retinal images |
CN116797463A (en) * | 2023-08-22 | 2023-09-22 | 佗道医疗科技有限公司 | Feature point pair extraction method and image stitching method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103914819A (en) * | 2014-03-26 | 2014-07-09 | 东华大学 | Infrared image splicing method based on improved RANSAC |
US20140314325A1 (en) * | 2011-11-30 | 2014-10-23 | Nokia Corporation | Method and apparatus for image stitching |
CN104596519A (en) * | 2015-02-17 | 2015-05-06 | 哈尔滨工业大学 | RANSAC algorithm-based visual localization method |
CN106649690A (en) * | 2016-12-16 | 2017-05-10 | 西安电子科技大学 | Security image retrieval method and system and image retrieval server |
-
2017
- 2017-09-05 CN CN201710792862.6A patent/CN107833179A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140314325A1 (en) * | 2011-11-30 | 2014-10-23 | Nokia Corporation | Method and apparatus for image stitching |
CN103914819A (en) * | 2014-03-26 | 2014-07-09 | 东华大学 | Infrared image splicing method based on improved RANSAC |
CN104596519A (en) * | 2015-02-17 | 2015-05-06 | 哈尔滨工业大学 | RANSAC algorithm-based visual localization method |
CN106649690A (en) * | 2016-12-16 | 2017-05-10 | 西安电子科技大学 | Security image retrieval method and system and image retrieval server |
Non-Patent Citations (3)
Title |
---|
李爱霞 等: ""基于空间约束的正则化流形学习影像匹配方法"", 《计算机工程与应用》 * |
赵小强 等: ""一种面向图像拼接的快速匹配算法"", 《南京理工大学学报》 * |
黄梅: ""基于改进 RANSAC 算法的图像拼接技术"", 《海南大学学报自然科学版》 * |
Cited By (24)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108510446B (en) * | 2018-04-10 | 2023-03-14 | 四川和生视界医药技术开发有限公司 | Method and device for superimposing retinal images |
CN108537833A (en) * | 2018-04-18 | 2018-09-14 | 昆明物理研究所 | A kind of quick joining method of infrared image |
CN108805799A (en) * | 2018-04-20 | 2018-11-13 | 平安科技(深圳)有限公司 | Panoramic picture synthesizer, method and computer readable storage medium |
CN108805799B (en) * | 2018-04-20 | 2021-04-23 | 平安科技(深圳)有限公司 | Panoramic image synthesis apparatus, panoramic image synthesis method, and computer-readable storage medium |
CN108648146A (en) * | 2018-05-08 | 2018-10-12 | 南京齿贝犀科技有限公司 | Tooth tongue palate side Panorama Mosaic method based on Local Optimization Algorithm |
CN108961322A (en) * | 2018-05-18 | 2018-12-07 | 辽宁工程技术大学 | A kind of error hiding elimination method suitable for the sequential images that land |
CN108961322B (en) * | 2018-05-18 | 2021-08-10 | 辽宁工程技术大学 | Mismatching elimination method suitable for landing sequence images |
CN109493278A (en) * | 2018-10-24 | 2019-03-19 | 北京工业大学 | A kind of large scene image mosaic system based on SIFT feature |
CN109712070A (en) * | 2018-12-04 | 2019-05-03 | 天津津航技术物理研究所 | A kind of infrared panoramic image split-joint method based on graph cut |
CN110084754A (en) * | 2019-06-25 | 2019-08-02 | 江苏德劭信息科技有限公司 | A kind of image superimposing method based on improvement SIFT feature point matching algorithm |
CN110443838B (en) * | 2019-07-09 | 2023-05-05 | 中山大学 | Associative graph construction method for stereoscopic image stitching |
CN110443838A (en) * | 2019-07-09 | 2019-11-12 | 中山大学 | A kind of associated diagram building method for stereo-picture splicing |
CN110807732B (en) * | 2019-10-11 | 2023-08-29 | 武汉兰丁智能医学股份有限公司 | Panoramic stitching system and method for microscopic images |
CN110807732A (en) * | 2019-10-11 | 2020-02-18 | 武汉兰丁医学高科技有限公司 | Panoramic stitching system and method for microscopic images |
CN110689485A (en) * | 2019-10-14 | 2020-01-14 | 中国空气动力研究与发展中心超高速空气动力研究所 | SIFT image splicing method applied to infrared nondestructive testing of large pressure container |
CN111179176A (en) * | 2019-12-30 | 2020-05-19 | 北京东宇宏达科技有限公司 | Automatic splicing calibration method for infrared panoramic imaging |
CN111680703B (en) * | 2020-06-01 | 2022-06-03 | 中国电建集团昆明勘测设计研究院有限公司 | 360-degree construction panorama linkage positioning method based on image feature point detection and matching |
CN111680703A (en) * | 2020-06-01 | 2020-09-18 | 中国电建集团昆明勘测设计研究院有限公司 | 360-degree construction panorama linkage positioning method based on image feature point detection and matching |
CN112162107A (en) * | 2020-10-10 | 2021-01-01 | 广西信路威科技发展有限公司 | Vehicle running speed measuring method and system based on spliced image |
CN112986329B (en) * | 2021-02-07 | 2022-03-25 | 电子科技大学 | Infrared thermal imaging detection method for ultra-high speed impact damage of large-size non-planar test piece |
CN112986329A (en) * | 2021-02-07 | 2021-06-18 | 电子科技大学 | Infrared thermal imaging detection method for ultra-high speed impact damage of large-size non-planar test piece |
CN113807451A (en) * | 2021-05-25 | 2021-12-17 | 中德(珠海)人工智能研究院有限公司 | Panoramic image feature point matching model training method and device and server |
CN116797463A (en) * | 2023-08-22 | 2023-09-22 | 佗道医疗科技有限公司 | Feature point pair extraction method and image stitching method |
CN116797463B (en) * | 2023-08-22 | 2023-11-21 | 佗道医疗科技有限公司 | Feature point pair extraction method and image stitching method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107833179A (en) | The quick joining method and system of a kind of infrared image | |
CN105245841B (en) | A kind of panoramic video monitoring system based on CUDA | |
Wang et al. | 360sd-net: 360 stereo depth estimation with learnable cost volume | |
CN103337094B (en) | A kind of method of applying binocular camera and realizing motion three-dimensional reconstruction | |
CN105957007B (en) | Image split-joint method based on characteristic point plane similarity | |
CN103150715B (en) | Image mosaic processing method and processing device | |
CN109064404A (en) | It is a kind of based on polyphaser calibration panorama mosaic method, panoramic mosaic system | |
Bae et al. | Computational rephotography. | |
CN104599258B (en) | A kind of image split-joint method based on anisotropic character descriptor | |
CN104699842B (en) | Picture display method and device | |
EP2328125B1 (en) | Image splicing method and device | |
CN106683045A (en) | Binocular camera-based panoramic image splicing method | |
CN107424181A (en) | A kind of improved image mosaic key frame rapid extracting method | |
CN106709868A (en) | Image stitching method and apparatus | |
CN110351494A (en) | A kind of panoramic video synthetic method, device and electronic equipment | |
CN107833181A (en) | A kind of three-dimensional panoramic image generation method and system based on zoom stereoscopic vision | |
CN107154014A (en) | A kind of real-time color and depth Panorama Mosaic method | |
CN107016646A (en) | One kind approaches projective transformation image split-joint method based on improved | |
CN103927785B (en) | A kind of characteristic point matching method towards up short stereoscopic image data | |
CN113221665A (en) | Video fusion algorithm based on dynamic optimal suture line and improved gradual-in and gradual-out method | |
CN114119739A (en) | Binocular vision-based hand key point space coordinate acquisition method | |
CN106534670A (en) | Panoramic video generating method based on fixedly connected fisheye lens camera unit | |
CN107146195A (en) | Sphere image split-joint method and device | |
CN106791800B (en) | The quality diagnosis method and device of panoramic video | |
CN108537831A (en) | The method and device of CT imagings is carried out to increasing material manufacturing workpiece |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180323 |