CN102034103A - Lineament extraction method of remote sensing image - Google Patents
Lineament extraction method of remote sensing image Download PDFInfo
- Publication number
- CN102034103A CN102034103A CN 201010578935 CN201010578935A CN102034103A CN 102034103 A CN102034103 A CN 102034103A CN 201010578935 CN201010578935 CN 201010578935 CN 201010578935 A CN201010578935 A CN 201010578935A CN 102034103 A CN102034103 A CN 102034103A
- Authority
- CN
- China
- Prior art keywords
- key element
- image
- linear key
- linear
- gradient direction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a lineament extraction method of a remote sensing image, comprising the following steps of: (1) inputting a remote sensing image and carrying out the preprocessing on the image enhancement; (2) establishing a multi-resolution structural image sequence with a pyramid structure; (3) extracting lineaments in different levels for all levels of images with different resolutions according to the consistency of gradient direction and the consistency of the regional gray; and (4) establishing characteristic database for the extracted lineaments and managing the extracted lineaments. In the method, by carrying out the pyramid classification on the remote sensing image and extracting different lineaments, the application range can be widened; by fusing the gradient direction of the lineaments with the gray linear characteristics, the precision for recognizing the linearity can be improved; the calculated quantity is reduced and the leaking detection and the false detection are overcome by using a fixed area while the lineaments are generated; and the lineament extraction method has favorable robustness and accuracy.
Description
Technical field
The present invention relates to remote sensing technology and technical field of image processing, particularly relate to the extracting method of the linear key element of high-resolution remote sensing image.
Background technology
Because the height diversity and the complicacy of target in the remote sensing image, successful object target automatic recognition system will provide theory and the method with general guidance meaning for the image understanding problem of other types.Therefore, the object target of how discerning and extracting in the remote sensing image is one of important research project in the remote sensing images Target Recognition.It is the important research content of image geometry content that linear key element is extracted.At first, most of culture all is the plane; The second, the shape of most of object targets is all represented based on the linear straight line of bottom.Detected edge line section is added up and classified, obtain dissimilar linear line segments, thereby realize different objects identification of targets and detection.
Along with the application of high-resolution remote sensing image, the means that linear key element is extracted as the basic configuration of evaluating objects object more and more are subjected to people's attention.
Linear key element extracting method relatively more commonly used at present mainly comprises canny operator algorithm (D.H.Ballard.Generalizingthe hough transform to detect arbitrary shapes.PR, 13 (2): 111-122,1981.), this method is by Hough conversion (Hough transform) (J.Canny.A computational approach to edge detection.IEEE Trans.Pattern Anal.Mach.Intell., 8 (6): 679-698,1986.) extract the lines that comprise a large amount of marginal points that all surpass certain threshold value.These lines are divided into a lot of line segments by thresholding and length threshold.The Hough conversion has very serious defective: at first, the texture region of image has very high marginal density, often causes a lot of wrong detections; Owing to ignore the direction of marginal point, algorithm obtains the linear partition result of abnormal direction simultaneously.In addition, these class methods need setting threshold, thereby cause this method intellectuality not carry out linear key element extraction.
Another classical algorithm equally also is to begin to calculate from marginal point, utilize marginal point to connect into curve, then be divided into segment of curve and straight line line segment (A.Etemadi.Robust segmentation of edgedata.Int.Conf.on Image Processing and its Applications by straight line standard (the chain code criterion of standard) then, pages 311-314,1992; O.Faugeras, R.Deriche, H.Mathieu, N.J.Ayache, and G.Randall.The depth and motion analysis machine.PRAI, 6:353-385,1992.).This method does not need to regulate parameter just can obtain more accurate result, and this algorithm detects lines and arc simultaneously, but understands a lot of straight lines of omission and little boundary curve usually.
Burns, Hanson and Riseman have introduced one based on linear session linear partition detection algorithm.This algorithm is not from marginal point, in fact consider the direction of gradient and do not considered mould value (the J.Brian Burns of gradient, Allen R.Hanson, and Edward M.Riseman.Extracting straight lines.IEEE Trans.PAMI, 8 (4): 425-455,1986.).Kahn, Kitchen and Riseman improve this algorithm, its line segment is cut apart good location, but still do not solve threshold value problem (P.Kahn, L.Kitchen, and E.M.Riseman.Real-time featureextraction:A fast line finder for vision-guided robot navigation.Technical Report 87-57, COINS, 1987; P.Kahn, L.Kitchen, and E.M.Riseman.A fast line finder for visionguided robot navigation.IEEE Trans.Pattern Anal.Mach.Intell., 12 (11): 1098-1102,1990.).Desolneux, Moisan and Moral have carried out detail analysis to the threshold value problem, their method is to go to calculate the point of a large amount of alignment and find linear partition as isolated point with a non-structural model, unfortunately, be the wrong simultaneously array (A.Desolneux that the alignment line segment is cut apart that produced like this, L.Moisan, and J.M.Morel.Meaningful alignments.International Journal ofComputer Vision, 40 (1): 7-23,2000.).
Summary of the invention
The extracting method that the purpose of this invention is to provide a kind of linear key element based on high resolution image, can detect linear line segment exactly under the situation of manual shift parameter there not being flase drop to survey and do not need, extract with the linearity that is used for remote sensing image, extract especially for the linearity of high-resolution remote sensing image.
For achieving the above object, the present invention adopts following technical scheme:
A kind of linear key element extracting method of remote sensing image may further comprise the steps:
1) input remote sensing image and carry out the pre-service that image strengthens;
2) pretreated image is set up the pyramid structure image sequence of multiresolution;
3) images at different levels of different resolution in the pyramid structure image sequence are extracted the linear key element of different stage according to gradient direction consistance and area grayscale consistance;
4) the linear key element of extracting being set up property data base manages.
Above-mentioned steps 1) at first be that original remote sensing image is input in the database, it is the attribute of the original remote sensing image of typing, set up the metadata management of remote sensing image simultaneously, the attribute of described original remote sensing image comprises base attributes such as the identifier of raw video, image name, precision, coordinate, projection pattern, imaging time, image type, memory location, presentation content description entry and quality of image description entry; Then remote sensing image is strengthened.The main contents that remote sensing image is strengthened comprise:
A. raw video is carried out histogram equalization: original remote sensing image is carried out closed operation, reduce the texture information of image to be identified, strengthen picture contrast.
B. the background of remote sensing image suppresses to handle: the image behind the histogram equalization is carried out morphology increase and the inhibition of morphology background.
For pretreated remote sensing image,, effectively extract linear key element in conjunction with the feature of gradient direction consistance and gray consistency by setting up the image sequence of pyramid structure.
Above-mentioned steps 2) set up multiresolution the pyramid structure image sequence method preferably: establish σ
x, σ
yThe multiple that the image of the low resolution that obtains after handling through layering for higher resolution image reduces in x direction and y direction size is with f
0(x y) represents the initial pictures that original remote sensing image obtains after the step 1) pre-service, it is the highest at the image sequence intermediate-resolution, obtains resolution each tomographic image (referring to Fig. 2) from high to low successively according to following recurrence relation:
f
l(x，y)＝(σ
x×σ
y)
-1×{f
l-1(2x-1，2y-1)+f
l-1(2x-1，2y)+f
l-1(2x，2y-1)+f
l-1(2x，2y)}
Wherein l is the integer of 0～L, and L represents the number of times that layering is handled.Initial pictures is handled the image sequence that obtains through the L secondary clearing is: f
0(x, y), f
1(x, y), f
2(x, y) ..., f
L-1(x, y), f
L(x, y).
Above-mentioned steps 3) effectively the image of different resolution is extracted the linear key element of different stage in conjunction with the feature of gradient direction consistance and gray consistency.Can extract linear key element to images at different levels by resolution order from high to low, promptly earlier the initial pictures of original resolution be extracted linear key element, successively the image of low resolution be extracted linear key element then.
Carry out region growing and gray consistency detection by images at different levels, set up linear key element at different levels, specifically comprise the steps: the pyramid image sequence
I) set the linear search width, carry out region growing according to the gradient direction consistance;
Ii) to step I) the linear key element of the gradient direction unanimity that obtains, extract linear key element according to gray consistency again.
Step I) at first needs the gradient direction of each pixel in the computed image, set up the consistent criterion of linear key element,, think that then these two pixels have linear relationship if promptly the angle of the gradient direction of two pixels is smaller or equal to π/8 by gradient direction.
The concrete grammar of the gradient direction of each pixel preferably in the computed image: at first be with the 2D gaussian filtering image to be carried out filtering (as shown in Equation 1), remove picture noise.
G (x, y)=(2
-1π σ
2) exp[-(x
2+ y
2)/2 σ
2] formula 1
Wherein, (x y) is pixel in the image.
Calculate the derivative G of the gray scale of each pixel in the image with the single order differential of Gauss operator then along x direction and y direction
xAnd G
y, calculate the gradient direction θ of pixel thus:
θ=arctan (G
y/ G
x) formula 2
We define linear criterion: the gradient direction of supposing pixel i is θ
i, the gradient direction of another pixel i+1 is θ
I+1, from the gradient direction θ of pixel i
iCarry out range of linearity growth, according to the definition of linear key element, if two gradient direction θ
iAnd θ
I+1Between angle
Smaller or equal to π/8, just can assert the gradient direction unanimity of these two pixels, have linear relationship, wherein:
P wherein
iBe two gradient direction θ
iAnd θ
I+1Linear relationship tolerance size, as shown in Figure 3.
After setting up the conforming linear criterion of gradient direction, treat surveyed area according to the characteristics of linearity and carry out linear key element search according to certain qualification width, promptly by setting a fixing linear search width, from certain pixel, in the width regions that limits, search for the pixel of gradient direction unanimity with it, when searching the pixel of gradient direction unanimity with it, carry out region growing towards the direction of this pixel.The characteristics of linear key element are to have certain width, and we search for linear key element in fixing width regions the inside just by the fixing width regions of definition, have reduced calculated amount.From certain pixel, detect the scope in 16 fields on every side according to fixed area, certain region conforms linear feature in 16 fields just carries out region growing according to this regional direction around finding.When a certain direction is determined to have linearity, along this direction with its linear relationship of fixed area size detection.
Step I i) image is passed through step I) the linear key element of the gradient that generates, carrying out gray consistency according to fixed area detects, if certain zone and the average gray value difference of adjacent area, then merge these two zones less than certain threshold value and join in the linear key element and go.
Above-mentioned steps 4) for extracting good linear key element, it is managed by setting up property data base.Mainly be to set up different linear key element storehouses, promptly linear key element classified according to the information under the linear key element (comprising attribute informations such as linear key element indications, resolution, linear width, lineal measure) at different resolution and linear goal.
The present invention can be used for the basic comprising element line segment of remote sensing image is extracted, and its application comprises the detection and the extraction of targets such as road and buildings.Method of the present invention is particularly suitable for the linearity of high resolving power (resolution≤2 meter) remote sensing image and extracts, adopted multiresolution hierarchical structure based on the high-resolution remote sensing image, extract linear key element by gradient direction consistance and area grayscale consistance, be different from the method that single employing gradient direction and area validation carry out parameter setting.Concrete, the major advantage of the inventive method comprises: by remote sensing image being carried out the pyramid classification, collinearity key element is not extracted, improved range of application; Utilize gradient direction, the area grayscale principle of correspondence, the gradient direction of linear key element is in the same place with the linear Feature Fusion of gray scale, fixed area by gradient direction and adaptive method and gray consistency are expressed linear feature, have improved the linear precision of identification; When generating linear key element, utilize fixing zone, reduced calculated amount, overcome omission and false retrieval, have good robustness and accuracy.
Description of drawings
Fig. 1 is the synoptic diagram that the linear key element of the present invention is extracted flow process;
Fig. 2 is a synoptic diagram of setting up the multi-resolution image series mould of pyramid structure;
Fig. 3 carries out linear direction consistency detection synoptic diagram according to gradient direction;
Fig. 4 is the fixed range search synoptic diagram of linear direction, the linear direction synoptic diagram that wherein left figure is right figure;
Fig. 5 utilizes the present invention to carry out the synoptic diagram of road extraction process among the embodiment.
Embodiment
Further specify technical matters related in the technical solution of the present invention below in conjunction with the navigation road extraction.Be to be noted that described embodiment only purport be convenient to the understanding of the present invention, and it is not played any qualification effect.
Present embodiment is that example illustrates the extraction of remote sensing image neutral line key element with the navigation road extraction.Road difference in the high resolving power raw video, caused the linear key element of different roads, pyramid structure image sequence by the present invention's proposition, we can extract the road of different brackets, can on the low image of resolution, extract for advanced road, for road smaller in the city, can on the high image of resolution, extract.
Specifically describe each step of this leaching process below.
At first, be the input of high-resolution remote sensing image:
High resolving power raw video data through image registration, inlay and after resampling, framing cut out step, be entered in the high resolving power raw video database, set up the metadata specification of high resolution image simultaneously, comprising: identifier, image name, precision, coordinate, projection pattern, imaging time, image type, memory location, presentation content description entry and quality of image description entry.
Managing and control by the input of database to high resolution image, mainly is to carry out necessary preparation for the pre-service and the feature extraction of image, operates laying the first stone for the extraction of linear key element.
Its two, the high resolution image pre-service:
A. raw video is carried out histogram equalization: raw video is carried out closed operation, reduce the texture information of image to be identified, strengthen picture contrast.
B. the background of high-resolution remote sensing image suppresses to handle: the image to histogram equalization carries out morphology increase and the inhibition of morphology background.
Its three, the linear key element of high resolution image is extracted, and extracts flow process as shown in Figure 1, mainly may further comprise the steps:
1) set up the image sequence of high resolution image pyramid structure: supposing nethermost is initial pictures, σ
x, σ
yThe multiple that the image of the low resolution that obtains after handling through layering for higher resolution image reduces in x direction and y direction size.If it is f that initial pictures is handled the image sequence that obtains through the L secondary clearing
0(x, y), f
1(x, y), f
2(x, y) ..., f
L-1(x, y), f
L(x, y), wherein: f
0(x y) is initial pictures.
Their recurrence relation is represented:
f
l(x，y)＝(σ
x×σ
y)
-1×{f
l-1(2x-1，2y-1)+f
l-1(2x-1，2y)+f
l-1(2x，2y-1)+f
l-1(2x，2y)}
As shown in Figure 2, when L=2, obtain the image of 2 low resolutions of initial pictures.
2) gradient direction of each pixel in the calculating initial pictures: concrete steps at first are with the 2D gaussian filtering image to be carried out filtering, and removal picture noise G (x, y)=(2
-1π σ
2) exp[-(x
2+ y
2)/2 σ
2] calculate the gray scale of each pixel in the image along the derivative G of level with the single order differential of Gauss operator again with vertical (being x and y) both direction
xAnd G
y, calculate gradient direction θ=arctan (G of each pixel
y/ G
x).
We are defined as linear criterion: suppose the gradient direction θ from certain pixel i
iCarry out range of linearity growth,, find the gradient direction θ of pixel i+1 according to the definition of linearity
I+1If, the angle of two gradient directions
Smaller or equal to π/8, just can assert it is linear key element, as shown in Figure 3.
3) treating surveyed area according to the characteristics of linearity searches for according to certain qualification width, be exactly at first to utilize a pixel to calculate this linear direction of growth, then by the fixing regional travel direction growth of definition, we do not need the search of growing of whole zone like this, thereby have reduced calculated amount.Detect the scope in 16 fields on every side according to fixing width regions, certain region conforms linear feature in 16 fields just carries out region growing according to this regional direction around finding.When a certain direction is determined to have linear relationship, carry out the zone according to this direction with the zone of fixed size and detect.As shown in Figure 4, at first determine linear direction, and then be 2 to carry out the fixing search of this direction according to width up and down by 16 fields.
4) the linear key element of the gradient that image is generated is carried out gray consistency according to fixed area and is detected, if should the zone and the difference of the average gray value of adjacent area less than certain threshold value, then be merged into a zone and join in the linear key element and go.
5) then the image of different resolution also carry out step 2)-4) operation, different image in different resolution is extracted the linear key element of different brackets.
Its four, linear feature warehouse-in:
For extracting good linear feature, by setting up property data base it is managed, wherein linear key element is classified by the information (comprising linear key element indications, resolution, linear width, length etc.) of linear key element attribute.
Fig. 5 has illustrated the process of whole road extraction, and wherein A is pretreated image; B is the gradient image that generates; C is that certain grade of image to pyramid series carries out the image that region growing obtains; D is the linear key element that generates in conjunction with gray consistency.
Above-mentioned linear element characteristic extracting method based on high-resolution remote sensing image has great importance, and its major advantage is as follows:
1. utilize the fusion of gradient direction and gray consistency, improved the precision of the extraction of linear key element.
2. utilize fixedly gradient direction zone, reduced amount of calculation, overcome the omission and the false retrieval that produce when utilizing the direction mould to extract, have good robustness and accuracy.
3. by image being carried out pyramid classification, collinearity key element is not extracted, improved the scope of using.
In sum, the present invention has adopted based on the gradient direction of high resolution remote sensing image and gray consistency, is different from the method that setting parameter is carried out in single employing gradient direction and zone checking. This invention utilizes the fusion of gradient direction and the gray scale linear character of linear key element, and the FX by gradient direction and adaptive method and gray consistency are expressed linear feature, can improve the linear precision of identification. The present invention can be used for the basic comprising element line segment of high resolution image is extracted, and its application comprises detection and the extraction of the targets such as road and building.
Claims (10)
1. the linear key element extracting method of a remote sensing image may further comprise the steps:
1) input remote sensing image and carry out the pre-service that image strengthens;
2) pretreated image is set up the pyramid structure image sequence of multiresolution;
3) images at different levels of different resolution are extracted the linear key element of different stage according to gradient direction consistance and area grayscale consistance;
4) the linear key element of extracting being set up property data base manages.
2. linear key element extracting method as claimed in claim 1 is characterized in that, the described pre-service of step 1) is earlier remote sensing image to be carried out histogram equalization, carries out morphology increase and morphology background then and suppresses to handle.
3. linear key element extracting method as claimed in claim 1 is characterized in that step 2) method of setting up the pyramid structure image sequence of multiresolution is: establish σ
x, σ
yThe multiple that the image of the low resolution that obtains after handling through layering for higher resolution image reduces in x direction and y direction size is with f
0(x y) represents the initial pictures that original remote sensing image obtains after the step 1) pre-service, obtain resolution each tomographic image from high to low: f successively according to following recurrence relation
l(x, y)=(σ
x* σ
y)
-1* { f
L-1(2x-1,2y-1)+f
L-1(2x-1,2y)+f
L-1(2x, 2y-1)+f
L-1(2x, 2y) } wherein, l is the integer of 0～L, L represents the number of times that layering is handled.
4. linear key element extracting method as claimed in claim 1 is characterized in that, step 3) is earlier extracted linear key element to the initial pictures of original resolution, successively the image of low resolution is extracted linear key element then.
5. linear key element extracting method as claimed in claim 1 is characterized in that step 3) comprises the process that images at different levels extract linear key element:
I) set the linear search width, carry out region growing according to the gradient direction consistance;
Ii) to step I) the linear key element of the gradient direction unanimity that obtains, extract linear key element according to gray consistency again.
6. linear key element extracting method as claimed in claim 5, it is characterized in that, described step I) gradient direction of each pixel in the computed image at first is if the angle of the gradient direction of two pixels, thinks then that the gradient direction of these two pixels is consistent smaller or equal to π/8; Set a fixing linear search width,, in the width regions that limits, search for the pixel of gradient direction unanimity with it, when searching the pixel of gradient direction unanimity with it, carry out region growing towards the direction of this pixel from certain pixel.
7. linear key element extracting method as claimed in claim 6, it is characterized in that, step I) in the computed image method of the gradient direction of each pixel be: with the 2D gaussian filtering image is carried out filtering, calculates the derivative G of the gray scale of each pixel in the image with the single order differential of Gauss operator then along x direction and y direction
xAnd G
y, calculate gradient direction θ=arctan (G of pixel thus
y/ G
x).
8. linear key element extracting method as claimed in claim 6 is characterized in that step I) in from certain pixel, the scope in 16 fields around detecting according to fixing width regions is to search for the pixel of gradient direction unanimity with it.
9. linear key element extracting method as claimed in claim 5, it is characterized in that, step I i) to passing through step I) the linear key element of the gradient that generates, carrying out gray consistency according to fixed area detects, if the difference of certain zone and the average gray value of adjacent area, then is merged into a zone less than certain threshold value and joins in the linear key element and go.
10. linear key element extracting method as claimed in claim 1 is characterized in that, described remote sensing image is the high-resolution remote sensing image of resolution≤2 meter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010578935XA CN102034103B (en) | 2010-12-03 | 2010-12-03 | Lineament extraction method of remote sensing image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010578935XA CN102034103B (en) | 2010-12-03 | 2010-12-03 | Lineament extraction method of remote sensing image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102034103A true CN102034103A (en) | 2011-04-27 |
| CN102034103B CN102034103B (en) | 2012-11-21 |
Family
ID=43886966
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010578935XA Expired - Fee Related CN102034103B (en) | 2010-12-03 | 2010-12-03 | Lineament extraction method of remote sensing image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102034103B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103218597A (en) * | 2013-03-13 | 2013-07-24 | 中国科学院对地观测与数字地球科学中心 | Method and device for extracting collapsed houses from high-spatial-resolution optical image |
| CN103488801A (en) * | 2013-10-17 | 2014-01-01 | 中国电子科技集团公司第五十四研究所 | Geographical information space database-based airport target detection method |
| CN108090929A (en) * | 2017-12-04 | 2018-05-29 | 国家***第海洋研究所 | The linear anomaly analysis extraction novel method in mining area |
| CN110207671A (en) * | 2018-12-29 | 2019-09-06 | 中国科学院软件研究所 | A kind of space-based intelligence imaging system |
| CN113763280A (en) * | 2021-09-15 | 2021-12-07 | 湖南科技大学 | Region growing algorithm based on spatial hierarchical topological relation for point cloud denoising |
| CN117764871A (en) * | 2024-02-22 | 2024-03-26 | 青岛萨纳斯智能科技股份有限公司 | unmanned aerial vehicle inspection image enhancement method based on artificial intelligence |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101126813A (en) * | 2007-09-29 | 2008-02-20 | 北京交通大学 | High resolution ratio satellite remote-sensing image architecture profile extraction method |
| US20090154792A1 (en) * | 2005-12-07 | 2009-06-18 | Commonwealth Scientific And Industrial Research Organisation | Linear Feature Detection Method and Apparatus |
-
2010
- 2010-12-03 CN CN201010578935XA patent/CN102034103B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090154792A1 (en) * | 2005-12-07 | 2009-06-18 | Commonwealth Scientific And Industrial Research Organisation | Linear Feature Detection Method and Apparatus |
| CN101126813A (en) * | 2007-09-29 | 2008-02-20 | 北京交通大学 | High resolution ratio satellite remote-sensing image architecture profile extraction method |
Non-Patent Citations (2)
| Title |
|---|
| 《仪器仪表学报》 20040831 高芳琴，吴健平 遥感图像识别中线性地物提取方法研究 653-656 1-10 第25卷, 第S1期 * |
| 《海洋科学进展》 20041031 肖晓柏，李培军 用方向性的形态变换方法换取遥感图像上线性特征 71-78 1-10 第22卷, * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103218597A (en) * | 2013-03-13 | 2013-07-24 | 中国科学院对地观测与数字地球科学中心 | Method and device for extracting collapsed houses from high-spatial-resolution optical image |
| CN103218597B (en) * | 2013-03-13 | 2016-05-11 | 中国科学院对地观测与数字地球科学中心 | From high spatial resolution optical image, extract method and the device of collapsed house |
| CN103488801A (en) * | 2013-10-17 | 2014-01-01 | 中国电子科技集团公司第五十四研究所 | Geographical information space database-based airport target detection method |
| CN103488801B (en) * | 2013-10-17 | 2016-09-28 | 中国电子科技集团公司第五十四研究所 | A kind of airport target detection method based on geographical information space database |
| CN108090929A (en) * | 2017-12-04 | 2018-05-29 | 国家***第海洋研究所 | The linear anomaly analysis extraction novel method in mining area |
| CN108090929B (en) * | 2017-12-04 | 2021-12-03 | 国家***第一海洋研究所 | Novel method for analyzing and extracting linear abnormity of mining area |
| CN110207671A (en) * | 2018-12-29 | 2019-09-06 | 中国科学院软件研究所 | A kind of space-based intelligence imaging system |
| CN113763280A (en) * | 2021-09-15 | 2021-12-07 | 湖南科技大学 | Region growing algorithm based on spatial hierarchical topological relation for point cloud denoising |
| CN113763280B (en) * | 2021-09-15 | 2023-11-21 | 湖南科技大学 | Region growing method based on space hierarchy topological relation for point cloud denoising |
| CN117764871A (en) * | 2024-02-22 | 2024-03-26 | 青岛萨纳斯智能科技股份有限公司 | unmanned aerial vehicle inspection image enhancement method based on artificial intelligence |
| CN117764871B (en) * | 2024-02-22 | 2024-04-30 | 青岛萨纳斯智能科技股份有限公司 | Unmanned aerial vehicle inspection image enhancement method based on artificial intelligence |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102034103B (en) | 2012-11-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Alcantarilla et al. | Street-view change detection with deconvolutional networks | |
| CN102034103B (en) | Lineament extraction method of remote sensing image | |
| CN104217427B (en) | Lane line localization method in a kind of Traffic Surveillance Video | |
| CN105184801A (en) | Optical and SAR image high-precision registration method based on multilevel strategy | |
| CN104239898A (en) | Method for carrying out fast vehicle comparison and vehicle type recognition at tollgate | |
| CN102243705B (en) | Method for positioning license plate based on edge detection | |
| CN110197173B (en) | Road edge detection method based on binocular vision | |
| Yuan et al. | Learning to count buildings in diverse aerial scenes | |
| Zhang et al. | Rapid inspection of pavement markings using mobile LiDAR point clouds | |
| CN109101981A (en) | Winding detection method based on global image bar code under a kind of streetscape scene | |
| CN103632376A (en) | Method for suppressing partial occlusion of vehicles by aid of double-level frames | |
| CN102446356A (en) | Parallel and adaptive matching method for acquiring remote sensing images with homogeneously-distributed matched points | |
| Yang et al. | A skeleton-based hierarchical method for detecting 3-D pole-like objects from mobile LiDAR point clouds | |
| Yao et al. | Automatic extraction of road markings from mobile laser-point cloud using intensity data | |
| Wu et al. | Recognition of airport runways in FLIR images based on knowledge | |
| Hu et al. | Generalized image recognition algorithm for sign inventory | |
| Ma et al. | Semi-automated framework for generating cycling lane centerlines on roads with roadside barriers from noisy MLS data | |
| CN102004921A (en) | Target identification method based on image characteristic analysis | |
| Zou et al. | PatchAugNet: Patch feature augmentation-based heterogeneous point cloud place recognition in large-scale street scenes | |
| Li et al. | An Effective Algorithm for Video‐Based Parking and Drop Event Detection | |
| Li et al. | Instance-aware semantic segmentation of road furniture in mobile laser scanning data | |
| Soilán et al. | Deep Learning-Based Road Segmentation of 3D Point Clouds for Assisting Road Alignment Parameterization | |
| Rahimi et al. | Automatic road extraction based on integration of high resolution lidar and aerial imagery | |
| Huang et al. | Expressway Guardrail Extraction and Classification Based on Slice Method. | |
| Volkov et al. | Objects description and extraction by the use of straight line segments in digital images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| 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: 20121121 Termination date: 20201203 |