CN109872371A - A kind of monocular vision three-dimensional rebuilding method based on improvement Sift algorithm - Google Patents
A kind of monocular vision three-dimensional rebuilding method based on improvement Sift algorithm Download PDFInfo
- Publication number
- CN109872371A CN109872371A CN201910069847.8A CN201910069847A CN109872371A CN 109872371 A CN109872371 A CN 109872371A CN 201910069847 A CN201910069847 A CN 201910069847A CN 109872371 A CN109872371 A CN 109872371A
- Authority
- CN
- China
- Prior art keywords
- image
- point
- dimensional
- matching
- algorithm
- 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
Abstract
The invention discloses a kind of based on the monocular vision three-dimensional rebuilding method for improving Sift algorithm comprising the steps of: A, camera calibration；B, image characteristics extraction and matching；C, feature extraction and SIFT algorithm improvement；D, three-dimensional reconstruction；The present invention improves feature extracting and matching algorithm on the basis of conventional three-dimensional reconstruction technique, matching precision is improved, to obtain better three-dimensional reconstruction effect.
Description
Technical field
It is specifically a kind of based on the monocular view for improving Sift algorithm the present invention relates to a kind of monocular vision three-dimensional rebuilding method
Feel three-dimensional rebuilding method.
Background technique
The object of objective world is three-dimensional, and we are two-dimensional with the image that video camera obtains, but we can be with
The three-dimensional information of target is perceived by two dimensional image.Three-dimensional reconstruction is that processing image is calculated in turn in some way
The three-dimensional information that machine can identify, thus analyzes target.The model that real scene is obtained from image is computer view
The important research content of feel, by the threedimensional model for reconstructing target, so that it may obtain digitized three-dimensional data, these all have
There are higher application value and extensive market prospects.
One complete three-dimensional reconstruction system normally comprises three characteristic matching, camera calibration, three-dimensional reconstruction parts.
Characteristic matching is the corresponding relationship established between feature, by image pair of the same feature (point or line) in space in different views
It should get up.The corresponding relationship between point on point in space and two dimensional image be determined by the imaging model of video camera, and
The process that forming model is determined by calculating is exactly camera calibration.The purpose of three-dimensional reconstruction is complete in order to restore scenery
Structural information has important application value, can be difficult to solve in some plane visuals to play weight with insurmountable occasion
The effect wanted.Such as several important events below:
(1) industrial automation: product testing positioning, detection 2D image are very difficult to the defect of identification, automatic production line life
It produces and a series of industry such as the detection in assembly line and positioning positions occasions.
(2) medical assistance application: a series of pairs of trouble such as surgical navigational, beauty and shaping, medical surgery simulation, 3D medical treatment printing
The beneficial occasion of person's rehabilitation.
(3) virtual reality applications: virtual product model, virtual battlefield environment, somatic sensation television game, virtual sports emulation, enhancing
Reality etc..
(4) vision guided navigation: Mobile Robotics Navigation, intelligent automobile navigation, aircraft navigation, moon craft navigate, certainly
The dynamic workplace for determining target and measuring distance, and overcome the influence of light.
Visual sensor has small in size, and price is low, is easy to use, and it is excellent to be capable of providing environmental information abundant etc.
Point, therefore how to realize that scene three-dimensional reconstruction is popular research direction in recent years by visual sensor.But vision passes
Sensor can be divided into many classifications again, there is monocular, binocular, depth camera etc..In recent years, it is gradually obtained based on stereovision technique
Extensive utilization.It restores the depth that two dimensional image is lost by the method that the video camera of two fixed positions simulates human vision system
Spend information.In contrast, the method for realizing 3 D scene rebuilding merely with a handheld digital camera is studied, there is equipment letter
It is single, it is applicable in the advantages that convenient and low in cost, and be very easy to be widely applied in other field.By monocular vision
Carrying out three-dimensional reconstruction to specific objective becomes a new hot spot, and is becoming a new demand.It is desirable in hand
Machine photographs picture, just can synthesize threedimensional model, and the technology of the 3-dimensional reconstruction of monocular vision is research heat recently at present
Point.But due to the accuracy constraint by shooting angle, extraction feature, three-dimensional reconstruction result is unsatisfactory, can not adequately show
The structure of scene out.How to obtain satisfied three-dimensional reconstruction effect becomes a significantly project, not only has multi-party
Face practical application scene, and there is important theoretical research value.
Summary of the invention
The purpose of the present invention is to provide a kind of based on the monocular vision three-dimensional rebuilding method for improving Sift algorithm, to solve
The problems mentioned above in the background art.
To achieve the above object, the invention provides the following technical scheme:
A kind of monocular vision three-dimensional rebuilding method based on improvement Sift algorithm comprising the steps of:
A, camera calibration；
B, image characteristics extraction and matching；
C, feature extraction and SIFT algorithm improvement；
D, three-dimensional reconstruction.
As further technical solution of the present invention: the step A is specifically: scaling board is chosen, from multiple angle shots
The image of scaling board is obtained, detects the characteristic point in image respectively, seeks video camera further according to selected calibrated and calculated method
Intrinsic parameter and external parameters of cameras complete camera calibration.
As further technical solution of the present invention: the step B is specifically: choosing suitable characteristics of image, research figure
As feature extracting method and new image representation method, the target signature point set of image is obtained, then carry out characteristic matching.It carries out
Software design writes program using Halcon image processing software, completes the realization of image characteristics extraction and matching technique, and
It chooses many algorithms and compares test.
As further technical solution of the present invention: the SIFT algorithm improvement specifically includes the following steps: S1, using list
Mesh camera acquires image from multi-angle, detects extreme point to wherein any two width target image S and image T application SIFT operator,
And obtained characteristic point is screened, SIFT feature is obtained to collection Rs and Rt, S2, calculating phase equalization energy information matrix；S3,
Value of the given threshold between 0.01 and 0.05 is compared with the phase equalization energy information matrix being calculated, and is rejected
Maximum square, to collection, obtains characteristic point to be matched to collection Rs ' and Rt ' from the SIFT feature lower than threshold value；S4, three-dimensional is carried out
Match, calculates Euclidean distance closest approach and compared with time ratio of near point and the threshold value of setting, judge the phase of SIFT feature pair
Like degree, if ratio is less than the threshold value of setting, otherwise feature point pair matching success fails；S5, agreeing for each match point is calculated
Dare coefficient, and calculate it and constrain threshold value, the match point that Kendall's coefficient is less than threshold value is rejected, to obtain final match point
To collection.
As further technical solution of the present invention: the step C also needs to complete comparative test, verifies innovatory algorithm
Feasibility.
As further technical solution of the present invention: the step D is specifically: projection matrix is utilized, in conjunction with being previously obtained
Matching double points collection calculate the three-dimensional space point coordinate of object, then to determining that the three-dimensional point cloud of coordinate value carries out triangle and cut open
Point, the surface information of reconstruction attractor object, will be selected in image that video camera is shot wherein one it is most suitable, will be on the figure
The texture of scene, is mapped in threedimensional model, recovers three-dimensional scene information.
As further technical solution of the present invention: the step D needs to carry out software design, at Halcon image
Software is managed, three-dimensional reconstruction is completed and realizes, and the three-dimensional reconstruction result by being extracted matching algorithm based on different characteristic is compared,
Further verify the feasibility of innovatory algorithm.
Compared with prior art, the beneficial effects of the present invention are: the present invention changes on the basis of conventional three-dimensional reconstruction technique
Into feature extracting and matching algorithm, matching precision is improved, to obtain better three-dimensional reconstruction effect.
Detailed description of the invention
Fig. 1 is rectangular scaling board schematic diagram；
Fig. 2 is gridiron pattern scaling board schematic diagram；
Fig. 3 is Local Extremum detection schematic diagram；
Fig. 4 is characterized extraction and matching algorithm schematic diagram；
Fig. 5 is innovatory algorithm schematic diagram.
Fig. 6 is systems solutions design diagram.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Embodiment 1: please referring to Fig. 1-6, a kind of based on the monocular vision three-dimensional rebuilding method for improving Sift algorithm, specific mistake
Journey is as follows: step A, camera calibration.Several scaling methods of video camera are studied, calibrated and calculated method the most suitable is selected.
High-precision scaling board is chosen, obtains the image of scaling board from multiple angle shots.The characteristic point in image is detected respectively, then
Camera intrinsic parameter and external parameters of cameras are sought according to selected calibrated and calculated method, completes camera calibration；
Step B, image characteristics extraction and matching.Choose suitable characteristics of image, study image characteristic extracting method and
New image representation method obtains the target signature point set of image, then carries out characteristic matching.Carry out software design, application
Halcon image processing software, writes program, completes the realization of image characteristics extraction and matching technique, and choose many algorithms into
Row comparative test.
Step C, feature extraction and SIFT algorithm improvement.Be directed to target image characteristics extract with it is matched during, can
The problem that error hiding is excessive, matching effect is undesirable can be will appear, matching algorithm is extracted to SIFT feature and is improved, raising
With efficiency, reduce error hiding rate.Using the iterative algorithm of random sampling consistency, the unstable characteristic point of screening is simultaneously picked
It removes, reduces the characteristic point number of mistake, improve recognition efficiency.The essence of characteristic matching is improved while not influencing matching speed
Degree.Program is write, the experiment of the feature extracting and matching of application enhancements algorithm is carried out, and analyzes experimental result.It completes to having a competition
It tests, verifies the feasibility of innovatory algorithm.
Step D, three-dimensional reconstruction: projection matrix is utilized, the three-dimensional of object is calculated in conjunction with the matching double points collection being previously obtained
Spatial point coordinate.Then to the three-dimensional point cloud progress triangulation for determining coordinate value, the surface information of reconstruction attractor object.It will take the photograph
Selected in the image that camera is shot wherein one it is most suitable, the texture of scene on the figure is mapped in threedimensional model, it is extensive
It appears again three-dimensional scene information.Software design is carried out, using Halcon image processing software, three-dimensional reconstruction is completed and realizes.And
By extracting the three-dimensional reconstruction result comparison of matching algorithm based on different characteristic, the feasibility of innovatory algorithm is further verified.
It is as follows to improve SIFT algorithm steps on the basis of embodiment 1 for embodiment 2:
Step 1: image being acquired from multi-angle using monocular camera, to wherein any two width target image and image application
SIFT operator detects extreme point, and screens obtained characteristic point, obtain SIFT feature to collection and；
Step 2: calculating phase equalization energy information matrix；
Step 3: value of the given threshold between 0.01 and 0.05, with the phase equalization energy information matrix being calculated
Be compared, reject the characteristic point pair that maximum square is lower than threshold value, obtain characteristic point to be matched to collection and；
Step 4: carrying out Stereo matching.It calculates Euclidean distance closest approach and time ratio of near point and the threshold value of setting carries out pair
Than judging the similarity degree of SIFT feature pair.If ratio is less than the threshold value of setting, otherwise feature point pair matching success is lost
It loses；
Step 5: calculating the Kendall's coefficient of each match point, and calculate it and constrain threshold value, reject Kendall's coefficient and be less than
The match point of threshold value, to obtain final matching double points collection.
It is obvious to a person skilled in the art that invention is not limited to the details of the above exemplary embodiments, Er Qie
In the case where without departing substantially from spirit or essential attributes of the invention, the present invention can be realized in other specific forms.Therefore, no matter
From the point of view of which point, the present embodiments are to be considered as illustrative and not restrictive, and the scope of the present invention is by appended power
Benefit requires rather than above description limits, it is intended that all by what is fallen within the meaning and scope of the equivalent elements of the claims
Variation is included within the present invention.Any reference signs in the claims should not be construed as limiting the involved claims.
In addition, it should be understood that although this specification is described in terms of embodiments, but not each embodiment is only wrapped
Containing an independent technical solution, this description of the specification is merely for the sake of clarity, and those skilled in the art should
It considers the specification as a whole, the technical solutions in the various embodiments may also be suitably combined, forms those skilled in the art
The other embodiments being understood that.
Claims (7)
1. a kind of based on the monocular vision three-dimensional rebuilding method for improving Sift algorithm, which is characterized in that comprise the steps of:
A, camera calibration；
B, image characteristics extraction and matching；
C, feature extraction and SIFT algorithm improvement；
D, three-dimensional reconstruction.
2. according to claim 1 a kind of based on the monocular vision three-dimensional rebuilding method for improving Sift algorithm, feature exists
In the step A is specifically: choosing scaling board, obtain the image of scaling board from multiple angle shots, detect in image respectively
Characteristic point, seek camera intrinsic parameter and external parameters of cameras further according to selected calibrated and calculated method, complete video camera mark
It is fixed.
3. according to claim 1 a kind of based on the monocular vision three-dimensional rebuilding method for improving Sift algorithm, feature exists
In the step B is specifically: choosing suitable characteristics of image, study image characteristic extracting method and characteristics of image description side
Method obtains the target signature point set of image, then carries out characteristic matching, carries out software design, using Halcon image processing software,
Program is write, the realization of image characteristics extraction and matching technique is completed, and chooses many algorithms and compares test.
4. according to claim 1 a kind of based on the monocular vision three-dimensional rebuilding method for improving Sift algorithm, feature exists
In, the SIFT algorithm improvement specifically includes the following steps: S1, using monocular camera acquiring image from multi-angle, to wherein appointing
It anticipates two width target imagesAnd imageExtreme point is detected using SIFT operator, and screens obtained characteristic point, obtains SIFT spy
Sign point is to collectionWith, S2, calculate phase equalization energy information matrix；S3, given threshold are between 0.01 and 0.05
Value, is compared with the phase equalization energy information matrix being calculated, and rejects maximum square from the SIFT feature for being lower than threshold value
Point obtains characteristic point to be matched to collection to collectionWith；S4, carry out Stereo matching, calculate Euclidean distance closest approach and time
The ratio of near point and the threshold value of setting compare, and judge the similarity degree of SIFT feature pair, if ratio is less than the threshold of setting
Value, then feature point pair matching success, otherwise fails；S5, the Kendall's coefficient for calculating each match point, and calculate it and constrain threshold
Value rejects the match point that Kendall's coefficient is less than threshold value, to obtain final matching double points collection.
5. according to claim 4 a kind of based on the monocular vision three-dimensional rebuilding method for improving Sift algorithm, feature exists
In the step C also needs to complete comparative test, verifies the feasibility of innovatory algorithm.
6. according to claim 1 a kind of based on the monocular vision three-dimensional rebuilding method for improving Sift algorithm, feature exists
In the step D is specifically: utilizing projection matrix, the three-dimensional space of object is calculated in conjunction with the matching double points collection being previously obtained
Point coordinate, then to the three-dimensional point cloud progress triangulation for determining coordinate value, the surface information of reconstruction attractor object, by video camera
Selected in the image shot wherein one it is most suitable, the texture of scene on the figure is mapped in threedimensional model, is recovered
Three-dimensional scene information.
7. according to claim 6 a kind of based on the monocular vision three-dimensional rebuilding method for improving Sift algorithm, feature exists
In the step D needs to carry out software design, using Halcon image processing software, completes three-dimensional reconstruction and realizes, and leads to
The three-dimensional reconstruction result comparison for extracting matching algorithm based on different characteristic is crossed, the feasibility of innovatory algorithm is further verified.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910069847.8A CN109872371A (en) | 2019-01-24 | 2019-01-24 | A kind of monocular vision three-dimensional rebuilding method based on improvement Sift algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910069847.8A CN109872371A (en) | 2019-01-24 | 2019-01-24 | A kind of monocular vision three-dimensional rebuilding method based on improvement Sift algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109872371A true CN109872371A (en) | 2019-06-11 |
Family
ID=66918104
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910069847.8A Pending CN109872371A (en) | 2019-01-24 | 2019-01-24 | A kind of monocular vision three-dimensional rebuilding method based on improvement Sift algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109872371A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110120096A (en) * | 2019-05-14 | 2019-08-13 | 东北大学秦皇岛分校 | A kind of unicellular three-dimensional rebuilding method based on micro- monocular vision |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101639947A (en) * | 2009-08-26 | 2010-02-03 | 北京农业信息技术研究中心 | Image-based plant three-dimensional shape measurement and reconstruction method and system |
| CN102982548A (en) * | 2012-12-11 | 2013-03-20 | 清华大学 | Multi-view stereoscopic video acquisition system and camera parameter calibrating method thereof |
| CN103914847A (en) * | 2014-04-10 | 2014-07-09 | 西安电子科技大学 | SAR image registration method based on phase congruency and SIFT |
| CN107481315A (en) * | 2017-06-29 | 2017-12-15 | 重庆邮电大学 | A kind of monocular vision three-dimensional environment method for reconstructing based on Harris SIFT BRIEF algorithms |
| CN108346162A (en) * | 2018-03-26 | 2018-07-31 | 西安电子科技大学 | Remote sensing image registration method based on structural information and space constraint |
-
2019
- 2019-01-24 CN CN201910069847.8A patent/CN109872371A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101639947A (en) * | 2009-08-26 | 2010-02-03 | 北京农业信息技术研究中心 | Image-based plant three-dimensional shape measurement and reconstruction method and system |
| CN102982548A (en) * | 2012-12-11 | 2013-03-20 | 清华大学 | Multi-view stereoscopic video acquisition system and camera parameter calibrating method thereof |
| CN103914847A (en) * | 2014-04-10 | 2014-07-09 | 西安电子科技大学 | SAR image registration method based on phase congruency and SIFT |
| CN107481315A (en) * | 2017-06-29 | 2017-12-15 | 重庆邮电大学 | A kind of monocular vision three-dimensional environment method for reconstructing based on Harris SIFT BRIEF algorithms |
| CN108346162A (en) * | 2018-03-26 | 2018-07-31 | 西安电子科技大学 | Remote sensing image registration method based on structural information and space constraint |
Non-Patent Citations (1)
| Title |
|---|
| GEORGIOS KORDELAS ET AL: "Robust SIFT-based feature matching using Kendall"s rank correlation measure", 《2009 16TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110120096A (en) * | 2019-05-14 | 2019-08-13 | 东北大学秦皇岛分校 | A kind of unicellular three-dimensional rebuilding method based on micro- monocular vision |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111243093B (en) | Three-dimensional face grid generation method, device, equipment and storage medium | |
| CN104933718B (en) | A kind of physical coordinates localization method based on binocular vision | |
| CN104463108B (en) | A kind of monocular real time target recognitio and pose measuring method | |
| CN106503671B (en) | The method and apparatus for determining human face posture | |
| Alexiadis et al. | An integrated platform for live 3D human reconstruction and motion capturing | |
| CN109242954B (en) | Multi-view three-dimensional human body reconstruction method based on template deformation | |
| CN102999942B (en) | Three-dimensional face reconstruction method | |
| CN108288292A (en) | A kind of three-dimensional rebuilding method, device and equipment | |
| TWI554980B (en) | Method and apparatus for sensing moving ball | |
| WO2021143282A1 (en) | Three-dimensional facial model generation method and apparatus, computer device and storage medium | |
| WO2019140945A1 (en) | Mixed reality method applied to flight simulator | |
| CN107862744A (en) | Aviation image three-dimensional modeling method and Related product | |
| CN109887030A (en) | Texture-free metal parts image position and posture detection method based on the sparse template of CAD | |
| CN106919944A (en) | A kind of wide-angle image method for quickly identifying based on ORB algorithms | |
| CN104599317A (en) | Mobile terminal and method for achieving 3D (three-dimensional) scanning modeling function | |
| CN106127743B (en) | The method and system of automatic Reconstruction bidimensional image and threedimensional model accurate relative location | |
| JP7459051B2 (en) | Method and apparatus for angle detection | |
| CN114119739A (en) | Binocular vision-based hand key point space coordinate acquisition method | |
| CN107015655A (en) | Museum virtual scene AR experiences eyeglass device and its implementation | |
| CN106997617A (en) | The virtual rendering method of mixed reality and device | |
| CN109389634A (en) | Virtual shopping system based on three-dimensional reconstruction and augmented reality | |
| CN110580720A (en) | camera pose estimation method based on panorama | |
| CN113160335A (en) | Model point cloud and three-dimensional surface reconstruction method based on binocular vision | |
| CN108734772A (en) | High accuracy depth image acquisition methods based on Kinect fusion | |
| CN109872371A (en) | A kind of monocular vision three-dimensional rebuilding method based on improvement Sift algorithm |
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: 20190611 |
|
| WD01 | Invention patent application deemed withdrawn after publication |