CN103413131A - Tower crane recognition method based on spectral and geometric characteristics - Google Patents
Tower crane recognition method based on spectral and geometric characteristics Download PDFInfo
- Publication number
- CN103413131A CN103413131A CN2013100254732A CN201310025473A CN103413131A CN 103413131 A CN103413131 A CN 103413131A CN 2013100254732 A CN2013100254732 A CN 2013100254732A CN 201310025473 A CN201310025473 A CN 201310025473A CN 103413131 A CN103413131 A CN 103413131A
- Authority
- CN
- China
- Prior art keywords
- tower crane
- value
- image
- target
- atural object
- 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)
- Image Processing (AREA)
Abstract
According to the invention, a tower crane is recognized from an aerial image by use of inherent spectral characteristics and special geometric characteristics of the tower crane to obtain a location coordinate and number information of the tower crane. Firstly, the tower crane and surface features similar to spectrum of the tower crane are extracted from the image by use of the spectral characteristics of the tower crane; then, the tower crane is further extracted by use of the geometric characteristics of the tower crane such as a unique length-width ratio, area and the like; and extraction results reach an accuracy of 100% and good resistance to the effects of noise is achieved.
Description
Technical field: the present invention relates to a kind of tower crane image-recognizing method of practicality, be specifically related to target identification and digital image processing techniques, it is having a wide range of applications aspect construction management monitoring.
Background technology: image object identification is the target signature of utilizing from extracting image, realizes detection, location and classification to target, is the integrated application of pattern-recognition and digital image processing techniques.(referring to document: Ma Yingjun. based on the military aircraft recognizer of rough set and support vector machine. Northwestern Polytechnical University, 2007) image object identification is mainly used in identification and the classification of the feature atural objects such as aircraft, bridge, highway, river and car plate, also do not identify at present the concrete grammar of tower crane, but tower crane is as the symbol of building ground, application is very widely being arranged aspect the monitoring of construction investment, therefore carrying out tower crane image object identification tool and be of great significance.
Target identification method mainly is divided into two kinds: based on identification and the structural method of decision-theoretic approach.(referring to document: Paul Gonzales. Digital Image Processing. Beijing: Electronic Industry Press, 2008) based on the identification of decision-theoretic approach, be based on the identification of decision-making (or differentiation) function.Make x=(x
1,
2..., x
n)
TRepresent a n dimensional pattern vector.To W Pattern Class ω
1, ω
2..., ω
W, the basic problem of decision theory pattern-recognition is to find W discriminant function d according to attribute
1(x), d
2(x) ..., d
W(x), if pattern x belongs to class ω
i,
d
i(x)＞d
j(x)j＝1，2，…，W；j≠i (1)
(referring to document: Duda, R.O, Hart, P.E., and Stork, D.G.Pattern Classification, John Wiley& Sons, New York, 2001)
Identification based on decision-theoretic approach can be divided into coupling, best statistical sorter and neural net method again.(referring to document: Paul Gonzales. Digital Image Processing. Beijing: Electronic Industry Press, 2008) based on the recognition technology of coupling by a kind of each class of prototype pattern vector representation.Unknown pattern is endowed one by the defined tolerance class the most close with it in advance.The simplest method is minimum distance classifier, and this method just as its name implies, be calculated the distance between (in Euclidean space) unknown quantity and each prototype vector.Select minor increment wherein to carry out decision-making.Also have the matching process that a kind of foundation is relevant, the size of finding coupling in size is the image of M*N is the subgraph of J*K, determines that the related coefficient between two width figure is mated.Formula is as follows:
Although related function can, by using related coefficient normalization, obtain normalized change in size and rotate while changing more difficult for changes in amplitude.When occurring that uncertain or free rotation changes, seldom use correlation technique.(referring to document: Tou, J.T., and Gonzalez, R.C.Pattern Recognition Principles, Addison-Wesley, Reading, Mass, 1974)
Best statistical sorter is a kind of probabilistic method.At great majority, process the occasion of measuring and judging physical event, usually can produce random pattern classification, comprising gaussian model class and Bayes classifier.(referring to document: Jain, A.K., Duin, R.P.W., and Mao, J.Statistical Pattern Recognition:A Review, IEEE Trans.Pattern Anal.Machine Intell., 2000,22 (1): 4-37)
Neural network be take and estimated that with sample mode the statistical parameter of each Pattern Class is basis, and training algorithm is divided into: linear separability from class, inseparable class.(referring to document: Principle, J.C., Euliano, N.R., and Lefebre, W.C.Neural and Adaptive Systems:Fun-damentals through Simulations, John Wiley& Sons, New York, 1999)
Above these methods all have been widely used in the aspects such as military and traffic, along with China is increasing to the supervision of house, factory construction, people strengthen gradually to the attention degree of living space and soil utilization, as the typical heavy construction equipment of construction site---and tower crane also should be as one of research object of target identification.Due to remote sensing images have Real-Time Monitoring and on a large scale observation characteristic, can fully utilize remote sensing images uses target identification method to detect position and the number of tower crane, and by the analysis to tower crane number and distribution situation in statistical regions, construction project is carried out to space orientation, judge simultaneously and monitor the construction speed of investment project in demonstration area.Its Research Significance and importance can compare favourably with the identification of the typical objects such as aircraft, highway.
Summary of the invention:
Based on the tower crane recognition methods of spectrum and geometric properties, it is characterized in that first utilizing spectral information to tower crane and close with the tower crane color
Atural object extract, the recycling tower crane geometrical property tower crane and other atural object are distinguished, concrete steps are as follows:
(1) according to spectral information, slightly extract target and target class like spectrum atural object
By observing atural object, find the atural object close with the tower crane color crooked dirt road and mound are arranged, and residual earth on ground; Because tower crane and its background image color characteristic differ larger, can to Target scalar, slightly extract according to spectral information, obtain target and target class like spectrum atural object; In the color table at R, G, B, yellow and blue complementary, so the R of gilvous, G, B value are respectively 255,255,0; Utilize respectively R, the G of aviation image and the histogram of tri-wave bands of B to carry out analyzing and processing;
1. according to the histogram distribution situation of R wave band, setting threshold is less than 160, and the pixel that is less than this value all is set to zero at the gray-scale value of three wave bands;
2. according to the histogram distribution situation of G-band, setting threshold is less than 100, and the pixel that is less than this value all is set to zero in the DN of three wave bands value;
3. according to B wave band histogram distribution situation, setting threshold is greater than 170, and the pixel that is greater than this value all is set to 0 in the DN of three wave bands value;
(2) according to the Extraction of Geometrical Features Target scalar
1. cromogram is changed into to gray-scale map, and it is carried out to the automatic threshold processing, then image is carried out to binaryzation, wherein, the method that the automatic threshold processing is adopted is maximum variance between clusters, by setting threshold, make the variance between target image and background image reach maximum, threshold value now is carries out by gray-scale map the threshold value that binaryzation adopts;
The binary map that 2. will obtain is carried out Labelling Regions, utilize neighbours territory method, each point in image is scanned, if in the scope of the neighbours territory of this point, there be the point identical with its DN value, so this point is classified as to same class with analyzing spot, is labeled as same numerical value;
3. the geometric properties special according to tower crane, distinguish thing like itself and its spectral class accurately, sets the length and width ratio and be greater than 6, and dirt road, hill and noise are removed; Area by being greater than 10 pixels and certain pitch angle can be removed the crack between fragment of brick; Adopt main axis length to limit, thick head part is removed, but in most cases, tower crane is blocked by the buildings major part, now to by the minimum and maximum value of tower crane main shaft width and length, extract according to the different characteristics of tower crane in image comprehensive above the analysis, respectively the subregion obtained is calculated to its feature, comprise area, focus point coordinate, trend angle and main axis length and main shaft width; And qualifications is set, accurately extract tower crane arm part;
4. the center point coordinate of tower crane and quantity thereof in calculating chart.
Of the present invention be mainly utilize tower crane in remote sensing images spectral information and geometric properties by it from aviation image, extracting and obtain its coordinate position and quantity.Take full advantage of the tower crane spectral characteristic different from other atural objects and special geometric configuration, both eliminated the impact of noise, strengthened again the information of tower crane, be convenient to carry out sample survey and the verification of the investment project information of large scale.
The accompanying drawing explanation: Fig. 1 is that this method utilizes remotely-sensed data to carry out the process flow diagram of tower crane position and number of extracted;
Fig. 2 .1 is the Yunnan Airport unmanned plane aerial images on Dec 30th, 2009;
Fig. 2 .2 utilizes the spectral characteristic of tower crane to carry out threshold value extraction figure as a result afterwards to image R wave band;
Fig. 2 .3 is that the G-band to Fig. 2 .2 carries out the as a result figure of threshold value after extracting;
Fig. 2 .4 is that the B wave band to Fig. 2 .3 carries out tower crane that threshold value obtains after extracting and the extraction result images of the similar spectrum atural object of tower crane;
Fig. 2 .5 carries out to Fig. 2 .4 the figure as a result that automatic threshold extracts rear binaryzation;
Fig. 2 .6 is to the net result of Fig. 2 .5 according to the Extraction of Geometrical Features tower crane.
Embodiment:
One, according to spectral information, slightly extract target and target class like spectrum atural object
At first observe the atural object in accompanying drawing 2.1, find that tower crane is yellow, the atural object close with the tower crane color has crooked dirt road and mound, and residual earth etc. on ground.And tower crane and its background image color characteristic differ larger, can to Target scalar, slightly extract according to spectral information, obtain target and target class like spectrum atural object.Spectral information is expressed by color model, and the color model in optics mainly contains: HSV model, RGB model, HIS model, CHL model, LAB model and CMY model etc.Wherein, RGB (red, green, blue) model is the model towards hardware the most general in reality.(referring to document: Fortner, B., and Meyer, T.E.Number by Colors, Springer-Verlag, New York, 1997).Due to ubiquity and the versatility of RGB model, we utilize the RGB model to analyze and extract the spectral information of tower crane image.In the color table of R, G, B, each spectral components is based on respectively three coordinate axis of cartesian coordinate system, and the span of gray shade scale is 0 to 255, and the rgb value of black is (0,0,0), and white is (255,255,255).In rgb space, be called pixel depth be used to the bit number that means each pixel.Consider the RGB image, wherein each width red, green, blue image is all 8 bit image, and under this condition, each RGB colour element is called 24 bit-depths.Full-color image is commonly used to define the coloured image of 24 bits.In 24 bit RGB images, the color sum is (256*256*256=16777216).(referring to document: Paul Gonzales. Digital Image Processing. Beijing: Electronic Industry Press, 2008) in the RGB model, yellow and blue complementary, so the R of gilvous, G, B value are respectively 255,255,0.Can find out, gilvous is all bigger than normal in the value of red and green spectral, less than normal at the blue color spectrum component value.The present invention utilizes this characteristic, according to R, the G of aviation image and the statistical distribution histogram of tri-wave bands of B, carries out analyzing and processing respectively.
1, the histogram distribution situation of 2.1 R wave band with reference to the accompanying drawings, it is 160 that threshold value is set, and the pixel that is less than this value all is set to zero in the DN of three wave bands value, the pixel of the low gray level of filtering R wave band, obtain accompanying drawing 2.2.Can see that part blueness and grey atural object are by filtering.
2, the histogram distribution situation of 2.2 G-band with reference to the accompanying drawings, setting threshold is 100, and the pixel that is less than this value all is set to zero in the DN of three wave bands value, the pixel of the low gray level of filtering G-band, obtain accompanying drawing 2.3.Red atural object is also by filtering.
3,2.3 B wave band histogram distribution situation with reference to the accompanying drawings, setting threshold is 170, and the pixel that is greater than this value all is set to 0 in the DN of three wave bands value, the pixel of filtering B wave band high grade grey level obtains meeting yellow RGB characteristic set of pixels---accompanying drawing 2.4.Can find out, yellow atural object nearly all has been extracted out, and large-area non-yellow atural object (blueness, grey etc.), all by successful filtering, presents black.
Two, according to the Extraction of Geometrical Features Target scalar
From accompanying drawing 2.4, finding out, the target class extracted also has a lot of tiny spots like spectrum atural object except dirt road, and they are all not identical with shape, the size characteristic of tower crane, and special geometric shape is accurately extracted therefore to utilize tower crane.
1, because Fig. 2 .4 is still cromogram, for convenient, extract, convert it into gray-scale map, and it is carried out to the automatic threshold processing, then image is carried out to binaryzation, obtain accompanying drawing 2.5.Wherein, the method for automatic threshold processing employing is maximum variance between clusters.The method is to make the variance between target image and background image reach maximum by setting threshold, and threshold value so now is carries out by gray-scale map the threshold value that binaryzation adopts.Its algorithm steps following (source document: Otsu, N., AThreshold Selection Method from Gray-Level Histograms, IEEE Transactions on Systems, Man, and Cybernetics, 1979,9 (1): 62-66):
(1) establish image comprise L gray level (0,1 ..., L-1), gray level is that the pixel number of i is N
i, the total pixel number of image is
N＝N
0+N
1+…N
(L-1) (3)
(2) gray level is that the some probability of i is P
i=N
i/ N (4)
(3) threshold value t is divided into dark space c by the view picture image
1With clear zone c
2Two classes, inter-class variance σ is the function of t:
σ＝a
1*a
2*(u
1-u
3)
2 (5)
In formula, a
jFor class c
jThe ratio of area and total image area, a
1=sum (P
i) i → t, a
2=1-a
1
U
1=sum (i*P
i)/a
10 → t, u
2=sum (i*P
i)/a
2T+1 → L-1 selects optimal threshold to make variances sigma reach maximum.
The binary map that 2, will obtain is carried out Labelling Regions.Utilize neighbours territory method, each point in image is scanned, if in the scope of the neighbours territory of this point, there be the point identical with its DN value, so this point is classified as to same class with analyzing spot, be labeled as same numerical value.Wherein, the neighbours territory refers to for each pixel in image, the pixel set of its upper and lower, left and right four direction.Incomplete for fear of image boundarg pixel neighbours territory, the problem that needs classification to discuss, in the process that algorithm is realized, the expansion of at first a Pixel-level being carried out in four borders of image scans again, so not only can improve the efficiency of scanning but also can reduce the complexity of code.
3, the geometric properties special according to tower crane, can distinguish thing like itself and its spectral class accurately.Utilize larger length and width dirt road, hill and noise can be removed than (being greater than 6); Larger area (being greater than 10 pixels) and certain pitch angle can be removed the crack between fragment of brick; Some is black at thicker head place due to tower crane, differs greatly with the tower crane color, therefore in the spectral detection process, has breakpoint, for the quantity tower crane not detected impacts, adopts main axis length to limit, and thick head part is removed.But in most cases, tower crane is blocked by the buildings major part, now to, according to the different characteristics of tower crane in image, extract by the minimum and maximum value of tower crane main shaft width and length.Comprehensive above the analysis, calculate its feature to the subregion obtained respectively, comprises area, focus point coordinate, trend angle and main axis length and main shaft width.And qualifications is set, accurately extract tower crane arm part.
4, the center point coordinate of tower crane and quantity thereof in calculating chart, tower crane quantity is 3, its position in the drawings is respectively: (149.6250 357.3594), (184.6061 107.9242), (214.3350 230.3299), with former Fig. 2, be analyzed and find that the tower crane position, shape and the number that extract fit like a glove, and have reached 100% precision.
The present invention is based on spectrum and geometric properties identifies the tower crane image, utilize spectral information to extract tower crane and the atural object close with the tower crane color, the recycling geometric properties is distinguished tower crane and other atural object, the accuracy rate of identification is high, noiseproof feature is good, can effectively extract the tower crane target in remote sensing images, have good stability and versatility, be with a wide range of applications in building monitoring field.
Claims (3)
1. based on the tower crane recognition methods of spectrum and geometric properties, it is characterized in that first utilizing spectral information to extract tower crane and the atural object close with the tower crane color, the geometrical property of recycling tower crane distinguishes tower crane and other atural object, and concrete steps are as follows:
(1) according to spectral information, slightly extract target and target class like spectrum atural object
By observing atural object, find the atural object close with the tower crane color crooked dirt road and mound are arranged, and residual earth on ground; Because tower crane and its background image color characteristic differ larger, can to Target scalar, slightly extract according to spectral information, obtain target and target class like spectrum atural object; In the color table at R, G, B, yellow and blue complementary, so the R of gilvous, G, B value are respectively 255,255,0; Utilize respectively R, the G of aviation image and the histogram of tri-wave bands of B to carry out analyzing and processing;
1. according to the histogram distribution situation of R wave band, setting threshold is less than 160, and the pixel that is less than this value all is set to zero at the gray-scale value of three wave bands;
2. according to the histogram distribution situation of G-band, setting threshold is less than 100, and the pixel that is less than this value all is set to zero in the DN of three wave bands value;
3. according to B wave band histogram distribution situation, setting threshold is greater than 170, and the pixel that is greater than this value all is set to 0 in the DN of three wave bands value;
(2) according to the Extraction of Geometrical Features Target scalar
1. cromogram is changed into to gray-scale map, and it is carried out to the automatic threshold processing, then image is carried out to binaryzation, wherein, the method that the automatic threshold processing is adopted is maximum variance between clusters, by setting threshold, make the variance between target image and background image reach maximum, threshold value now is carries out by gray-scale map the threshold value that binaryzation adopts;
The binary map that 2. will obtain is carried out Labelling Regions, utilize neighbours territory method, each point in image is scanned, if in the scope of the neighbours territory of this point, there be the point identical with its DN value, so this point is classified as to same class with analyzing spot, is labeled as same numerical value;
3. the geometric properties special according to tower crane, distinguish thing like itself and its spectral class accurately, sets the length and width ratio and be greater than 6, and dirt road, hill and noise are removed; Area by being greater than 10 pixels and certain pitch angle can be removed the crack between fragment of brick; Adopt main axis length to limit, thick head part is removed, but in most cases, tower crane is blocked by the buildings major part, now to by the minimum and maximum value of tower crane main shaft width and length, extract according to the different characteristics of tower crane in image comprehensive above the analysis, respectively the subregion obtained is calculated to its feature, comprise area, focus point coordinate, trend angle and main axis length and main shaft width; And qualifications is set, accurately extract tower crane arm part;
4. the center point coordinate of tower crane and quantity thereof in calculating chart.
2. method according to claim 1, is characterized in that, tower crane presents yellow in true color image, and background is the concrete floor that color differs greatly, and can clearly differentiate.
3. according to the described method of claim 1 or 2, it is characterized in that, the length breadth ratio of tower crane is 6, and the length of main shaft is 20 Pixel-level, and threshold value can be set to be greater than 10 by other trifling atural object filterings.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310025473.2A CN103413131B (en) | 2013-01-24 | 2013-01-24 | Tower crane recognition method based on spectrum and geometric properties |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310025473.2A CN103413131B (en) | 2013-01-24 | 2013-01-24 | Tower crane recognition method based on spectrum and geometric properties |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103413131A true CN103413131A (en) | 2013-11-27 |
| CN103413131B CN103413131B (en) | 2018-05-11 |
Family
ID=49606138
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310025473.2A Expired - Fee Related CN103413131B (en) | 2013-01-24 | 2013-01-24 | Tower crane recognition method based on spectrum and geometric properties |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103413131B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104240196A (en) * | 2014-08-21 | 2014-12-24 | 陕西理工学院 | Four-neighborhood classification method based on FCM |
| CN104299002A (en) * | 2014-10-11 | 2015-01-21 | 嘉兴学院 | Tower crane image detection method based on monitoring system |
| CN106529430A (en) * | 2016-10-31 | 2017-03-22 | 武汉大学 | High spatial resolution image residential area extraction method based on characteristics of spatial correlation and heterogeneity |
| CN106629394A (en) * | 2015-10-28 | 2017-05-10 | 上海振华重工电气有限公司 | Camera external parameter calibration system and method applied to rail-mounted gantry crane hanger pose detection |
| CN111061265A (en) * | 2019-12-06 | 2020-04-24 | 常州节卡智能装备有限公司 | Object carrying method, device and system |
| CN112561487A (en) * | 2020-12-21 | 2021-03-26 | 广联达科技股份有限公司 | Method, device and equipment for calculating construction progress and readable storage medium |
| CN113255626A (en) * | 2021-07-14 | 2021-08-13 | 杭州大杰智能传动科技有限公司 | Intelligent tower crane structure state detection method and device based on scanned image analysis |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101599120A (en) * | 2009-07-07 | 2009-12-09 | 华中科技大学 | A kind of identification method of remote sensing image building |
| US20110150324A1 (en) * | 2009-12-22 | 2011-06-23 | The Chinese University Of Hong Kong | Method and apparatus for recognizing and localizing landmarks from an image onto a map |
| CN102385694A (en) * | 2010-09-06 | 2012-03-21 | 邬明权 | Hyperspectral identification method for land parcel-based crop variety |
-
2013
- 2013-01-24 CN CN201310025473.2A patent/CN103413131B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101599120A (en) * | 2009-07-07 | 2009-12-09 | 华中科技大学 | A kind of identification method of remote sensing image building |
| US20110150324A1 (en) * | 2009-12-22 | 2011-06-23 | The Chinese University Of Hong Kong | Method and apparatus for recognizing and localizing landmarks from an image onto a map |
| CN102385694A (en) * | 2010-09-06 | 2012-03-21 | 邬明权 | Hyperspectral identification method for land parcel-based crop variety |
Non-Patent Citations (2)
| Title |
|---|
| BO YU ET AL: "An automatic and effective approach in identifying tower cranes", 《PROC. SPIE 8334, FOURTH INTERNATIONAL CONFERENCE ON DIGITAL IMAGE PROCESSING》 * |
| 周小成 等: "结合对象关系特征的高分辨率卫星影像建筑物识别方法", 《国土资源遥感》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104240196A (en) * | 2014-08-21 | 2014-12-24 | 陕西理工学院 | Four-neighborhood classification method based on FCM |
| CN104299002A (en) * | 2014-10-11 | 2015-01-21 | 嘉兴学院 | Tower crane image detection method based on monitoring system |
| CN104299002B (en) * | 2014-10-11 | 2017-06-23 | 嘉兴学院 | A kind of tower crane image detecting method based on monitoring system |
| CN106629394A (en) * | 2015-10-28 | 2017-05-10 | 上海振华重工电气有限公司 | Camera external parameter calibration system and method applied to rail-mounted gantry crane hanger pose detection |
| CN106629394B (en) * | 2015-10-28 | 2018-01-16 | 上海振华重工电气有限公司 | Camera extrinsic number calibration system and method applied to the detection of track sling pose |
| CN106529430A (en) * | 2016-10-31 | 2017-03-22 | 武汉大学 | High spatial resolution image residential area extraction method based on characteristics of spatial correlation and heterogeneity |
| CN111061265A (en) * | 2019-12-06 | 2020-04-24 | 常州节卡智能装备有限公司 | Object carrying method, device and system |
| CN112561487A (en) * | 2020-12-21 | 2021-03-26 | 广联达科技股份有限公司 | Method, device and equipment for calculating construction progress and readable storage medium |
| CN112561487B (en) * | 2020-12-21 | 2024-01-23 | 广联达科技股份有限公司 | Method, device and equipment for calculating construction progress and readable storage medium |
| CN113255626A (en) * | 2021-07-14 | 2021-08-13 | 杭州大杰智能传动科技有限公司 | Intelligent tower crane structure state detection method and device based on scanned image analysis |
| CN113255626B (en) * | 2021-07-14 | 2021-12-03 | 杭州大杰智能传动科技有限公司 | Intelligent tower crane structure state detection method and device based on scanned image analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103413131B (en) | 2018-05-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen et al. | A self organizing map optimization based image recognition and processing model for bridge crack inspection | |
| CN103413131A (en) | Tower crane recognition method based on spectral and geometric characteristics | |
| CN103020605B (en) | Bridge identification method based on decision-making layer fusion | |
| Li et al. | Novel approach to pavement image segmentation based on neighboring difference histogram method | |
| Lei et al. | Automated pavement distress detection and deterioration analysis using street view map | |
| CN114998852A (en) | Intelligent detection method for road pavement diseases based on deep learning | |
| CN102096821B (en) | Number plate identification method under strong interference environment on basis of complex network theory | |
| CN111696123A (en) | Remote sensing image water area segmentation and extraction method based on super-pixel classification and identification | |
| CN105975913B (en) | Road network extraction method based on adaptive cluster learning | |
| CN113963222B (en) | High-resolution remote sensing image change detection method based on multi-strategy combination | |
| CN104182763A (en) | Plant type identification system based on flower characteristics | |
| Yuan et al. | Learning to count buildings in diverse aerial scenes | |
| CN109635733B (en) | Parking lot and vehicle target detection method based on visual saliency and queue correction | |
| CN101950359A (en) | Method for recognizing rock type | |
| Bharadwaj Sundra Murthy et al. | Detection of potholes in autonomous vehicle | |
| CN103927511A (en) | Image identification method based on difference feature description | |
| CN107992856B (en) | High-resolution remote sensing building shadow detection method under urban scene | |
| CN104182985A (en) | Remote sensing image change detection method | |
| CN106529461A (en) | Vehicle model identifying algorithm based on integral characteristic channel and SVM training device | |
| CN109635722B (en) | Automatic identification method for high-resolution remote sensing image intersection | |
| Tian et al. | An algorithm combined with color differential models for license-plate location | |
| KR101294347B1 (en) | Mehtod for brick path recognition using image shape pattern and texture feature | |
| Mokhtari et al. | Statistical selection and interpretation of imagery features for computer vision-based pavement crack–detection systems | |
| Wu et al. | Recognition of airport runways in FLIR images based on knowledge | |
| Ding et al. | Efficient vanishing point detection method in unstructured road environments based on dark channel prior |
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 |
Granted publication date: 20180511 Termination date: 20200124 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |