CN106815824A - A kind of image neighbour's optimization method for improving extensive three-dimensional reconstruction efficiency - Google Patents
A kind of image neighbour's optimization method for improving extensive three-dimensional reconstruction efficiency Download PDFInfo
- Publication number
- CN106815824A CN106815824A CN201611124130.1A CN201611124130A CN106815824A CN 106815824 A CN106815824 A CN 106815824A CN 201611124130 A CN201611124130 A CN 201611124130A CN 106815824 A CN106815824 A CN 106815824A
- Authority
- CN
- China
- Prior art keywords
- image
- image pair
- pair
- characteristic point
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000005457 optimization Methods 0.000 title claims abstract description 12
- 230000008859 change Effects 0.000 claims abstract description 31
- 239000011159 matrix material Substances 0.000 claims abstract description 22
- 238000001914 filtration Methods 0.000 claims abstract description 12
- 238000012795 verification Methods 0.000 claims description 13
- 238000000605 extraction Methods 0.000 claims description 5
- 230000000717 retained effect Effects 0.000 claims description 4
- 239000000203 mixture Substances 0.000 claims description 3
- 238000004321 preservation Methods 0.000 claims description 3
- 238000005303 weighing Methods 0.000 claims description 3
- 230000006872 improvement Effects 0.000 description 5
- 230000008901 benefit Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 239000002699 waste material Substances 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 201000011243 gastrointestinal stromal tumor Diseases 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- 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
- G06T2207/10004—Still image; Photographic image
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of image neighbour's optimization method for improving extensive three-dimensional reconstruction efficiency, neighborhood matching image pair is obtained first, to image to carrying out Feature Points Matching；Geometry is carried out by basis matrix and verified being unsatisfactory for the error hiding of epipolar-line constraint to reject and obtaining interior points, and calculate homography matrix obtaining homograph rate；Then the change between statistical match characteristic point on direction and yardstick, obtains corresponding histogram；Measure the similitude of image and mark redundant image therein by the triple constraints of the change histogram in interior points, homograph rate, yardstick and direction；The image pair comprising redundant image is rejected, and narrow baseline image pair is rejected by interior points, homograph rate；The image after filtering is finally preserved to match information, redundant image pair and the narrower image pair of baseline has been eliminated, the precision and efficiency of follow-up three-dimensional reconstruction is further increased.
Description
Technical field
The invention belongs to computer vision field, more particularly, to a kind of figure for improving extensive three-dimensional reconstruction efficiency
As neighbour's optimization method.
Background technology
The three-dimensional scenic of extensive pictures is the popular research field of a comparing in recent years.Three-dimensional reconstruction is current
Algorithm generally is exercise recovery structure (Structure from Motion, SFM) algorithm of increment type, it is main include with
Lower four parts：1) picture feature point is extracted, 2) characteristic matching between image, 3) to images match to carrying out geometry verification,
4) camera attitude and sparse three-dimensional point cloud are estimated according to matching.For large-scale dataset, key issue is efficiency.Press
According to flow above, the bottleneck of current efficiency of algorithm mainly appears on second step and the 3rd step, the wherein original mode of second step
To be matched two-by-two, but for large-scale dataset, most of picture be no scene overlap, it is incoherent,
If these pictures carry out matching and will waste the substantial amounts of time.Therefore for the improvement of second step, main approach is exactly to pass through
Certain high efficiency mode approx finds the image pair for having scene to overlap, so as to reduce follow-up match time.Change in this respect
The space entered is very big, and in fact many scholars are exactly to do a lot of work in this respect.Frahm GISTs a kind of to image zooming-out is special
Levy, and clustered according to the similitude of this feature, find presentation graphics therein, so as to reduce image pair.Agarwal
Trained by characteristics of image and obtain a words tree, the near of each image is found with a kind of Image Retrieval Mechanism by this tree
Neighbour, matching is only carried out between neighbour.Chao has done a kind of improvement on the basis of Agarwal work, with the method for on-line study
To be ranked up to retrieval result so as to improve the accuracy of neighbour.Wu match obtaining figure with the characteristic point after down-sampled
Similarity as between, so as to find image pair.Each characteristic point in picture is mapped to a spy for training by Havlena
Certain word of dictionary is levied, the characteristic point of the same word of correspondence is matching, all of picture is all matched together.
For the 3rd step, the basis matrix F between image is mainly estimated in geometry verification, then carries out pole to matching
Line checksum filter error hiding.Many people propose efficient RANSAC modified versions to estimate basis matrix so as to improve efficiency, separately
Outer Raguram proposes a kind of mode of on-line study to improve the efficiency of geometry verification.
Because the bottleneck of second step can cause the efficiency entire lowering of follow-up process, many work at present is concentrated on to the
The improvement of two steps.It is exactly to train to obtain a words tree by the SIFT feature of picture wherein to compare classical method
(vocabulary tree), k neighbour of every pictures is found according to this words tree, and this k neighbour is the picture
Matching image.Although the method can preferably reject some invalid images pair, the neighbour for wherein finding also contains
Some problems：1) image of redundancy, it is assumed that a kind of extreme situation, to same width picture reproduction many times over (equivalent to same
Individual position, many pictures of same angle shot), then will constitute neighbour between these pictures, but these pictures it
Between match information be nonsensical.The neighbour that others picture finds in addition may be the duplicate of same pictures,
The matching of these neighbours contains many redundancies, wastes many times.2) the very narrow neighbour of baseline, baseline is too narrow
Image is to that can cause the decline of reconstruction precision.
The content of the invention
For the disadvantages described above or Improvement requirement of prior art, extensive three-dimensional reconstruction effect is improved the invention provides one kind
Image neighbour's optimization method of rate, the method is being carried out on the basis of image neighbor search obtains initial pictures pair, by feature
Point matching, by geometry verification, the yardstick between homograph rate and characteristic point and the triple constraints of direction change come near to image
Neighbour is further optimized, and substantially increases the precision and efficiency of follow-up three-dimensional reconstruction.Thus solve in the prior art to warp
Cross the redundancy and the narrower technical problem of baseline existed after image neighbor search.
To achieve the above object, according to one aspect of the present invention, there is provided one kind improves extensive three-dimensional reconstruction efficiency
Image neighbour's optimization method, including：
(1) inceptive filtering of feature point extraction and image pair：Characteristic point is extracted to each image, then using image retrieval
Method find the neighbour of each image, obtain the image pair of scene overlap；
(2) Feature Points Matching and related data prepare：For each image pair, the image to image pair carries out characteristic point
Matching, carries out geometry verification to reject error hiding by the basis matrix between image pair image, is met basis matrix
Feature Points Matching number, as interior points, and characteristic point according to matching calculates optimal homography matrix and obtain homograph rate,
Then change of the characteristic point of statistical match on direction and yardstick, obtains the change histogram in direction and the change Nogata of yardstick
Figure；
(3) redundant image is marked：By the interior points of each image pair, homograph rate, direction change histogram
And similarity of the triple constraints of change histogram of yardstick to measure between each image pair image, similarity is met into default rule
The image composition multiple images set of all image pairs then, chooses an image in each image collection as representative graph
Picture, it is redundant image to mark other images；
(4) further filtering optimizes image pair：The image pair comprising redundant image is rejected, and by interior points, Dan Ying
Interconversion rate rejects narrow baseline image pair；
(5) preservation of information after optimizing：Preserve after filtering between the image pair that obtains and image pair image
With information.
Preferably, step (2) specifically includes following sub-step：
(2.1) for each image pair, the characteristic point between image pair image is carried out using Feature Points Matching algorithm
Matching；
(2.2) geometry verification is carried out by estimating the basis matrix F between image pair image, error hiding is rejected, is obtained
Interior points m is to the Feature Points Matching number for meeting F；
(2.3) calculate optimal homography matrix H and obtain corresponding homograph rate h according to the matching characteristic point for meeting F,
Wherein, h is used for weighing the baseline width between image pair image；
(2.4) direction for asking difference operation to obtain characteristic point is carried out on the direction o of characteristic point to the matching characteristic point for meeting F
Changes delta o on o, carries out the changes delta s, Ran Houli on the yardstick s that division arithmetic obtains characteristic point on the yardstick s of characteristic point
Dispersion obtains the histogram of Δ o and the histogram of Δ s.
Preferably, step (3) specifically includes following sub-step：
(3.1) for each image pair, if the interior points m of image pair is more than threshold alpha and homograph rate h is more than threshold value
β, then perform step (3.2)；
(3.2) if peak value p of the Δ s histograms of image pair near 1 is more than threshold value η and the histogrammic tops of Δ o
Value q is more than threshold θ, then by the image of the image pair alternately redundancy；
(3.3) all alternative redundancys are constituted into independent multiple images set, for each image collection, only retaining should
Used as representative image, remaining image tagged is redundant image to a most image of matching number in image collection.
Preferably, step (4) specifically includes following sub-step：
(4.1) for each image pair, if image pair includes redundant image, the image comprising redundant image is rejected
It is right, otherwise perform step (4.2)；
(4.2) if the interior points m of image pair is more than threshold gamma and homograph rate h is less than threshold value δ, the image is retained
It is right, otherwise reject the image pair；
(4.3) judge whether that traversal completes all of image pair, if not completing, perform step (4.1), otherwise terminate
Flow.
In general, there is following skill compared with prior art, mainly by the contemplated above technical scheme of the present invention
Art advantage：
(1) a kind of method of new measurement image similarity is proposed to filter the too high redundant image of similitude,
Image is solved to the Redundancy by still existing after traditional image neighbor search.
(2) combine geometry verification and homograph rate rejects narrow baseline image pair, solve image to by traditional figure
As the narrower problem of the baseline still existed after neighbor search.In the case where almost time complexity is not increased, significantly
Improve the precision and efficiency of follow-up three-dimensional reconstruction.
Brief description of the drawings
Fig. 1 is a kind of image neighbour's optimization method for improving extensive three-dimensional reconstruction efficiency disclosed in the embodiment of the present invention
Schematic flow sheet；
Fig. 2 is a kind of schematic flow sheet of mark redundant image disclosed in the embodiment of the present invention；
Fig. 3 is the yardstick and the histogrammic specific embodiment of direction change of a kind of two nonredundancy images；
Fig. 4 is the yardstick and the histogrammic specific embodiment of direction change of a kind of two redundant images.
Specific 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 long as additionally, technical characteristic involved in invention described below each implementation method
Not constituting conflict each other can just be mutually combined.
The method is related to feature point extraction, and large-scale image is to filtering, quick hash Feature Points Matchings algorithm, geometry school
Test, the calculating of homograph rate, direction and the histogrammic structure of change of scale between matching characteristic point, a kind of new measurement image
The technologies, the image pair of optimization such as the method for similitude, a kind of further method for rejecting redundant image pair and narrow baseline image pair
The form for preserving into file with corresponding match information can be directly used for follow-up three-dimensional reconstruction process, optimized reconstruction precision and effect
Rate.
Fig. 1 is a kind of image neighbour's optimization method for improving extensive three-dimensional reconstruction efficiency disclosed in the embodiment of the present invention
Schematic flow sheet, comprises the following steps in the method shown in Fig. 1：
S1：The inceptive filtering of feature point extraction and image pair；
Wherein, step S1 specifically includes following operation：Characteristic point is extracted to each image, then using the side of image retrieval
Method finds the neighbour of each image, obtains the image pair of scene overlap.
Wherein, step S1 specifically includes following sub-step：
(S1.1) Scale invariant features transform (Scale-invariant feature transform, SIFT) feature is used
Extraction algorithm extracts the characteristic point of each image；
(1.2) trained with SIFT feature and obtain a words tree (vocabulary tree), searched for by words tree and obtained
K neighbour of each image, obtains the images match pair of scene overlap.
S2：Feature Points Matching and related data prepare；
Wherein, step S2 specifically includes following operation：For each image pair, the image to image pair carries out characteristic point
Matching, carries out geometry verification to reject error hiding by the basis matrix between image pair image, is met basis matrix
Feature Points Matching number, as interior points, and characteristic point according to matching calculates optimal homography matrix and obtain homograph rate,
Then change of the characteristic point of statistical match on direction and yardstick, obtains the change histogram in direction and the change Nogata of yardstick
Figure.
Wherein, the change histogram in the interior points of each image pair, homograph rate, direction can be obtained by step S2
With the change histogram of yardstick.
Wherein, step S2 specifically includes following sub-step：
(S2.1) for each image pair, the characteristic point between image pair image is carried out using Feature Points Matching algorithm
Matching；
Wherein it is possible to be carried out to the characteristic point between image pair image using quick hash Feature Points Matchings algorithm
Match somebody with somebody.
(S2.2) geometry verification is carried out by estimating the basis matrix F between image pair image, error hiding is rejected, is obtained
Interior points m is to the Feature Points Matching number for meeting F；
Wherein it is possible to combine at 8 points using consistent (RANdom SAmple Consensus, the RANSAC) algorithm of random sampling
Method estimates the basis matrix F between image pair image.
(S2.3) calculate optimal homography matrix H and obtain corresponding homograph rate according to the matching characteristic point for meeting F
H, wherein, h is used for weighing the baseline width between image pair image；
Wherein it is possible to calculate optimal homography matrix H by RANSAC strategies
(S2.4) side for asking difference operation to obtain characteristic point is carried out on the direction o of characteristic point to the matching characteristic point for meeting F
To the changes delta o on o, the changes delta s on the yardstick s that division arithmetic obtains characteristic point is carried out on the yardstick s of characteristic point, then
Discretization obtains the histogram of Δ o and the histogram of Δ s.
Wherein it is possible to the histogram of Δ o and the histogram of Δ s are obtained in the following manner, for each image pair, to full
The matching characteristic point of sufficient F carries out asking difference operation and enterprising in s (yardstick of representative feature point) on o (direction of representative feature point)
Row division arithmetic, obtains corresponding changes delta o, Δ s in two dimensions.Δ s is normalized to [0,4], Δ o and normalizes to [- 2
π, 2 π], n bucket is then divided into respectively.For each matching characteristic point, changes delta o, the Δ s in two dimensions are calculated, so
Amplitude adds 1 inside a bucket of its nearest neighbours afterwards, and the characteristic point for counting all matchings can be just obtained on two kinds of changes delta o,
The histogram of Δ s.As shown in Figure 3, Figure 4, two images are the image being compared above, below the left side figure be statistics o
The Δ o histograms that difference change is obtained；The right is to count the Δ s histograms that the ratio value changes of s are obtained.Δ o abscissa scopes are
[- 2 π, 2 π], the abscissa scope of Δ s is [0,4].
It should be noted that above-mentioned be only enumerated the histogrammic specific of a kind of histogram for how obtaining Δ o and Δ s
Embodiment, above-mentioned number range should not be construed as limiting the uniqueness of the embodiment of the present invention.
Wherein, direction o and yardstick s：The dimensional information and directional information of characteristic point are contained in SIFT feature.
Δ o, Δ s：Change on direction and yardstick.
S3：Mark redundant image；
Wherein, step S3 specifically includes following operation：By the interior points of each image pair, homograph rate, direction
Triple similarities of the constraint to measure between each image pair image of the change histogram of change histogram and yardstick, will be similar
Degree meets the image composition multiple images set of all image pairs of preset rules, chooses an image in each image collection
Used as representative image, it is redundant image to mark other images.
Wherein, the redundant image in all images can be obtained by step S3.
Wherein, step S3 specifically includes following sub-step, is illustrated in figure 2 a kind of mark disclosed in the embodiment of the present invention superfluous
The schematic flow sheet of remaining image, the method shown in Fig. 2 includes following operation：
(S3.1) for each image pair, if the interior points m of image pair is more than threshold alpha and homograph rate h is more than threshold
Value β, then perform step (3.2)；
Wherein, α and β can rule of thumb be configured.
Wherein, if the interior points m of image pair represents that picture material is more similar more than threshold alpha.
The similitude of picture material：The number counted out according to matching characteristic judges, matching characteristic count out it is more, it is interior
Rong Yue is similar.
Wherein, homograph rate h represents that baseline is narrower more than threshold value beta.
Baseline width：The homograph rate tried to achieve according to homography matrix judges that homograph rate is bigger, and baseline is narrower.
(S3.2) if peak value p of the Δ s histograms of image pair near 1 is more than threshold value η and the histogrammic tops of Δ o
Value q is more than threshold θ, then by the image of the image pair alternately redundancy；
Wherein, peak value p of the Δ s histograms of image pair near 1 represents the yardstick between matching characteristic point more than threshold value η
Close, the histogrammic peak-peak q of Δ o represent that the anglec of rotation change between matching characteristic point is consistent more than threshold θ.If both
All meet, then it represents that the image is alternately superfluous by the two images of the image pair to being similitude image pair very high
Remainder.
The close dimensional variation represented between matching characteristic point of yardstick less, can be judged by Δ s histograms.
The consistent anglec of rotation change represented between matching characteristic point of anglec of rotation change is unanimous on the whole, can be straight by Δ o
Square figure is judged.
Wherein, η and θ can rule of thumb be configured.
As shown in figure 3, be can be seen by following histogram, although Δ o histograms have high peaks in a certain position, but
It is do not have obvious peak value in Δ s histograms, and without integrated distribution near 1.Can be apparent by original image
Have obvious different scale between image, therefore, it is determined that for non-similarity compared with hi-vision；Tested by another set, such as Fig. 4 institutes
Show.It may be seen that Δ o histograms have obvious peak value, and concentration point in having high peaks, and Δ s histograms near 0
Cloth is near 1, therefore, it is determined that this pair of image is similitude compared with hi-vision.
Fig. 3, Fig. 4 represent a kind of experimental verification to image neighbour optimization method proposed by the present invention, and it is right to should not be construed as
Uniqueness of the invention is limited.
(S3.3) all alternative redundancys are constituted into independent multiple images set, for each image collection, is only retained
Used as representative image, remaining image tagged is redundant image to a most image of matching number in the image collection.
Wherein, representative image represents the most information comprising other images in set where it, it is ensured that rebuild knot
The information that fruit is lost is minimum.
S4：Further filtering optimizes image pair；
Wherein, step S4 specifically includes following operation：Reject comprising redundant image image pair, and by interior points,
Homograph rate rejects narrow baseline image pair.
Wherein, step (4) specifically includes following sub-step：
(S4.1) for each image pair, if image pair includes redundant image, the figure comprising redundant image is rejected
As right, step (S4.2) is otherwise performed；
(S4.2) if the interior points m of image pair is more than threshold gamma and homograph rate h is less than threshold value δ, the figure is retained
As right, the image pair is otherwise rejected；
Wherein, γ and δ can rule of thumb be configured.
(S4.3) judge whether that traversal completes all of image pair, if not completing, perform step (4.1), otherwise tie
Line journey.
S5：The preservation of information after optimization.
Wherein, step S5 specifically includes following operation：Preserve after filtering the image pair that obtains and image pair image it
Between match information.
Match information between the image pair and image pair image that are obtained after final filtration is saved in associated documents
In, these files can be used for subsequent reconstruction, and geometry verification need not be again carried out during reconstruction and single strain is calculated
Change.
As it will be easily appreciated by one skilled in the art that the foregoing is only presently preferred embodiments of the present invention, it is not used to
The limitation present invention, all any modification, equivalent and improvement made within the spirit and principles in the present invention etc., all should include
Within protection scope of the present invention.
Claims (4)
1. a kind of image neighbour's optimization method for improving extensive three-dimensional reconstruction efficiency, it is characterised in that including：
(1) inceptive filtering of feature point extraction and image pair：Characteristic point is extracted to each image, then using the side of image retrieval
Method finds the neighbour of each image, obtains the image pair of scene overlap；
(2) Feature Points Matching and related data prepare：For each image pair, the image to image pair carries out characteristic point
Match somebody with somebody, carry out geometry verification by the basis matrix between image pair image to reject error hiding, be met basis matrix
Feature Points Matching number, as interior points, and homograph rate is obtained according to the characteristic point optimal homography matrix of calculating for matching, connect
Change of the characteristic point of statistical match on direction and yardstick, the change histogram in direction and the change Nogata of yardstick is obtained
Figure；
(3) redundant image is marked：By the interior points of each image pair, homograph rate, direction change histogram and chi
Similarity of the triple constraints of change histogram of degree to measure between each image pair image, preset rules are met by similarity
The image composition multiple images set of all image pairs, chooses an image in each image collection as representative image, mark
Remember that other images are redundant image；
(4) further filtering optimizes image pair：The image pair comprising redundant image is rejected, and by interior points, homograph
Rate rejects narrow baseline image pair；
(5) preservation of information after optimizing：Preserve the matching letter between the image pair and image pair image obtained after filtering
Breath.
2. method according to claim 1, it is characterised in that step (2) specifically includes following sub-step：
(2.1) for each image pair, the characteristic point between image pair image is matched using Feature Points Matching algorithm；
(2.2) geometry verification is carried out by estimating the basis matrix F between image pair image, error hiding is rejected, is expired
The Feature Points Matching number of sufficient F is interior points m；
(2.3) calculate optimal homography matrix H and obtain corresponding homograph rate h according to the matching characteristic point for meeting F, its
In, h is used for weighing the baseline width between image pair image；
(2.4) on the direction o for carrying out asking difference operation to obtain characteristic point on the direction o of characteristic point to the matching characteristic point for meeting F
Changes delta o, the changes delta s on the yardstick s that division arithmetic obtains characteristic point is carried out on the yardstick s of characteristic point, it is then discrete
Change obtains the histogram of Δ o and the histogram of Δ s.
3. method according to claim 2, it is characterised in that step (3) specifically includes following sub-step：
(3.1) for each image pair, if the interior points m of image pair is more than threshold alpha and homograph rate h is more than threshold value beta,
Perform step (3.2)；
(3.2) if peak value p of the Δ s histograms of image pair near 1 is more than threshold value η and the histogrammic peak-peak q of Δ o are big
In threshold θ, then by the image of the image pair alternately redundancy；
(3.3) all alternative redundancys are constituted into independent multiple images set, for each image collection, only retains the image
Used as representative image, remaining image tagged is redundant image to a most image of matching number in set.
4. method according to claim 3, it is characterised in that step (4) specifically includes following sub-step：
(4.1) for each image pair, if image pair includes redundant image, the image pair comprising redundant image is rejected,
Otherwise perform step (4.2)；
(4.2) if the interior points m of image pair is more than threshold gamma and homograph rate h is less than threshold value δ, the image pair is retained,
Otherwise reject the image pair；
(4.3) judge whether that traversal completes all of image pair, if not completing, perform step (4.1), otherwise terminate stream
Journey.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611124130.1A CN106815824B (en) | 2016-12-08 | 2016-12-08 | A kind of image neighbour's optimization method improving extensive three-dimensional reconstruction efficiency |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611124130.1A CN106815824B (en) | 2016-12-08 | 2016-12-08 | A kind of image neighbour's optimization method improving extensive three-dimensional reconstruction efficiency |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106815824A true CN106815824A (en) | 2017-06-09 |
| CN106815824B CN106815824B (en) | 2019-07-05 |
Family
ID=59107113
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611124130.1A Active CN106815824B (en) | 2016-12-08 | 2016-12-08 | A kind of image neighbour's optimization method improving extensive three-dimensional reconstruction efficiency |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106815824B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108154531A (en) * | 2018-01-03 | 2018-06-12 | 深圳北航新兴产业技术研究院 | A kind of method and apparatus for calculating body-surface rauma region area |
| CN110176060A (en) * | 2019-04-28 | 2019-08-27 | 华中科技大学 | Dense three-dimensional rebuilding method and system based on the guidance of multiple dimensioned Geometrical consistency |
| CN111133474A (en) * | 2017-09-29 | 2020-05-08 | 日本电气方案创新株式会社 | Image processing apparatus, image processing method, and computer-readable recording medium |
| CN111402412A (en) * | 2020-04-16 | 2020-07-10 | Oppo广东移动通信有限公司 | Data acquisition method and device, equipment and storage medium |
| WO2020176238A1 (en) * | 2019-02-25 | 2020-09-03 | Sony Corporation | Method of merging point clouds that identifies and retains preferred points |
| CN112729254A (en) * | 2020-12-23 | 2021-04-30 | 西南交通大学 | Aerial triangulation calculation method suitable for multi-gradient aerial belt oblique photogrammetry |
| CN112822478A (en) * | 2020-12-31 | 2021-05-18 | 杭州电子科技大学 | High-quality photo sequence acquisition method for three-dimensional reconstruction |
| CN113936085A (en) * | 2021-12-17 | 2022-01-14 | 荣耀终端有限公司 | Three-dimensional reconstruction method and device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101650178A (en) * | 2009-09-09 | 2010-02-17 | 中国人民解放军国防科学技术大学 | Method for image matching guided by control feature point and optimal partial homography in three-dimensional reconstruction of sequence images |
| CN101877140A (en) * | 2009-12-18 | 2010-11-03 | 北京邮电大学 | Panorama-based panoramic virtual tour method |
| CN101908231A (en) * | 2010-07-27 | 2010-12-08 | 清华大学 | Reconstruction method and system for processing three-dimensional point cloud containing main plane scene |
| CN103617616A (en) * | 2013-12-03 | 2014-03-05 | 康江科技(北京)有限责任公司 | Affine invariant image matching method |
| CN104392453A (en) * | 2014-12-01 | 2015-03-04 | 南京邮电大学 | Method of matching and optimizing ransac features based on polar-line insertion image |
-
2016
- 2016-12-08 CN CN201611124130.1A patent/CN106815824B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101650178A (en) * | 2009-09-09 | 2010-02-17 | 中国人民解放军国防科学技术大学 | Method for image matching guided by control feature point and optimal partial homography in three-dimensional reconstruction of sequence images |
| CN101877140A (en) * | 2009-12-18 | 2010-11-03 | 北京邮电大学 | Panorama-based panoramic virtual tour method |
| CN101908231A (en) * | 2010-07-27 | 2010-12-08 | 清华大学 | Reconstruction method and system for processing three-dimensional point cloud containing main plane scene |
| CN103617616A (en) * | 2013-12-03 | 2014-03-05 | 康江科技(北京)有限责任公司 | Affine invariant image matching method |
| CN104392453A (en) * | 2014-12-01 | 2015-03-04 | 南京邮电大学 | Method of matching and optimizing ransac features based on polar-line insertion image |
Non-Patent Citations (5)
| Title |
|---|
| KUN SUN 等: "Point Sets Matching by Feature-Aware Mixture Point Matching Algorithm", 《EMMCVPR 2015》 * |
| KUN SUN 等: "Progressive match expansion via coherent subspace constraint", 《INFORMATION SCIENCES》 * |
| 仲明: "基于特征点精确配准的图像拼接技术的研究", 《中国优秀硕士学位论文全文数据库 信息科技辑（月刊）》 * |
| 曹芳 等: "基于匹配点相似度引导采样的图像多平面检测", 《计算机应用研究》 * |
| 蒋昭炎: "基于图像的大场景三维重建并行算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑（月刊）》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111133474A (en) * | 2017-09-29 | 2020-05-08 | 日本电气方案创新株式会社 | Image processing apparatus, image processing method, and computer-readable recording medium |
| CN111133474B (en) * | 2017-09-29 | 2023-09-19 | 日本电气方案创新株式会社 | Image processing apparatus, image processing method, and computer-readable recording medium |
| CN108154531A (en) * | 2018-01-03 | 2018-06-12 | 深圳北航新兴产业技术研究院 | A kind of method and apparatus for calculating body-surface rauma region area |
| CN108154531B (en) * | 2018-01-03 | 2021-10-08 | 深圳北航新兴产业技术研究院 | Method and device for calculating area of body surface damage region |
| WO2020176238A1 (en) * | 2019-02-25 | 2020-09-03 | Sony Corporation | Method of merging point clouds that identifies and retains preferred points |
| CN110176060A (en) * | 2019-04-28 | 2019-08-27 | 华中科技大学 | Dense three-dimensional rebuilding method and system based on the guidance of multiple dimensioned Geometrical consistency |
| CN110176060B (en) * | 2019-04-28 | 2020-09-18 | 华中科技大学 | Dense three-dimensional reconstruction method and system based on multi-scale geometric consistency guidance |
| CN111402412A (en) * | 2020-04-16 | 2020-07-10 | Oppo广东移动通信有限公司 | Data acquisition method and device, equipment and storage medium |
| CN111402412B (en) * | 2020-04-16 | 2023-06-09 | Oppo广东移动通信有限公司 | Data acquisition method and device, equipment and storage medium |
| CN112729254A (en) * | 2020-12-23 | 2021-04-30 | 西南交通大学 | Aerial triangulation calculation method suitable for multi-gradient aerial belt oblique photogrammetry |
| CN112822478A (en) * | 2020-12-31 | 2021-05-18 | 杭州电子科技大学 | High-quality photo sequence acquisition method for three-dimensional reconstruction |
| CN113936085A (en) * | 2021-12-17 | 2022-01-14 | 荣耀终端有限公司 | Three-dimensional reconstruction method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106815824B (en) | 2019-07-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106815824B (en) | A kind of image neighbour's optimization method improving extensive three-dimensional reconstruction efficiency | |
| Diem et al. | cBAD: ICDAR2017 competition on baseline detection | |
| Xie et al. | Discriminative object tracking via sparse representation and online dictionary learning | |
| Ding et al. | Multi-scale fully convolutional network for gland segmentation using three-class classification | |
| Reddy Mopuri et al. | Object level deep feature pooling for compact image representation | |
| Li et al. | A weighted sparse coding framework for saliency detection | |
| Liang et al. | Multi-spectral fusion based approach for arbitrarily oriented scene text detection in video images | |
| Pape et al. | Utilizing machine learning approaches to improve the prediction of leaf counts and individual leaf segmentation of rosette plant images | |
| CN108334848A (en) | A kind of small face identification method based on generation confrontation network | |
| WO2020253063A1 (en) | Method and device for searching for similar images | |
| CN104850822B (en) | Leaf identification method under simple background based on multi-feature fusion | |
| Pirrone et al. | Papy-s-net: A siamese network to match papyrus fragments | |
| CN103839074B (en) | Image classification method based on matching of sketch line segment information and space pyramid | |
| CN107368803A (en) | A kind of face identification method and system based on classification rarefaction representation | |
| Islam et al. | Geometry and statistics-preserving manifold embedding for nonlinear dimensionality reduction | |
| Xu et al. | A robust hierarchical detection method for scene text based on convolutional neural networks | |
| Gonde et al. | Volumetric local directional triplet patterns for biomedical image retrieval | |
| CN111199801A (en) | Construction method and application of model for identifying disease types of medical records | |
| Shi et al. | MGSNet: A multi-scale and gated spatial attention network for crowd counting | |
| Xu et al. | Robust seed localization and growing with deep convolutional features for scene text detection | |
| CN110728214B (en) | Weak and small figure target detection method based on scale matching | |
| Li | Saliency prediction based on multi-channel models of visual processing | |
| Lakshmi et al. | Robust algorithm for Telugu word image retrieval and recognition | |
| Jindal et al. | Analysis and Classification of Medicinal Plants found in India | |
| Grouver et al. | A spatial density and phase angle based correlation for multi-type family photo identification |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |