CN105160684B - A kind of on-line automatic matching process for remotely sensing image geometric correction - Google Patents
A kind of on-line automatic matching process for remotely sensing image geometric correction Download PDFInfo
- Publication number
- CN105160684B CN105160684B CN201510634043.XA CN201510634043A CN105160684B CN 105160684 B CN105160684 B CN 105160684B CN 201510634043 A CN201510634043 A CN 201510634043A CN 105160684 B CN105160684 B CN 105160684B
- Authority
- CN
- China
- Prior art keywords
- image
- matched
- matching
- image blocks
- point
- 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.)
- Expired - Fee Related
Links
Classifications
-
- 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
- G06T2207/10032—Satellite or aerial image; Remote sensing
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
Landscapes
- Image Processing (AREA)
Abstract
A kind of on-line automatic matching process for remotely sensing image geometric correction, is related to Remote Sensing Images Matching technical field.It is advantageous that making full use of the priori geological information of remote sensing image, by free or low cost network video map resource, equally distributed control point is obtained to high efficient and reliable.Using the initial imaging model of remote sensing image, image to be matched is divided into several regions by the number at the control point acquired as needed, and using a certain size image blocks as processing unit in each region；The approximate range for referring to image blocks is determined according to the range of image unit to be matched and initial imaging model, is then and resolution ratio similar in image unit to be matched by network downloading with reference to image blocks and by its resampling；SIFT matching, and excluding gross error point are carried out to image unit to be matched and with reference to image blocks, optimal matching points are finally selected from remaining match point and Least squares matching is carried out to it, correct its position coordinate.
Description
Technical field
The present invention relates to a kind of practical remote sensing image automatic matching methods, can using network Aeronautics and Astronautics image
Figure carries out on-line automatic matching to remote sensing image.It can be applied to the fields such as remote sensing, photogrammetric, mapping, image procossing.
Background technique
Image Auto-matching and geometric correction are the key links in Photogrammetry and Remote Sensing task, they are that image melts
The basis of advanced application such as close, inlay, changing detection, map rejuvenation.Although in the past in decades to image Auto-matching
Research is very more, but the Auto-matching of remote sensing image is still very challenging.Practical automatic matching method is answered
This efficiency, robustness, in terms of the performance that has all had, but since remote sensing image data amount is big, scene is big, obtains item
The features such as part is changeable, geometry deformation is complicated, existing method are difficult to take into account this several respect.In addition, the preparation with reference to image is also
One difficult point of remote sensing image Auto-matching and geometric correction, especially high-resolution generally require very high with reference to the acquisition of image
Cost.
In view of the limitation of existing automatic matching method, a kind of on-line automatic match party for remotely sensing image geometric correction
Method has important practical value.
Summary of the invention
It is an object of the invention to solve the deficiencies in the prior art, propose a kind of for the fast of remotely sensing image geometric correction
Speed, steady, accurately on-line automatic matching process, this method are capable of handling the remote sensing image of arbitrary size, utilize network video
Map is automatically matched to a certain number of, well-distributed high-precision control point in a relatively short period of time, is directly used in remote sensing shadow
The geometric accurate correction of picture.The advantage of this method essentially consists in the priori geological information for making full use of remote sensing image, by free or
The network video map resource of low cost, obtains to high efficient and reliable equally distributed control point.
To solve the above problems, the present invention provides a kind of on-line automatic match parties for remotely sensing image geometric correction
Method, the method comprising the steps of:
Remote sensing image to be matched is evenly dividing as several regions by the number at the control point S1. acquired as needed；
S2. if all imagery zones are processed, entire Image Matching process is completed；Otherwise, it starts to process next
A imagery zone；
S3. current image region is divided into several image units by 256 pixels × 256 pixels size；
S4. if all image units in current image region are processed, mark the imagery zone that place is completed
Reason, and it is transferred to step S2；Otherwise, next image unit is started to process；
S5: it is calculated according to the range of current image unit to be matched and initial imaging model and refers to shadow in network video map
As the approximate range of block, then by network download it is corresponding with reference to image blocks and by its resampling be and image unit to be matched
Similar resolution ratio；
S6: being matched using SIFT matching operator to image unit to be matched and with reference to image blocks, and excluding gross error point,
If the step obtains 4 or more match points, step S7 is carried out, is otherwise transferred to step S4；
S7. optimal matching points are selected from the match point that step S6 is obtained and Least squares matching is carried out to it, are corrected
This is added in matching result match point by the position coordinate of SIFT feature.
Wherein, step S5 further comprises:
S5.1 is according to the resolution ratio of image unit to be matched calculating network video map closest to level of zoom；
S5.2 calculates the corresponding width and height for referring to image blocks；
S5.3 calculates the latitude and longitude coordinates of the corresponding central point with reference to image blocks；
S5.4 sends static map service request and downloads corresponding with reference to image blocks；
S5.5 by the reference image blocks resampling of downloading be and resolution ratio similar in image unit to be matched.
Wherein, step S6 further comprises:
S6.1 carries out SIFT matching to image unit to be matched and with reference to image blocks；
S6.2 passes through dimensional constraints excluding gross error point；
S6.3 utilizes rotation angle restriction excluding gross error；
S6.4 constrains excluding gross error using RANSAC estimation similarity transformation；
S6.5 constrains excluding gross error point using affine transformation.
Detailed description of the invention
Fig. 1 is the on-line automatic matching process flow chart according to one embodiment of the present invention；
Fig. 2 is imagery zone and image unit schematic diagram in the method for the present invention；
Fig. 3 is according to acquisition network video map reference point shadow in the on-line automatic matching process of one embodiment of the present invention
As block flow chart；
Fig. 4 be according in the on-line automatic matching process of one embodiment of the present invention to image unit to be matched and reference
Image Block- matching flow chart；
Specific embodiment
On-line automatic matching process proposed by the present invention, is described with reference to the accompanying drawings as follows.
As shown in Figure 1, according to one embodiment of the present invention automatic matching method comprising steps of
Remote sensing image to be matched is evenly dividing as several regions by the number at the control point S1. acquired as needed, and
Each imagery zone is labeled as untreated state, imagery zone schematic diagram is shown in Fig. 2；
S2. it if all imagery zones are processed, completes and terminates entire Image Matching process；Otherwise, beginning
Manage next imagery zone；
S3. current image region is divided into several image units by 256 pixels × 256 pixels size, and will be each
Image unit is denoted as untreated state, and image unit schematic diagram is shown in Fig. 2；
S4. if all image units in current image region are processed, mark the imagery zone that place is completed
Reason, and it is transferred to step S2；Otherwise, next image unit is started to process；
S5. network video map is calculated according to the range of current image unit to be matched and initial imaging model (at present may be used
Network video map include Google satellite image map, Bing Aerial image map, MapQuest satellite photomap and
Mapbox satellite photomap) in reference to image blocks approximate range, then by network download it is corresponding refer to image blocks simultaneously
It is and resolution ratio similar in image unit to be matched by its resampling；
S6. it is matched using SIFT matching operator to image unit to be matched and with reference to image blocks, and excluding gross error point,
If the step obtains 4 or more match points, step S7 is carried out, is otherwise transferred to step S4；
S7. from the match point that step S6 is obtained, the maximum a pair of local contrast is chosen, and SIFT is matched
Local geometric transformation model carries out Least squares matching to match point to this as initial value, and the point for SIFT feature of refining is sat
This, is finally added in matching result match point, conditional equation formula (1) table of the Least squares matching to certain point by mark
Show,
k1Is(xs, ys)+k2-Ir(xr, yr)=0 (1)
Wherein xs, ysFor image unit pixel coordinate to be matched, xr, yrTo refer to image blocks pixel coordinate,
xr=a0+a1xs+a2ys, yr=b0+b1xs+b2ys,
a0, a1, a2, b0, b1, b2For 6 geometric transformation parameters, k1, k2For 2 radiation transformation parameters,
Is(xs, ys) and Ir(xr, yr) be respectively image unit to be matched and with reference to image blocks gray value,
It treats in 11 pixels around match point × 11 pixels region each point and establishes error equation according to formula (1), then use
Levenberg-Marquardt algorithm carries out optimization, optimal geometric transformation parameter is obtained, so that it is determined that after refining
With a coordinate.
After image to be matched is divided into image unit, need according to the range of current image unit to be matched and initially at
As model calculates the approximate range in network video map with reference to image blocks.Specifically, as shown in figure 3, step S5 is further wrapped
It includes:
S5.1 is according to the resolution ratio of image unit to be matched calculating network video map closest to level of zoom, zoom-level
Not according to the latitude and longitude coordinates and resolution ratio located where image unit to be matched using formula (2) calculating, at image unit place
Longitude λ and latitude φ is calculated by the initial imaging model of image to be matched,
R in formulaearthRice is earth radius, and approximation is 6378137 meters,
GSD is the resolution ratio at image unit place,
N is level of zoom,
[] indicates to take the operation of closest integer；
S5.2 calculates the corresponding width width and height height for referring to image blocks: given level of zoom n, network video
Map image coordinate x, y and longitude λ, the conversion method such as formula (3) of latitude φ and formula (4) are shown,
The latitude and longitude coordinates for being utilized respectively 4 vertex of image unit to be matched calculate corresponding 4 picture points by formula (3)
Coordinate, then seeks the minimum circumscribed rectangle of this 4 points again, and the width and height of the rectangle are the width with reference to image blocks
Width and height height；
S5.3 calculates the latitude and longitude coordinates λ of the corresponding central point with reference to image blocksrc, φrc, according to obtained in S5.2 most
Small boundary rectangle calculates its center point image coordinate, then substitutes into formula (4) and calculates corresponding latitude and longitude coordinates λrc, φrc；
S5.4 sends static map service request and downloads and corresponds to reference to image blocks, and static map service request is provided with unified
The mode of source finger URL (URL) issues, and the static map service request format for commonly using network video map is as follows:
(1) Google satellite image map:
Https: //maps.googleapis.com/maps/api/staticmap? maptype=satellite&
Zoom={ zoomLevel } &ce nter={ lat }, { lon } &size={ width } x { height } &key=
{googleKey}
(2) Bing Aerial image map:
Http:// dev.virtualearth.net/RES T/v1/Imagery/Map/Aerial/ { lat }, { lon }/
{ zoomLevel }? mapSize={ width }, { height } &key={ BingMapsKey }
(3) MapQuest satellite photomap:
Http:// www.mapquestapi.com/staticmap/v4/getmap? type=sat&zoom=
{ zoomLevel } ¢er={ lat }, { lon } &size={ width }, { height } &key={ mapquestKey }
(4) Mapbox satellite photomap:
Http:// api.tiles.mapbox.com/v4/mapbox.satellite/ { lon }, { lat },
{ zoomLevel }/{ width } x { height } .p ng? access_token={ mapboxKey }
In above each URL, the content in " { } " is filled according to the calculated result of step S5.1, S5.2 and S5.3, specifically,
ZoomLevel is to be calculated in step S5.1 closest to level of zoom n,
Width, height are the width width and height height for the reference image blocks being calculated in step S5.2,
Lon, lat are the central point longitude λ for the reference image blocks being calculated in step S5.3rcWith latitude φrc,
GoogleKey, BingMapsKey, mapquestKey, mapboxKey are respectively the api of each Map Service of Network
Key can be applied obtaining on corresponding website；
S5.5 by the reference image blocks resampling of downloading be and resolution ratio similar in image unit to be matched.
Downloading obtains after referring to image blocks accordingly, needs using SIFT matching operator to image unit to be matched and reference
Image blocks are matched, and excluding gross error point.Specifically, as shown in figure 4, step S6 further comprises:
S6.1 from image unit to be matched and with reference to extraction SIFT feature in image blocks and calculates corresponding SIFT respectively
Description vectors, using the Euclidean distance of SIFT description vectors as the distance metric of two SIFT features, to image list to be matched
Each SIFT feature in member judges following two condition, meet alternative one or simultaneously meet 2 conditions i.e. as
With candidate point:
(1) this into reference image blocks the ratio of each SIFT feature minimum range and time small distance less than 0.75；
(2) with reference in image blocks with the point apart from the smallest characteristic point other SIFT features into image unit to be matched
Distance be not less than the distance of the point；
Its corresponding match point be with reference in image blocks with it apart from nearest SIFT feature；
S6.2 checks the SIFT scale coordinate ratio T of each candidate point and its match pointσIf 0.8≤Tσ≤ 1.25, then it protects
It stays this to candidate point, otherwise rejects this to candidate point；
S6.3 calculates the SIFT principal direction differential seat angle of each candidate point and its match point, establishes intensity histogram with 10 degree for interval
Figure, using the angle where histogram peak as image unit to be matched and with reference to the rotation angle θ between image blockspeak, then
Check the SIFT principal direction differential seat angle Δ θ of each candidate point and its match point, if | Δ θ-θpeak|≤15 °, then retain this to time
Otherwise reconnaissance rejects this to candidate point；
S6.4 estimates the similarity transformation relationship (5) between remaining candidate matches point using RANSAC algorithm, while rejecting not
Meet the rough error point of the variation relation,
Wherein xs, ysFor image unit pixel coordinate to be matched, xr, yrTo refer to image blocks pixel coordinate, s and θ are similar
The scale and rotation angle parameter of transformation, tx, tyRespectively translation parameters of the similarity transformation in the direction x and the direction y；
S6.5 calculates affine Transform Model to candidate point remaining in step S6.4,
Wherein xs, ysFor image unit pixel coordinate to be matched, xr, yrTo refer to image blocks pixel coordinate, a0, a1, a2, b0,
b1, b2For 6 parameters of affine transformation；
Each pair of matching candidate point is checked using obtained affine Transform Model, if maximum residul difference is greater than 1 pixel, is picked
Except this to match point after, recalculate affine Transform Model, and residual test is carried out to remaining candidate point；If maximum residul difference is not
Greater than 1 pixel, then export remaining matching double points, end step S6.
Claims (1)
1. a kind of on-line automatic matching process for remotely sensing image geometric correction, it is characterized in that: include the following steps,
Remote sensing image to be matched is evenly dividing as several regions by the number at the control point that step S1. is acquired as needed；
If all imagery zones of step S2. are processed, entire Image Matching process is completed；Otherwise, it starts to process next
A imagery zone；
Current image region is divided into several image units by 256 pixels × 256 pixels size by step S3.；
If all image units in step S4. current image region are processed, mark the imagery zone that place is completed
Reason, and it is transferred to step S2；Otherwise, next image unit is started to process；
Step S5. calculates free or low cost network according to the range and initial imaging model of current image unit to be matched
The approximate range of image blocks is referred in photomap, and corresponding reference image blocks are then downloaded by network and is by its resampling
With resolution ratio similar in image unit to be matched；
Step S6. is matched using SIFT matching operator to image unit to be matched and with reference to image blocks, and excluding gross error
Point carries out step S7 if the step obtains 4 or more match points, is otherwise transferred to step S4；
Step S7. selects optimal matching points from the match point that step S6 is obtained and carries out Least squares matching to it, amendment
This is added in matching result match point by the position coordinate of SIFT feature；
Step S5 further comprises: S5.1 is according to the resolution ratio of image unit to be matched calculating network video map closest to contracting
Put rank, level of zoom utilizes formula (2) calculating, image according to the latitude and longitude coordinates and resolution ratio located where image unit to be matched
Longitude λ and latitude Φ at where unit are calculated by the initial imaging model of image to be matched,
R in formulaearthRice is earth radius, and approximation is 6378137 meters, and GSD is the resolution ratio at image unit place, and n is
Level of zoom, [] indicate to take the operation of closest integer；
Step S5.2 calculates the corresponding width width and height height for referring to image blocks: given level of zoom n, network video
Map image coordinate x, y and longitude λ, the conversion method such as formula (3) of latitude φ and formula (4) are shown,
4 picture point coordinates on 4 vertex are calculated by formula (3) using the latitude and longitude coordinates on 4 vertex of image unit to be matched,
Then the minimum circumscribed rectangle of 4 picture points is sought again, and the width and height of the minimum circumscribed rectangle are the reference
The width width and the height height of image blocks；
Step S5.3 calculates the longitude coordinate λ of the corresponding central point with reference to image blocksrcWith latitude coordinate φrc, obtained according in S5.2
To the minimum circumscribed rectangle calculate the central point image coordinate with reference to image blocks, then substitute into formula (4) calculate correspond to
Longitude coordinate λrcWith latitude coordinate φrc；
Step S5.4 sends static map service request and downloads and corresponds to reference to image blocks, and static map service request is provided with unified
The mode of source finger URL (URL) issues；
Step S5.5 by the reference image blocks resampling of downloading be and resolution ratio similar in image unit to be matched.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510634043.XA CN105160684B (en) | 2015-09-30 | 2015-09-30 | A kind of on-line automatic matching process for remotely sensing image geometric correction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510634043.XA CN105160684B (en) | 2015-09-30 | 2015-09-30 | A kind of on-line automatic matching process for remotely sensing image geometric correction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105160684A CN105160684A (en) | 2015-12-16 |
| CN105160684B true CN105160684B (en) | 2019-01-18 |
Family
ID=54801526
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510634043.XA Expired - Fee Related CN105160684B (en) | 2015-09-30 | 2015-09-30 | A kind of on-line automatic matching process for remotely sensing image geometric correction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105160684B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106408028B (en) * | 2016-09-26 | 2020-04-14 | 珠海市测绘院 | Urban and rural planning inspection surveying and mapping data processing method |
| CN106595598B (en) * | 2016-12-21 | 2019-03-19 | 上海航天控制技术研究所 | A kind of first optical remote sensing imaging method in permanent ground of wide visual field |
| CN108428220B (en) * | 2018-03-05 | 2020-12-01 | 武汉大学 | Automatic geometric correction method for ocean island reef area of remote sensing image of geostationary orbit satellite sequence |
| CN108492711A (en) * | 2018-04-08 | 2018-09-04 | 黑龙江工业学院 | A kind of drawing electronic map method and device |
| CN109242894B (en) * | 2018-08-06 | 2021-04-09 | 广州视源电子科技股份有限公司 | Image alignment method and system based on mobile least square method |
| CN109508674B (en) * | 2018-11-13 | 2021-08-13 | 佳木斯大学 | Airborne downward-looking heterogeneous image matching method based on region division |
| CN114004770B (en) * | 2022-01-04 | 2022-04-26 | 成都国星宇航科技有限公司 | Method and device for accurately correcting satellite space-time diagram and storage medium |
| CN116521927B (en) * | 2023-06-30 | 2024-02-13 | 成都智遥云图信息技术有限公司 | Remote sensing image matching method and system based on network map tiles |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103337052A (en) * | 2013-04-17 | 2013-10-02 | 国家测绘地理信息局卫星测绘应用中心 | Automatic geometric correction method for wide remote-sensing images |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| FR2879791B1 (en) * | 2004-12-16 | 2007-03-16 | Cnes Epic | METHOD FOR PROCESSING IMAGES USING AUTOMATIC GEOREFERENCING OF IMAGES FROM A COUPLE OF IMAGES TAKEN IN THE SAME FOCAL PLAN |
| CN102663680B (en) * | 2012-03-06 | 2017-04-05 | 中国科学院对地观测与数字地球科学中心 | Geometric image correction method based on region feature |
| CN103218783B (en) * | 2013-04-17 | 2016-05-25 | 国家测绘地理信息局卫星测绘应用中心 | Satellite remote sensing images fast geometric correcting method based on control point image database |
-
2015
- 2015-09-30 CN CN201510634043.XA patent/CN105160684B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103337052A (en) * | 2013-04-17 | 2013-10-02 | 国家测绘地理信息局卫星测绘应用中心 | Automatic geometric correction method for wide remote-sensing images |
Non-Patent Citations (1)
| Title |
|---|
| An Automated Method for Geo-Referencing Satellite Images Using Google Tile Scheme;Khalil Al-Joburi;《World Applied Sciences Journal》;20141231;第32卷(第2期);第267-277页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105160684A (en) | 2015-12-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105160684B (en) | A kind of on-line automatic matching process for remotely sensing image geometric correction | |
| CN107316325B (en) | Airborne laser point cloud and image registration fusion method based on image registration | |
| CN105761242B (en) | Blind person walking positioning method based on computer binocular vision and inertial measurement | |
| CN109029444A (en) | One kind is based on images match and sterically defined indoor navigation system and air navigation aid | |
| CN109523585B (en) | Multisource remote sensing image feature matching method based on direction phase consistency | |
| CN106709944B (en) | Satellite remote sensing image registration method | |
| KR20190026452A (en) | A method of automatic geometric correction of digital elevation model made from satellite images and provided rpc | |
| CN112750203B (en) | Model reconstruction method, device, equipment and storage medium | |
| CN112419374A (en) | Unmanned aerial vehicle positioning method based on image registration | |
| CN110660098A (en) | Positioning method and device based on monocular vision | |
| CN112200203A (en) | Matching method of weak correlation speckle images in oblique field of view | |
| Bruno et al. | Accuracy assessment of 3d models generated from google street view imagery | |
| Kim et al. | Investigating applicability of unmanned aerial vehicle to the tidal flat zone | |
| CN111693025A (en) | Remote sensing image data generation method, system and equipment | |
| CN113984039A (en) | Method, device and system for correcting motion trail and storage medium | |
| Zhang et al. | Automatic fusion of hyperspectral images and laser scans using feature points | |
| Duan et al. | A combined image matching method for Chinese optical satellite imagery | |
| KR101711575B1 (en) | Method for Detecting of the Unified Control Points for RPC Adjustment of Satellite Image | |
| Feng et al. | Mountainous remote sensing images registration based on improved optical flow estimation | |
| Tatar et al. | Quasi-epipolar resampling of high resolution satellite stereo imagery for semi global matching | |
| US20210110195A1 (en) | Fast determinant of hessian filtering for image tiepoint candidate area assessment | |
| CN113932793A (en) | Three-dimensional coordinate positioning method and device, electronic equipment and storage medium | |
| CN109696656A (en) | Localization method and its system based on phase focusing | |
| Alsubaie et al. | The feasibility of 3D point cloud generation from smartphones | |
| KR101775124B1 (en) | System and method for automatic satellite image processing for improvement of location accuracy |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190118 Termination date: 20190930 |