CN106952227A - A kind of method for sequence image auto-sequencing of taking photo by plane - Google Patents
A kind of method for sequence image auto-sequencing of taking photo by plane Download PDFInfo
- Publication number
- CN106952227A CN106952227A CN201710138301.4A CN201710138301A CN106952227A CN 106952227 A CN106952227 A CN 106952227A CN 201710138301 A CN201710138301 A CN 201710138301A CN 106952227 A CN106952227 A CN 106952227A
- Authority
- CN
- China
- Prior art keywords
- image
- sequence
- sigma
- translation
- correlation
- 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 48
- 238000012163 sequencing technique Methods 0.000 title claims abstract description 17
- 238000013519 translation Methods 0.000 claims abstract description 38
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 23
- 238000001228 spectrum Methods 0.000 claims description 7
- 230000009466 transformation Effects 0.000 claims description 6
- 238000013461 design Methods 0.000 claims description 4
- 241000208340 Araliaceae Species 0.000 claims description 3
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims description 3
- 235000003140 Panax quinquefolius Nutrition 0.000 claims description 3
- 235000008434 ginseng Nutrition 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 2
- 238000005259 measurement Methods 0.000 claims 2
- 238000006243 chemical reaction Methods 0.000 claims 1
- 239000011159 matrix material Substances 0.000 claims 1
- 230000008859 change Effects 0.000 abstract description 5
- 238000004064 recycling Methods 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000010977 jade Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 230000000007 visual effect Effects 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
- G06T3/4084—Scaling of whole images or parts thereof, e.g. expanding or contracting in the transform domain, e.g. fast Fourier transform [FFT] domain scaling
-
- 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
- G06T3/60—Rotation of whole images or parts thereof
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of method for sequence image auto-sequencing of taking photo by plane, limitation of the conventional phase correlation method to picture size uniformity is breached, it is proposed that one kind is based on the improved image sequence Sorting algorthm scheme of phase correlation method.The program proposes to represent the relation of two images to be matched using the mode of log-polar, establish rotation, translation, the order models of change of scale, sequence head, tail image are determined by maximum relation degree, coordinate where recycling peak value determines translation parameters, image right position relation is judged according to given criterion, manual intervention is avoided, the confounding issues of left and right translation can be also eliminated, can accurately carry out the sequence of sequence image.Instant invention overcomes the auto-sequencing that conventional phase correlation method can not complete to translate between image, rotate, scale complex relationship, the scope of application of algorithm is enhanced, certain basis has been established for Panorama Mosaic of taking photo by plane, with very strong practical value.
Description
Technical field
The present invention relates to Aerial Images process field, the method for specifically a kind of sequence image auto-sequencing of taking photo by plane.
Background technology
In recent years, because unmanned plane has the advantage of high flexibility, high efficiency and low cost, it is extensive
Applied to Natural Disaster Evaluation, battle reconnaissance, the field such as environmental monitoring.In order to expand the visual field, fully understand and analysis shooting letter
Breath, effective image mosaic technology is extremely important.The image obtained by unmanned aerial vehicle remote sensing platform has that data volume is big, phase is shaken
It is small, the features such as degree of overlapping is high, the order of confusing image sequence is easy in post processing image, follow-up figure can be so given
As registration and fusion band carry out certain puzzlement, it is impossible to reach preferable splicing effect.And the calculation that splicings many at present is related to
Method[3-4]Require that artificial sequence image sequence could effectively be spliced, such artificial setting be it is very time-consuming,
Particularly for Aerial Images data, it is impossible to meet the requirement of real-time.The automatic of image sequence is completed using computer
Ordering techniques have been suggested, and are widely studied.Kuglin and Hines (Kuglin C, Hines D.The
phase correlation image alignment method[A].Conference on Cybernetics and
Society[C].1975:163-165.) 1975 find that phase correlation method is unrelated with scene, can be exactly in good condition
Under two-dimension translational image is alignd., Hua Shungang et al. (Hua Shungang, Zeng Lingyi, a kind of Europe ancestor's quick posts of beautiful jade in 2006
Face Panorama Mosaic algorithm [J] data acquisition and processions, 2006,04:434-438.) propose one kind and utilize equidistant
The thinking matched somebody with somebody is used for realizing the auto-sequencing of image sequence, and this method is not only computationally intensive, to the robustness of various interference environments
Also it is poor；Zhao Hui in 2007 et al. (Zhao Hui, Chen Hui, in a kind of Chinese images of improved panorama sketch automatic Mosaic algorithm [J] of deep
Figure journal, 2007,02:The auto-sequencing of image sequence 336-342) is realized using phase correlation method, this method can have
There is certain antijamming capability, but need manually selected threshold value, the adaptation of algorithm and automaticity are substantially reduced.The same year, Zhao Wan
A kind of gold et al. (image sequence auto-sequencings for image mosaic of [8] Zhao Wanjin, Gong Shengrong, Liu Quan, Shen Xiangjun, Liu Chunping
Algorithm [J] Journal of Image and Graphics, 2007,10:1861-1864) differentiated using the correlation between phase correlation method image
Position relationship, but require that the size of sequence image must be identical, and the position relationship of image affirms and bright not enough understands really.
, a kind of Wu Xianxiang et al. (improved sequence image Sorting algorthm [J] light of [9] Wu Xianxiang, Guo Baolong, Wang Juan in 2009
Electronics laser, 2009,08:1114-1117.) make two images size identical again with phase phase using the method for afterbody zero padding
Pass method ordering chart picture, but this method does not provide processing to rotation image.
The content of the invention
Problem to be solved by this invention is when overcoming existing Aerial Images using phase correlation method progress auto-sequencing
To the necessary identical limitation of image size, it is proposed that the translation between image, rotation are represented using the mode of log-polar, is contracted
The relation put, sets up image sequence order models and is verified by the derivation of equation, while providing maximum relation degree criterion and peak
Value coordinate judges the principle of adjacent image position relationship, enhances the scope of application of algorithm.
The present invention specifically uses following technical scheme：
Invention one：For the Aerial Images extracted from sensor, not only comprising translation, rotation, also greatly may be used
Can there is a situation where scaling.The relation of two images to be matched is represented set forth herein the mode of log-polar, is built
Rotation, translation, the order models of change of scale have been found, the scope of application of algorithm is enhanced.
Principles illustrated：Define the translation parameter x between two images to be spliced0、y0, anglec of rotation α, zoom factor σ, then f1
(x, y) and f2The position relationship of (x, y) is represented by：
f1(x, y)=f2(σxcosα+σysinα-x0,-σxsinα+σycosα-y0)
Corresponding Fourier transformation is：
Step B, make F1、F2Modulus value be respectively M1、M2, then above formula is turned to：
U=ρ cos β, v=ρ sin β are made, carrying out coordinate to above formula is transformed to：
I.e.：
When two images are only existed, translation, rotation transformation when, now twiddle factor is x=x0, σ=σ0, then have
For above formula, anglec of rotation α=α can be calculated using common phase correlation method0。
When deciding anglec of rotation α0When, then have：
Log ρ=m, log σ=n are made, then ρ=em, σ=enDai Huike is obtained：
M1(em, α) and=e-2nM2(en-m,α-α0)
I.e.：
M1(m, α)=e-2kM2(m-n,α-α0)
Equally n and twiddle factor σ can be calculated with common phase correlation method.The anglec of rotation α calculated0With
Zoom factor σ0Generation returns former formula can just calculate translation parameters with phase correlation method again.
Invent two, according to improved phase related algorithm scheme, propose that maximum relation degree criterion and peak coordinate judge
The principle of adjacent image position relationship, designs following Sorting algorthm：
(1) Two-dimensional Maximum degree of correlation array is built.The phase related algorithm that every piece image is proposed according to upper section is calculated
With the normalization crosspower spectrum of other images, and the peak value of its inverse Fourier transform is tried to achieve as the degree of correlation between image, so
The N-1 degree of correlation can be arrived per piece image, so that the two-dimensional array for setting up N*N is used for the degree of correlation between storage image.
(2) head image and tail image are determined.For each width to I haven't seen you for ages adjacent with wherein piece image (head image and tail figure
Picture), at most can be adjacent with wherein two images (intermediate image).And two adjacent images, its δ impulse function is in the spatial domain
Very sharp peak value, as maximum relation degree are correspond to, the translation ginseng between two images can be just calculated according to the peak value
Amount.Therefore 2 maximum relation degrees are found out according to the row, column of each width image ordered series of numbers, the 2N maximum degree of correlation can be obtained.And
Only adjacent with piece image for head, tail two images, its maximum relation degree is significantly less than other degrees of correlation, then its corresponding figure
Picture as head image and tail image.If the corresponding horizontal translation amount Δ x ＜ 0 of the maximum relation degree of the image, it is head figure
Picture；Conversely, being tail head portrait；
(3) left-right relation of adjacent two images is determined.Can according to the corresponding translational movement of maximum relation degree along head image
Whole image chain is determined successively, if x0＞ 0, then image should come the right, otherwise come the left side.
Thus obtained sequence image corresponds to the result that camera is shot from left to right.
The present invention overcomes conventional phase correlation method to image limitation of the same size, proposes that the mode of log-polar carrys out table
Show the relation of two images to be matched, rotation, translation, the order models of change of scale are established, while giving maximal correlation
Degree criterion and peak coordinate judge the principle of adjacent image position relationship, for judging the position of sequence image, enhance calculation
The scope of application of method.
Brief description of the drawings
Fig. 1 is phase correlation method flow chart；
Fig. 2, which is one group, has overlapping region Aerial Images；
Fig. 3 is the impulse function that Fig. 3 is detected in the spatial domain using phase correlation method；
Fig. 4 is the algorithm course diagram of sequence image auto-sequencing of taking photo by plane under log-polar；
Fig. 5 is one group of sequence image of taking photo by plane being not in the right order；
Fig. 6 is the image after one group of sequence image sequence of taking photo by plane；
Maximum relation degree between Fig. 7 image sequences；
The corresponding horizontal translation amount of Fig. 8 maximum relation degrees.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, detailed
Carefully describe.Described embodiment is only a part of embodiment of the present invention.
1st, phase correlation method converts by using Fourier's series and obtains the phase information of crosspower spectrum, its ash to image
Spend information dependence smaller, therefore with certain antijamming capability, design flow diagram such as Fig. 1：
Phase correlation method principle can be described as：
If two images A (x, y) and B (x, y), they only exist translation relation, relative level, vertical translation amount before
For x0、y0, then have：
f1(x, y)=f2(x-x0,y-y0)
(1) formula is carried out to be fourier transformed into frequency domain：
Normalized power is composed：
F in above formula*It is F complex conjugate function, from normalized power spectrum it is seen that an exponential function, and to it
Progress is fourier transformed into spatial domain, obtains an impulse function：
The position of the corresponding peak value of impulse function is found out in the spatial domain, and its value reflects the correlation of two images, together
When determine translation parameters (x0,y0).When having noise, complicated perspective between image, even there is moving object, impulse function
Energy can be distributed to from single peak value on other small leaks, but the position where peak-peak still have certain stabilization
Property, it is ensured that the consistency of translational movement.
Fig. 2, which is one group, has the Aerial Images of overlapping region, and size is that Fig. 2 uses phase correlation method for 500 × 750, Fig. 3
The impulse function δ detected in the spatial domain, can represent that right figure is right relative to left figure in Fig. 2 with calculated level amount (124,23)
124pixel is moved, is moved up as 23pixel.
Phase related algorithm under log-polar
Picture size size to be sorted is strict with according to classical phase correlation method described above identical, by mutual
The normalizated correlation coefficient that power spectrum progress inverse Fourier transform is obtained is one and f1(x,y)、f2(x, y) size identical square
Battle array, so there are the confounding issues of left and right translation.Simultaneously for the Aerial Images extracted from sensor, not only wrap
It is also greatly possible to there is a situation where scaling containing translation, rotation.Two width are represented set forth herein the mode of log-polar
The relation of image to be matched, establishes rotation, translation, the order models of change of scale, enhances the scope of application of algorithm.
Principles illustrated：Define the translation parameter x between two images to be spliced0、y0, anglec of rotation α, zoom factor σ, then f1
(x, y) and f2The position relationship of (x, y) is represented by：
f1(x, y)=f2(σxcosα+σysinα-x0,-σxsinα+σycosα-y0)
Fourier transformation is：
Make F1、F2Modulus value be respectively M1、M2, then above formula is turned to：
U=ρ cos β, v=ρ sin β are made, carrying out coordinate to above formula is transformed to：
I.e.：
(1) when two images are only existed, translation, rotation transformation when, now twiddle factor is σ=σ0, then have：
Above formula has been converted to the similar form of simple translation, and rotation can be calculated using the common phase correlation method of upper section
Gyration α=α0。
(2) when deciding anglec of rotation α0When, it can equally obtain：
Log ρ=m, log σ=n are made, then ρ=em, σ=enDai Huike is obtained：
M1(em, α) and=e-2nM2(en-m,α-α0)
I.e.：
M1(m, α)=e-2kM2(m-n,α-α0)
Same above formula turns to the form simply translated, with common phase correlation method can calculate n and rotation because
Sub- σ.
So far the anglec of rotation α calculated0With zoom factor σ0Generation is returned in former formula again can with phase correlation method
To calculate translation parameters.
Improvement phase related algorithm protocol procedures figure such as Fig. 4 can be designed according to above-mentioned principle：
3rd, according to improved phase related algorithm scheme, it is proposed that maximum relation degree criterion and peak coordinate judge adjacent
The principle of image positional relationship, designs following Sorting algorthm：
(1) Two-dimensional Maximum degree of correlation array is built.The phase related algorithm that every piece image is proposed according to upper section is calculated
With the normalization crosspower spectrum of other images, and the peak value of its inverse Fourier transform is tried to achieve as the degree of correlation between image, so
The N-1 degree of correlation can be arrived per piece image, so that the two-dimensional array for setting up N*N is used for the degree of correlation between storage image.
(2) head image and tail image are determined.For each width to I haven't seen you for ages adjacent with wherein piece image (head image and tail figure
Picture), at most can be adjacent with wherein two images (intermediate image).And two adjacent images, its δ impulse function is in the spatial domain
Very sharp peak value, as maximum relation degree are correspond to, the translation ginseng between two images can be just calculated according to the peak value
Amount.Therefore 2 maximum relation degrees are found out according to the row, column of each width image ordered series of numbers, the 2N maximum degree of correlation can be obtained.And
Only adjacent with piece image for head, tail two images, its maximum relation degree is significantly less than other degrees of correlation, then its is corresponding
Image is an image and tail image.If the corresponding horizontal translation amount Δ x ＜ 0 of the maximum relation degree of the image, it is head
Image；Conversely, being tail head portrait；
(3) left-right relation of adjacent two images is determined.Can according to the corresponding translational movement of maximum relation degree along head image
Whole image chain is determined successively, if x0＞ 0, then image should come the right, otherwise come the left side.
Thus obtained sequence image corresponds to the result that camera is shot from left to right.
Fig. 5 is one group of sequence image of taking photo by plane being not in the right order, wherein there is 323 × 370,224 × 256,313 × 302 3 kinds
Obvious translation, rotation are there is between the image of size type, and these images and is scaled, with phase set forth above
Correlation method calculates the degree of correlation between image two-by-two, statistics such as Fig. 7, according to maximum relation degree can preferentially determine head image and
Tail image (is marked) in table with *；The horizontal translation amount between the associated picture of each image maximum relation degree is calculated simultaneously
x0, such as Fig. 8, the left-right relation of two images is can determine that according to the size of translational movement, so as to complete the automatic of experimental image sequence
Sequence sequence, image sequence such as Fig. 6 after sequence.
A kind of present invention improved phase related algorithm scheme according to the feature extractions of Aerial Images, realizes sequence chart
The auto-sequencing of picture.The main complex relationship for being represented to translate, rotate, scale between image by the way of log-polar, is broken through
Limitation of the existing algorithm to picture size, and give the concrete scheme of algorithm realization, passes through maximum relation degree sequencing really
Row head, tail image, coordinate where recycling peak value determine translation parameters, judge that image right position is closed according to given criterion
System, it is to avoid manual intervention.Taken when calculating the degree of correlation in a frequency domain due to algorithm it is larger, can by FFT method.And change
Algorithm after entering can immediately arrive at the translation parameters between image, and the algorithm arrangement can fast and effectively complete aerial image sequence
Auto-sequencing, there is this follow-up key technology to image mosaic very big research to have very big value.
Claims (5)
1. a kind of method for sequence image auto-sequencing of taking photo by plane, it is characterised in that include following steps：
Step A:Translation between image, rotation, scaling relation are represented using the phase correlation method under log-polar, figure is set up
Order models as between；
Step B：Power spectrum between normalized image, the range value of impulse function is used as the relativity measurement between image；
Step C：Designed phase correlation criterion judges that image and peak coordinate judge the position relationships of adjacent coordinates end to end.
2. a kind of method for sequence image auto-sequencing of taking photo by plane according to claim 1, it is characterised in that for Aerial Images
The translation that exists between sequence, rotation, the relation of scaling, rotation, translation, yardstick are established using the mode of log-polar
The order models of conversion, step is as follows：
Translation parameter x between step A, two images to be spliced of definition0、y0, anglec of rotation α, zoom factor σ, then f1(x, y) and f2
The position relationship of (x, y) is represented by：
f1(x, y)=f2(σx cosα+σy sinα-x0,-σx sinα+σy cosα-y0)
Corresponding Fourier transformation is：
Step B, make F1、F2Modulus value be respectively M1、M2, then above formula is turned to：
U=ρ cos β, v=ρ sin β are made, carrying out coordinate to above formula is transformed to：
I.e.：
Step C, only exist when two images, translation, rotation transformation when, now twiddle factor is x=x0, σ=σ0, then have
For above formula, anglec of rotation α=α can be calculated using common phase correlation method0；
Step D, when deciding anglec of rotation α0When, then have：
Log ρ=m, log σ=n are made, then ρ=em, σ=enDai Huike is obtained：
M1(em, α) and=e-2nM2(en-m,α-α0)
I.e.：
M1(m, α)=e-2kM2(m-n,α-α0)
N and twiddle factor σ equally can be calculated with common phase correlation method, the anglec of rotation α calculated0And scaling
Factor sigma0Generation returns former formula can just calculate translation parameters with phase correlation method again；
3) sample matrix, sample are calculatedFeature beWherein
3. a kind of method for sequence image auto-sequencing of taking photo by plane according to claim 1, it is characterised in that related with phase
Method only exists translation to N width and image of the same size is ranked up, and the range value by the use of impulse function is used as the phase between image
Closing property measurement, and translational movement is calculated with coordinate where peak value, so that the position relationship between process decision chart picture.
4. a kind of method for sequence image auto-sequencing of taking photo by plane according to claim 1, it is characterised in that design is a kind of automatic
Sort algorithm：
The degree of correlation that step A, structure Two-dimensional Maximum degree of correlation array are used between storage image；
Step B, head, tail image are relatively determined to the row, column maximum relation degree of each image；
Step C, along head image whole image chain is determined according to the corresponding translational movement of maximum relation degree successively.
5. a kind of method for sequence image auto-sequencing of taking photo by plane according to claim 4, it is characterised in that according to given phase
Position correlation criterion and peak coordinate judge the position relationship of adjacent coordinates, so as to complete the automatic row of whole sequence image of taking photo by plane
Sequence, is comprised the following steps that：
Step A：Build Two-dimensional Maximum degree of correlation array, to every piece image according to it is upper section propose phase related algorithm calculate with
The normalization crosspower spectrum of other images, and the peak value of its inverse Fourier transform is tried to achieve as the degree of correlation between image, it is so every
Piece image can arrive the N-1 degree of correlation, so that the two-dimensional array for setting up N*N is used for the degree of correlation between storage image；
Step B：It is determined that head image and tail image；For each width to I haven't seen you for ages adjacent with wherein piece image (head image and tail figure
Picture), at most can be adjacent with wherein two images (intermediate image)；And two adjacent images, its δ impulse function is in the spatial domain
Very sharp peak value, as maximum relation degree are correspond to, the translation ginseng between two images can be just calculated according to the peak value
Amount；Therefore 2 maximum relation degrees are found out according to the row, column of each width image ordered series of numbers, the 2N maximum degree of correlation can be obtained；And
Only adjacent with piece image for head, tail two images, its maximum relation degree is significantly less than other degrees of correlation, then its corresponding figure
Picture as head image and tail image；If the corresponding horizontal translation amount Δ x ＜ 0 of the maximum relation degree of the image, it is head figure
Picture；Conversely, being tail head portrait；
Step C：Determine the left-right relation of adjacent two images；Can be according to according to the corresponding translational movement of maximum relation degree along head image
Secondary determination whole image chain, if x0＞ 0, then image should come the right, otherwise come the left side.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710138301.4A CN106952227A (en) | 2017-03-09 | 2017-03-09 | A kind of method for sequence image auto-sequencing of taking photo by plane |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710138301.4A CN106952227A (en) | 2017-03-09 | 2017-03-09 | A kind of method for sequence image auto-sequencing of taking photo by plane |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106952227A true CN106952227A (en) | 2017-07-14 |
Family
ID=59468200
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710138301.4A Pending CN106952227A (en) | 2017-03-09 | 2017-03-09 | A kind of method for sequence image auto-sequencing of taking photo by plane |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106952227A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108335314A (en) * | 2018-02-28 | 2018-07-27 | 百度在线网络技术（北京）有限公司 | Method and apparatus for generating information |
| CN110361560A (en) * | 2019-06-25 | 2019-10-22 | 中电科技（合肥）博微信息发展有限责任公司 | A kind of shipping sail speed measurement method, device, terminal device and computer readable storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102938143A (en) * | 2012-09-28 | 2013-02-20 | 河海大学 | Image sequence ordering method |
| CN103606139A (en) * | 2013-09-09 | 2014-02-26 | 上海大学 | Sonar image splicing method |
| CN104866823A (en) * | 2015-05-11 | 2015-08-26 | 重庆邮电大学 | Vehicle detection and tracking method based on monocular vision |
| CN105359184A (en) * | 2013-03-14 | 2016-02-24 | 微软技术许可有限责任公司 | Image capture and ordering |
| CN105931185A (en) * | 2016-04-20 | 2016-09-07 | 中国矿业大学 | Automatic splicing method of multiple view angle image |
-
2017
- 2017-03-09 CN CN201710138301.4A patent/CN106952227A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102938143A (en) * | 2012-09-28 | 2013-02-20 | 河海大学 | Image sequence ordering method |
| CN105359184A (en) * | 2013-03-14 | 2016-02-24 | 微软技术许可有限责任公司 | Image capture and ordering |
| CN103606139A (en) * | 2013-09-09 | 2014-02-26 | 上海大学 | Sonar image splicing method |
| CN104866823A (en) * | 2015-05-11 | 2015-08-26 | 重庆邮电大学 | Vehicle detection and tracking method based on monocular vision |
| CN105931185A (en) * | 2016-04-20 | 2016-09-07 | 中国矿业大学 | Automatic splicing method of multiple view angle image |
Non-Patent Citations (3)
| Title |
|---|
| ALFONSO ALBA等: "Phase correlation based image alignment with subpixel accuracy", 《LECTURE NOTES IN COMPUTER SCIENCE》 * |
| 赵萌萌: "基于特征点的图像拼接算法", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 陈丽莉等: "一种有效的序列图像自动拼接方法", 《光电子激光》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108335314A (en) * | 2018-02-28 | 2018-07-27 | 百度在线网络技术（北京）有限公司 | Method and apparatus for generating information |
| CN110361560A (en) * | 2019-06-25 | 2019-10-22 | 中电科技（合肥）博微信息发展有限责任公司 | A kind of shipping sail speed measurement method, device, terminal device and computer readable storage medium |
| CN110361560B (en) * | 2019-06-25 | 2021-10-26 | 中电科技（合肥）博微信息发展有限责任公司 | Ship navigation speed measuring method and device, terminal equipment and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| He et al. | Epipolar transformers | |
| CN106780543B (en) | A kind of double frame estimating depths and movement technique based on convolutional neural networks | |
| Pirsiavash et al. | Assessing the quality of actions | |
| CN106469299B (en) | A kind of vehicle search method and device | |
| Svarm et al. | Accurate localization and pose estimation for large 3d models | |
| Sheikh et al. | Exploring the space of a human action | |
| Torii et al. | From google street view to 3d city models | |
| CN103679674B (en) | Method and system for splicing images of unmanned aircrafts in real time | |
| Baboud et al. | Automatic photo-to-terrain alignment for the annotation of mountain pictures | |
| CN103530881B (en) | Be applicable to the Outdoor Augmented Reality no marks point Tracing Registration method of mobile terminal | |
| CN107533370B (en) | Image processing apparatus, image processing method, and program | |
| WO2017133009A1 (en) | Method for positioning human joint using depth image of convolutional neural network | |
| CN104599258B (en) | A kind of image split-joint method based on anisotropic character descriptor | |
| CN105957007A (en) | Image stitching method based on characteristic point plane similarity | |
| CN106295526B (en) | The method and device of Car image matching | |
| CN106228507A (en) | A kind of depth image processing method based on light field | |
| CN105160323B (en) | A kind of gesture identification method | |
| US10755433B2 (en) | Method and system for scanning an object using an RGB-D sensor | |
| CN105427298A (en) | Remote sensing image registration method based on anisotropic gradient dimension space | |
| CN105761234A (en) | Structure sparse representation-based remote sensing image fusion method | |
| JP2007538318A5 (en) | ||
| CN107833250A (en) | Semantic space map constructing method and device | |
| CN107766864B (en) | Method and device for extracting features and method and device for object recognition | |
| Zhang et al. | Retargeting semantically-rich photos | |
| CN106952227A (en) | A kind of method for sequence image auto-sequencing of taking photo by plane |
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: 20170714 |
|
| WD01 | Invention patent application deemed withdrawn after publication |