CN109271995A - A kind of high-precision image matching method and system - Google Patents
A kind of high-precision image matching method and system Download PDFInfo
- Publication number
- CN109271995A CN109271995A CN201710584412.8A CN201710584412A CN109271995A CN 109271995 A CN109271995 A CN 109271995A CN 201710584412 A CN201710584412 A CN 201710584412A CN 109271995 A CN109271995 A CN 109271995A
- Authority
- CN
- China
- Prior art keywords
- point
- match
- image
- match point
- interest label
- 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 19
- 238000000605 extraction Methods 0.000 claims abstract description 16
- 238000009826 distribution Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 abstract description 4
- 238000001514 detection method Methods 0.000 description 4
- 238000012706 support-vector machine Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012887 quadratic function Methods 0.000 description 2
- 230000000717 retained effect Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Processing Or Creating Images (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of high-precision image matching method and systems, are related to data processing field, in particular to map datum processing and fabricating.High-precision image matching method disclosed by the invention comprises determining that target image and compares the point of interest label region in image；Feature extraction is carried out to point of interest label region；The feature extracted is matched, match point is obtained；Images match is carried out according to the distributing position of match point and/or match point quantity.The present invention is designed by algorithm optimization, improves images match precision, so as to remove redundant image, improves the producing efficiency of map especially high-precision electronic map.
Description
Technical field
The present invention relates to data processing fields, in particular to map datum processing and fabricating.
Background technique
Electronic map (English: Electronic map), i.e. numerical map are using computer technology, in a digital manner
Storage and the map consulted.Electronic map data generally comprises road, point of interest, road network etc..Electronic map be cartography and
One system of application, is to control map generated by electronic computer, is the screen map based on the digital graphics, is
Visually scheme on the spot." visualizing on the computer screen " is the basic feature of electronic map.
The characteristics of electronic map, there is following 6:
1. display can be accessed quickly.
2. animation may be implemented.
3. can be by map elements Layering manifestation.
4. enabling user have sense on the spot in person map three-dimensional, mobilism using virtual reality technology.
5. electronic map can be transferred to elsewhere using data transmission technology.
6. the automatic measurement of the length on figure, angle, area etc. may be implemented.
Electronic map data generally passes through field data collector and is acquired, such as acquisition target is put down,
The modes such as take pictures, photograph acquire, and then carry out data mart modeling by interior industry data mart modeling personnel.In order to guarantee collected target
It is accurate and complete, when generally acquiring on the spot, multiple pictures can be shot to acquisition target.
When carrying out map datum processing, because acquisition target has plurality of pictures, if not removing the image of repeated and redundant, need
Interior industry data mart modeling personnel handle every image, it will cause the huge waste of interior industry resource.And remove repeated and redundant
Image just need a kind of high-precision image matching method can judge containing it is identical acquisition target multiimage.Future
Intelligent driving is highly dependent on high-precision electronic map, and the production of high-precision electronic map also needs high-precision image
Method of completing the square improves high-precision electronic cartography efficiency to remove redundant image.
Summary of the invention
The object of the present invention is to provide a kind of high-precision image matching method and systems, to improve the essence of images match
Degree improves the producing efficiency of map, especially high-precision electronic map to remove redundant image.
The present invention provides a kind of high-precision image matching method, comprising:
It determines target image and compares the point of interest label region in image；
Feature extraction is carried out to point of interest label region；
The feature extracted is matched, match point is obtained；
Images match is carried out according to the distributing position of match point and/or match point quantity.
The present invention also provides a kind of high-precision image matching systems, comprising:
Point of interest label area determination unit, for determining target image and comparing the point of interest label region in image；
Feature extraction unit, for carrying out feature extraction to point of interest label region；
Characteristic matching unit obtains match point for matching the feature extracted；
Image matching unit, for carrying out images match according to the distributing position and/or match point quantity of match point.
The present invention is designed by algorithm optimization, improves images match precision, so as to remove redundant image, improves ground
The producing efficiency of figure especially high-precision electronic map.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, required use in being described below to embodiment
Attached drawing be briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for this
For the those of ordinary skill in field, without creative efforts, it can also be obtained according to these attached drawings other
Attached drawing.
Fig. 1 is the flow diagram of high-precision image matching method provided in an embodiment of the present invention；
Fig. 2 is the structural schematic diagram of high-precision image matching system provided in an embodiment of the present invention.
Specific embodiment
With reference to the attached drawing in the embodiment of the present invention, technical solution in the embodiment of the present invention carries out clear, complete
Ground description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.Based on this
The embodiment of invention, every other implementation obtained by those of ordinary skill in the art without making creative efforts
Example, belongs to protection scope of the present invention.
The embodiment of the present invention is described in further detail below in conjunction with attached drawing.
In map datum process, to freshly harvested image (target image), in the database images data of magnanimity
In (compare image) be compared and match, the image of successful match, then be judged as YES include identical acquisition target repetition
Image.The comparison can be target image with compare image and be compared two-by-two, be also possible to target image and compare image
Concentration is disposably compared.
Embodiment one
The embodiment of the present invention provides a kind of high-precision image matching method, and specific implementation process is as shown in Figure 1, include following
Step:
101, it determines target image and compares the point of interest label region in image, point of interest (POI) is in GIS-Geographic Information System
A term, geographic object a little can be abstracted as by referring to all, more especially be lived closely related geography with people
Entity, such as school, bank, restaurant, gas station, hospital, supermarket, point of interest label region, which refers in image, has point of interest letter
The region of breath, if point of interest is a restaurant, point of interest label region is the region for having restaurant name in image.It specifically includes:
Firstly, using MSER algorithm (Maximally Stable Extremal Regions maximum stable extremal region) in image
Point of interest label region carry out coarse positioning, recycle HOG feature (Histogram of Oriented Gridients feature
Detection algorithm) and SVM (Support Vector Machine support vector machines) to detection target be accurately positioned.
102, feature extraction is carried out to point of interest label region determined by step 101.Specifically, Root- can be used
Sift algorithm (Root-Scale-invariant feature transform, scale invariant feature conversion) carries out feature and mentions
It takes.Specifically include: the picture construction scale space compared firstly, for needs is exactly the Gaussian difference bundling using different scale
Product core and image carry out convolution algorithm.Secondly, being compared using each sampled point with all the points around it, Gaussian difference is found
Divide the extreme point in space, and removes Gaussian convolution space local curvature very asymmetric image by being fitted three-dimensional quadratic function
Vegetarian refreshments.Again, it is each key point assigned direction parameter using the gradient direction distribution characteristic of key point field pixel, makes operator
Have 128 dimension directioin parameters of rotational invariance, here it is the Feature Descriptors of key point, are normalized and are taken to description
Root obtains final Expressive Features.
103, the feature for being extracted step 102 matches, and obtains match point.It will be in the image that feature be extracted
Sub- matching is described in each scale image, can be judged as successful match, acquisition when the distance of description is less than threshold value
With point.
104, images match is carried out according to the distributing position of match point and/or match point quantity.It include: according to match point
Distributing position and/or match point quantity judge whether image matches, if match point spatial distribution is consistent and match point quantity reaches
Preset value then confirms images match success.Match point quantity includes following one kind: all match point quantity, or removal away from
From the match point quantity after the match point in the pre-determined distance of point of interest label zone boundary；Or the match point that removal line intersects
Match point quantity afterwards；Or remove that match point and line in the pre-determined distance of point of interest label zone boundary intersect
With the match point quantity after point.Match point quantity is preferably the matching removed in the pre-determined distance of point of interest label zone boundary
Match point quantity after the match point that point and line intersect.Match point at the edge in point of interest label region has little significance,
Therefore it can remove it, the matching line is interdigital: the line that the match point in two images connects is had intersection.
Matching line is generally parallel line, can if the matching line of certain match point intersects with the matching line of other match points
With the match point removal for intersecting line.
When the target image and comparing image successful match, then judge that image repeats, deletes multiimage to only retaining
One image.Specifically, can only retain the most image of feature quantity when being deleted and being retained, alternatively, retaining
The newest image of shooting time.Feature means that image is more clear, and the electronic map precision processed is also higher therefore excellent
Select the image that keeping characteristics are most.
High-precision image matching method provided in an embodiment of the present invention is suitable for electronic map, especially high precision electro
The production of sub- map.The embodiment of the present invention is designed by algorithm optimization, images match precision is improved, so as to remove redundancy
Image improves the producing efficiency of map especially high-precision electronic map.
Embodiment two
The embodiment of the present invention provides a kind of high-precision image matching system, as shown in Figure 2, comprising:
Point of interest label area determination unit 201, for determining target image and comparing the point of interest label region in image.It is emerging
Interesting point (POI) is a term in GIS-Geographic Information System, and geographic object a little can be abstracted as by referring to all, more especially
It lives closely related geographical entity, such as school, bank, restaurant, gas station, hospital, supermarket with people, point of interest label area
Domain refers to the region in image with interest point information, and if point of interest is a restaurant, point of interest label region is that have in image
The region of restaurant name.It specifically includes: firstly, (Maximally Stable Extremal Regions is most using MSER algorithm
Big stable extremal region) coarse positioning is carried out to the point of interest label region in image, recycle HOG feature (Histogram of
Oriented Gridients feature detection algorithm) and SVM (Support Vector Machine support vector machines) to detection
Target is accurately positioned.
Feature extraction unit 202, for point of interest label region determined by point of interest label area determination unit 201
Carry out feature extraction.Specifically, Root-sift algorithm (Root-Scale-invariant feature can be used
Transform, scale invariant feature conversion) carry out feature extraction.It specifically includes: the picture construction compared firstly, for needs
Scale space is exactly to carry out convolution algorithm using the difference of Gaussian convolution kernel and image of different scale.Secondly, utilizing each sampling
Point is compared with all the points around it, finds the extreme point in difference of Gaussian space, and by being fitted three-dimensional quadratic function
Remove the very asymmetric pixel of Gaussian convolution space local curvature.Again, the gradient direction of key point field pixel is utilized
Distribution character is each key point assigned direction parameter, so that operator is had 128 dimension directioin parameters of rotational invariance, here it is passes
The Feature Descriptor of key point is normalized description and root mean square is taken to obtain final Expressive Features.
Characteristic matching unit 203, the feature for being extracted feature extraction unit 202 are matched, are matched
Point.Sub- matching is described in each scale image in the image for extracting feature, when the distance of description is less than threshold value
It can be judged as successful match, obtain match point.
Image matching unit 204, for carrying out images match according to the distributing position and/or match point quantity of match point.
It include: to judge whether image matches according to the distributing position and/or match point quantity of match point, if match point spatial distribution is consistent
And match point quantity reaches preset value, then confirms images match success.Match point quantity includes following one kind: all match points
Quantity, or remove the match point quantity after the match point in the pre-determined distance of point of interest label zone boundary；Or removal
Match point quantity after the match point that line intersects；Or remove the matching in the pre-determined distance of point of interest label zone boundary
Match point quantity after the match point that point and line intersect.Match point quantity is preferably removed apart from point of interest label zone boundary
Match point quantity after the match point that match point and line in pre-determined distance intersect.At the edge in point of interest label region
It with having little significance for point, therefore can remove it, the matching line is interdigital: the match point in two images is connected
The line come has intersection.Matching line is generally parallel line, if of the matching line and other match points of certain match point
Intersect with line, then the match point that can intersect line removes.
When the target image and comparing image successful match, then judge that image repeats, deletes multiimage to only retaining
One image.Specifically, can only retain the most image of feature quantity when being deleted and being retained, alternatively, retaining
The newest image of shooting time.Feature means that image is more clear, and the electronic map precision processed is also higher therefore excellent
Select the image that keeping characteristics are most.
High-precision image matching system provided in an embodiment of the present invention is suitable for electronic map, especially high precision electro
The production of sub- map.The embodiment of the present invention is designed by algorithm optimization, images match precision is improved, so as to remove redundancy
Image improves the producing efficiency of map especially high-precision electronic map.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
Within the technical scope of the present disclosure, any changes or substitutions that can be easily thought of by anyone skilled in the art,
It should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with the protection model of claims
Subject to enclosing.
Claims (10)
1. a kind of high-precision image matching method characterized by comprising
It determines target image and compares the point of interest label region in image；
Feature extraction is carried out to point of interest label region；
The feature extracted is matched, match point is obtained；
Images match is carried out according to the distributing position of match point and/or match point quantity.
2. the method according to claim 1, wherein
The determining target image and the point of interest label region compared in image include:
Coarse positioning is carried out to the point of interest label region in image using MSER algorithm, HOG algorithm and SVM algorithm is recycled to carry out
It is accurately positioned.
3. the method according to claim 1, wherein
Described includes: using Root-sift algorithm to the point of interest mark to the progress feature extraction of point of interest label region
Board region carries out feature extraction.
4. the method according to claim 1, wherein
The feature extracted is matched, obtaining match point includes: by each scale in the image for extracting feature
Sub- matching is described in image, can be judged as successful match when the distance of description is less than threshold value, obtain match point.
5. the method according to claim 1, wherein
It is described to include: according to the distributing position and/or match point quantity of match point progress images match
Judge whether image matches according to the distributing position of match point and/or match point quantity, if match point spatial distribution is consistent
And match point quantity reaches preset value, then judges images match success.
6. a kind of high-precision image matching system characterized by comprising
Point of interest label area determination unit, for determining target image and comparing the point of interest label region in image；
Feature extraction unit, for carrying out feature extraction to point of interest label region；
Characteristic matching unit obtains match point for matching the feature extracted；
Image matching unit, for carrying out images match according to the distributing position and/or match point quantity of match point.
7. system according to claim 6, which is characterized in that
The determining target image and the point of interest label region compared in image include:
Coarse positioning is carried out to the point of interest label region in image using MSER algorithm, HOG algorithm and SVM algorithm is recycled to carry out
It is accurately positioned.
8. system according to claim 6, which is characterized in that
Described includes: using Root-sift algorithm to the point of interest mark to the progress feature extraction of point of interest label region
Board region carries out feature extraction.
9. system according to claim 6, which is characterized in that
The feature extracted is matched, obtaining match point includes: by each scale in the image for extracting feature
Sub- matching is described in image, can be judged as successful match when the distance of description is less than threshold value, obtain match point.
10. system according to claim 6, which is characterized in that
It is described to include: according to the distributing position and/or match point quantity of match point progress images match
Judge whether image matches according to the distributing position of match point and/or match point quantity, if match point spatial distribution is consistent
And match point quantity reaches preset value, then judges images match success.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710584412.8A CN109271995A (en) | 2017-07-18 | 2017-07-18 | A kind of high-precision image matching method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710584412.8A CN109271995A (en) | 2017-07-18 | 2017-07-18 | A kind of high-precision image matching method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109271995A true CN109271995A (en) | 2019-01-25 |
Family
ID=65152424
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710584412.8A Pending CN109271995A (en) | 2017-07-18 | 2017-07-18 | A kind of high-precision image matching method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109271995A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116309741A (en) * | 2023-05-22 | 2023-06-23 | 中南大学 | TVDS image registration method, segmentation method, device and medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101275854A (en) * | 2007-03-26 | 2008-10-01 | 日电（中国）有限公司 | Method and equipment for updating map data |
| CN103218427A (en) * | 2013-04-08 | 2013-07-24 | 北京大学 | Local descriptor extracting method, image searching method and image matching method |
| CN105069144A (en) * | 2015-08-20 | 2015-11-18 | 华南理工大学 | Similar image search method |
| CN106203342A (en) * | 2016-07-01 | 2016-12-07 | 广东技术师范学院 | Target identification method based on multi-angle local feature coupling |
| CN106529591A (en) * | 2016-11-07 | 2017-03-22 | 湖南源信光电科技有限公司 | Improved MSER image matching algorithm |
| CN106846608A (en) * | 2017-01-25 | 2017-06-13 | 杭州视氪科技有限公司 | A kind of visually impaired people's paper money recognition glasses based on RGB D cameras |
-
2017
- 2017-07-18 CN CN201710584412.8A patent/CN109271995A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101275854A (en) * | 2007-03-26 | 2008-10-01 | 日电（中国）有限公司 | Method and equipment for updating map data |
| CN103218427A (en) * | 2013-04-08 | 2013-07-24 | 北京大学 | Local descriptor extracting method, image searching method and image matching method |
| CN105069144A (en) * | 2015-08-20 | 2015-11-18 | 华南理工大学 | Similar image search method |
| CN106203342A (en) * | 2016-07-01 | 2016-12-07 | 广东技术师范学院 | Target identification method based on multi-angle local feature coupling |
| CN106529591A (en) * | 2016-11-07 | 2017-03-22 | 湖南源信光电科技有限公司 | Improved MSER image matching algorithm |
| CN106846608A (en) * | 2017-01-25 | 2017-06-13 | 杭州视氪科技有限公司 | A kind of visually impaired people's paper money recognition glasses based on RGB D cameras |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116309741A (en) * | 2023-05-22 | 2023-06-23 | 中南大学 | TVDS image registration method, segmentation method, device and medium |
| CN116309741B (en) * | 2023-05-22 | 2023-08-11 | 中南大学 | TVDS image registration method, segmentation method, device and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108763287B (en) | Construction method of large-scale passable regional driving map and unmanned application method thereof | |
| CN109753885B (en) | Target detection method and device and pedestrian detection method and system | |
| CN111462275A (en) | Map production method and device based on laser point cloud | |
| Karantzalos et al. | Large-scale building reconstruction through information fusion and 3-d priors | |
| CN105701798A (en) | Point cloud extraction method and device for columnar object | |
| CN102804231A (en) | Piecewise planar reconstruction of three-dimensional scenes | |
| CN111435421B (en) | Traffic-target-oriented vehicle re-identification method and device | |
| CN105953773B (en) | Ramp slope angle acquisition methods and device | |
| CN108225334A (en) | A kind of localization method and device based on three-dimensional live-action data | |
| Xu et al. | A new clustering-based framework to the stem estimation and growth fitting of street trees from mobile laser scanning data | |
| Yadav et al. | Identification of trees and their trunks from mobile laser scanning data of roadway scenes | |
| CN113255578B (en) | Traffic identification recognition method and device, electronic equipment and storage medium | |
| CN110689573A (en) | Edge model-based augmented reality label-free tracking registration method and device | |
| CN109146773B (en) | Method and device for mapping river channel map to Web map | |
| CN110798805A (en) | Data processing method and device based on GPS track and storage medium | |
| Liu et al. | Deep-learning and depth-map based approach for detection and 3-D localization of small traffic signs | |
| CN107563366A (en) | A kind of localization method and device, electronic equipment | |
| CN113963259A (en) | Street view ground object multi-dimensional extraction method and system based on point cloud data | |
| Liu et al. | Image-translation-based road marking extraction from mobile laser point clouds | |
| CN115375857A (en) | Three-dimensional scene reconstruction method, device, equipment and storage medium | |
| CN103744903B (en) | A kind of scene image search method based on sketch | |
| CN111611900A (en) | Target point cloud identification method and device, electronic equipment and storage medium | |
| CN111383286A (en) | Positioning method, positioning device, electronic equipment and readable storage medium | |
| CN114758086A (en) | Method and device for constructing urban road information model | |
| Wu et al. | A stepwise minimum spanning tree matching method for registering vehicle-borne and backpack LiDAR point clouds |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190125 |
|
| RJ01 | Rejection of invention patent application after publication |