CN102339388A - Method for identifying classification of image-based ground state - Google Patents
Method for identifying classification of image-based ground state Download PDFInfo
- Publication number
- CN102339388A CN102339388A CN2011101756084A CN201110175608A CN102339388A CN 102339388 A CN102339388 A CN 102339388A CN 2011101756084 A CN2011101756084 A CN 2011101756084A CN 201110175608 A CN201110175608 A CN 201110175608A CN 102339388 A CN102339388 A CN 102339388A
- Authority
- CN
- China
- Prior art keywords
- image
- histogram
- ground
- lbp
- state
- 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 method for identifying a classification of an image-based ground state. The method comprises the following steps: (1) preparing a ground image sample set; (2) acquiring a to-be-identified ground state image from an imaging device; (3) acquiring a feature vector of the to-be-identified ground state image; (4) respectively calculating a distance between the feature vector of the to-be-identified ground state image and the feature vector of each sample image; and (5) judging the ground state, specifically, (5.1) sorting out K minimum distances from the distances and recording the classification of the sample image corresponding to each distance, wherein K is a positive integer and (5.2) respectively counting the emerging times of the classification of the ground state of the corresponding sample image in the K distances, wherein the classification emerging for most times is confirmed as the classification of the to-be-identified ground state image. According to the method, the state of a ground image can be efficiently identified, thereby promoting the degree of automation and precision of ground state observation.
Description
Technical field
The invention belongs to Digital Image Processing and area of pattern recognition, be specifically related to a kind of method, be used for discerning automatically the state of ground classification of ground image based on image recognition state of ground kind.
Background technology
The state of ground is meant at observation time, near the surface condition terrain surface specifications without the maintenance nature of turning over the observation station, the consequence that particularly such as weather such as rain, snow, low temperature ground is caused.State of ground observation is the important component part of surface weather observation, is the required important step of dust and sand weather forecast, in development of the national economy service, important effect is arranged also simultaneously.
In " surface weather observation standard " that China Meteorological Administration published in 1979, just do not comprise the related content of state of ground observation.Along with China meteorological department expands the demand of service field and the development of surface weather observation new technology, more and more highlight state of ground importance of monitoring, in " the surface weather observation standard " of the up-to-date publication of 2003 China Meteorological Administration in the end of the year; The state of ground is classified as a chapter separately; The contents such as kind (two types, 20 kinds of situations), observation procedure of the state of ground have been introduced in detail, at present; Research about state of ground robotization identification aspect both at home and abroad almost is in blank state; So the observation of the state of ground of present stage also rests on the artificial observation, meteorologic observer's workload is big, and the accuracy of observed result is subject to observation person's experience level.If can utilize the digital picture recognition technology to discern the state on ground automatically, will reduce cost of human resources greatly, also can improve the automaticity and the precision of state of ground observation.
The LBP operator is a kind of operator of effective processing image texture; Not only theoretical simple easy to understand; And computation process is convenient and swift; Combine the proper vector of formation with the color characteristic of color histogram [Y.Gong, H.J.Zhang and T.C.Chua, An image database system with content capturing and fast image indexing abilities]; Can effectively identify the state of ground image of each classification, thereby make the identification of the state of ground can realize robotization.LBP (the Local Binary Pattern) texture [Timo Ojala, Matti Pietikainen.Multiresolution Gray-Scale and Rotation Invariant Texture Classification with Local Binary Patterns] that present widely used a kind of textural characteristics is an image.
Summary of the invention
The present invention provides a kind of state of ground kind recognition methods based on image, through the proper vector with the color characteristic composing images of image texture features and image, with χ
2Probability statistics are as the similarity measurement of proper vector; Can identify the state of ground image effectively in conjunction with the k nearest neighbor algorithm for pattern recognition; Automatically discern the state of ground classification in the state of ground image; To improve the automaticity and the precision of state of ground observation, overcome problems such as existing state of ground artificial observation method wastes time and energy, inefficiency.
A kind of state of ground kind recognition methods based on image of the present invention comprises the steps:
(1) image pattern collection preparation process, the identification of the state of ground are based on calculates that between image to be identified and the sample image similarity on characteristic is carried out, so at first need prepare the sample set of ground image, step is following:
(1.1) sample image obtaining step obtains several ground image samples under each kind, and writes down the affiliated classification of each ground image pattern, ask for each width of cloth sample image I (detailed process is for x, proper vector y):
(1.2) obtain the color histogram of sample image, the detailed process of obtaining color histogram is:
(1.2.1) image I (x, color space y) by the RGB color space conversion to the hsv color space IHsv (x, y);
(1.2.2) extract hsv color space hypograph IHsv (x, H y) (x, y) and V (x, y) two color components.
(1.2.3) with H (x, y) and V (x, y) 2 element quantizations become between several little chromatic zoneses, each minizone becomes between a histogrammic chromatic zones, the span of the color interval number N of each color component is [200,250];
(1.2.4) color histogram normalization;
(1.2.5) two normalized color histograms are according to the synthetic color histogram of the der group of H, V component, and this color histogram is the color feature vector of image;
(1.3) obtain the LBP histogram of sample image, obtain the histogrammic detailed process of LBP and be:
(1.3.1) to the sample image I of the state of ground (x y) carries out gray processing, obtain result images B (x, y);
(1.3.2) to gray-scale map B (x y) carries out the median filter smoothness of image denoising, obtain result images T (x, y);
(1.3.3) ask for image T (x respectively; Y) the LBP histogram under
and
operator; The synthetic histogram of three set of histograms; This histogram is image T (x, y) final LBP histogram.Ask for image T (x, y) the histogrammic detailed process of LBP under operator
is:
1. ask for T (x by pixel; Y) the LBP value of all pixels in; Obtain image T (x; The LBP characteristic pattern LBP of
operator y) (x, y);
2. ask for characteristic pattern LBP (x, histogram y),
3. LBP histogram normalization,
(1.4) be combined into a histogram to LBP histogram and color histogram, this histogram is the proper vector of image;
(1.5) structure sample set step, all sample images that dispose, the affiliated sample class number of specimen types number, image, the characteristics of image that write down every image are vectorial, and this record is the image pattern collection.
(2) ground image obtaining step to be identified obtains state of ground image to be identified from imaging device;
(3) proper vector obtaining step obtains the proper vector of ground image to be identified, and the THE ACQUISITION OF FEATURE VECTOR step is identical in method and the image pattern collection preparation process;
(4) similarity measurement calculation procedure; Sample class number, characteristics of image vector under the specimen types number of every image of reading images sample centralized recording, the image; Calculate characteristics of image vector to be identified then respectively and open the distance of sample image proper vector with each, characteristics of image vector to be identified with the computing formula of the distance of sample image proper vector is:
Wherein, S, M represent the histogram distribution of two images, S
bAnd M
bRepresent b interval probable value in the histogram of two images respectively, B representes histogrammic total interval number.
(5) state of ground decision steps, detailed process is:
(5.1) pick out in the middle of the distance calculated in the step (4) minimumly to K little distance of K, and write down each affiliated classification apart from pairing sample image, the value of K is set at [1,4];
(5.2) add up the number of times that classification occurs under the state of ground of corresponding sample image in K the distance respectively;
(5.3) the classification number that occurrence number is maximum is the affiliated classification of state of ground image to be identified.
The present invention provides a kind of state of ground kind recognition methods based on image; Be intended to discern automatically the state of ground classification in the state of ground image; To improve the automaticity and the precision of state of ground observation, overcome problems such as existing state of ground artificial observation method wastes time and energy, inefficiency.At present, for 4 types of state of ground images that collected (bare area, snowfield, dry and cracked, meadow), the discrimination of this invention is up to more than 97%.
Description of drawings
Fig. 1 is a kind of state of ground recognizer process flow diagram based on image;
Fig. 2 is dry and cracked state of ground image;
Fig. 3 is the dirt floor status image;
Fig. 4 is a sandy soil state of ground image;
Fig. 5 is an accumulated snow state of ground image;
Fig. 6 is the color histogram of Fig. 5;
Embodiment
The present invention is with the color characteristic of image texture features and the image characteristic as image, and the present invention is with the proper vector of the color characteristic composing images of image texture features and image, with χ
2Probability statistics are as the similarity measurement of proper vector, can identify the state of ground image effectively in conjunction with the nearest patterns recognizer, and a kind of treatment scheme of the state of ground recognizer based on image is as shown in Figure 1.
A kind of state of ground recognition methods based on image, step is following:
(1) image pattern collection preparation process, the identification of the state of ground are based on calculates that between image to be identified and the sample image similarity on characteristic is carried out, so at first need prepare the sample set of ground image, step is following:
(1.1) sample image obtaining step obtains several ground image samples under each kind, and writes down the affiliated classification of each ground image pattern, asks for each width of cloth sample image I (x, proper vector y).Fig. 1-Fig. 4 is 4 kinds of common state of ground images, and the process of asking for the sample image proper vector is following;
(1.2) obtain the color histogram of sample image, the detailed process of obtaining color histogram is:
(1.2.1) image I (to the hsv color space, available conversion formula as follows is realized conversion by the RGB color space conversion for x, color space y):
In the middle of formula, r, g, the gray level of b will normalize to interval [0,1] from interval [0,255] originally, max (r, g, b) expression r, g, the maximal value during b three numbers are worthwhile, min (r, g, b) expression r, g, the minimum value during b three numbers are worthwhile;
(1.2.2) extract hsv color space hypograph IHsv (x, H y) (x, y) and V (x, y) two color components.
(1.2.3) with H (x, y) and V (x, y) 2 components are divided between several little chromatic zoneses; Each minizone becomes a histogrammic interval; In this recognition methods, the span of the interval number N of each color component is [200,250]; In the view picture component image, the number of pixels in dropping between certain chromatic zones is H
i, the statistical value H of one group of pixel then
1, H
2..., H
NBe the color histogram of this component just.
(1.2.4) color histogram normalization, middle number of pixels shared ratio in entire image is h between certain chromatic zones
i:
h
i=H
i/(width*height)
Width and heigh are respectively image I (x, y) wide and high, h
1, h
2..., h
NBe normalized color histogram.
(1.2.5) two normalized color histograms are according to the synthetic color histogram of the der group of H, V component, and this color histogram is the color feature vector of image;
HV
1,HV
2,...HV
2N=H
1,H
2,...H
N,V
1,V
2,...,V
N
In following formula, HV
1, HV
3... HV
2N(its total dimension is 2N, H for x, color feature vector y) for image I
1, H
2..., H
N(promptly the preceding N dimension of color feature vector is H component color histogram, V for x, H component color histogram y) for image I
1, V
2..., V
N(promptly the back N dimension of color feature vector is V component color histogram for x, V component color histogram y) for image I.
(1.3) obtain the LBP histogram of sample image, the LBP operator that uses among the present invention does
With
The multiple dimensioned LBP operator that is combined to form,
Be illustrated in radius and be in the circular neighborhood of R and get P pixel, R, P are positive integer, u
2The expression equivalent formulations; Equivalent formulations is defined as: when the pairing binary number of certain partial binary pattern from 0 to 1 or from 1 to 0 has twice saltus step at most; The pairing scale-of-two of this partial binary pattern just is called an equivalent formulations class, does not satisfy other binary number systems of equivalent formulations and all claims the mixed mode class.
Obtaining the histogrammic detailed process of LBP is:
(1.3.1) to the sample image I of the state of ground (x y) carries out gray processing, obtain result images B (x, y);
B(x,y)=R(x,y)*0.299+G(x,y)*0.587+B(x,y)*0.114
In the following formula, and R (x, y), (x is y) with B (x, y) presentation video I (x, y) the R color component under the RGB color space, G color component and B color component respectively for G.
(1.3.2) (x y) carries out the median filter smoothness of image denoising, obtains result images T (x to gray-scale map B; Y); Compare with the image smoothing denoising method of linearity, the advantage of medium filtering is both to remove effectively noise, can keep edge of image and important detailed information preferably again; In practical implementation, the size of median filter window is preferably 3*3 or 5*5;
(1.3.3) ask for image T (x respectively; Y) the LBP histogram under
and
operator; The synthetic histogram of three set of histograms; This histogram is image T (x, y) final LBP histogram.Ask for image T (x, y) the histogrammic detailed process of LBP under operator
is:
1. for T (x, any pixel F (x in certain regional area in y)
c, y
c), be central point g with it
c, be P some g in the window of R to radius
0... g
P-1, with window center point gray-scale value g
cFor threshold value to other pixels g in the window
0... g
P-1Do binary conversion treatment, obtain the binary number of a P position, the pixel diverse location is carried out weighted sum, can obtain the LBP value of this window by following formula:
Ask for T (x by pixel; Y) the LBP value of all pixels in; Obtain image T (x; The LBP characteristic pattern LBP of
operator y) (x, y);
2. ask for characteristic pattern LBP (x, histogram y), the corresponding binary number of certain pixel value in the image, whether can find the solution according to following formula is the equivalent formulations class:
If the U (G that calculates
P) be less than or equal to 2, when promptly this binary number from 0 to 1 or from 1 to 0 has twice saltus step at most, then it is belonged to equivalent formulations, otherwise ownership mixed mode class.For
Operator, the number of equivalent formulations class are P
2-P+2, the number of mixed mode class is 1, promptly (x, histogram dimension y) is P to characteristic pattern LBP
2-P+3.(x, y), its pixel value span is by 2 for characteristic pattern LBP
PLevel is formed, and each pixel value is c
i(i=0,1 ..., 2
P-1).In entire image, if pixel value c
iBe equivalent formulations, then statistics has c
iThe number of pixels H of the equivalent formulations class of value
j(j=1,2 ..., P
2-P+2), if pixel value c
iBe the mixed mode class, then add up the number of pixels H of mixed mode class
j(j=P
2-P+3), the statistical value H of one group of pixel then
1, H
2..., H
j(j=1,2 ..., P
2-P+3) be this characteristic pattern LBP (x, histogram y) just.
3. LBP histogram normalization, the pixel number of promptly trying to achieve characteristic pattern account for image I (x, y) in the ratio of number of pixels:
h
i=H
i/(width*height)
Ask for image T (x; Y) after the LBP histogram under
and
operator, in such a way the synthetic histogram of three set of histograms:
In following formula,
For image T (x, final LBP histogram y),
(x is y) at operator for image T
Under the LBP histogram,
(x is y) at operator for image T
Under the LBP histogram,
(x is y) at operator for image T
Under the LBP histogram, P wherein
1=8, P
2=16, P
3=24.
(1.4) be combined into a histogram to LBP histogram and color histogram, this histogram is the proper vector of image;
In the following formula,
Be image I (x, proper vector y), HV
1, HV
2... HV
2NFor image I (x, color histogram y),
Be image I (x, LBP histogram y).
(1.5) structure sample set step, behind all sample images that dispose, sample class number, characteristics of image vector change record and are the image pattern collection under record specimen types number, the image.
(2) ground image obtaining step to be identified obtains state of ground image to be identified from imaging device;
(3) proper vector obtaining step obtains the proper vector of ground image to be identified, and the THE ACQUISITION OF FEATURE VECTOR step is identical in method and the image pattern collection preparation process;
(4) similarity measurement calculation procedure; Sample class number, characteristics of image vector under the specimen types number of every image of reading images sample centralized recording, the image; Calculate characteristics of image vector to be identified then respectively and open the distance of sample image proper vector with each, characteristics of image vector to be identified with the computing formula of the distance of sample image proper vector is:
Wherein, S, M represent the histogram distribution of two images, S
bAnd M
bRepresent b interval probable value in the histogram of two images respectively, B representes histogrammic total interval number.
(5) state of ground decision steps, detailed process is:
(5.1) pick out in the middle of the distance calculated in the step (4) minimumly to K little distance of K, and write down each affiliated classification apart from pairing sample image, the value of K is set at [1,4];
(5.2) add up the number of times that classification occurs under the state of ground of corresponding sample image in K the distance respectively;
(5.3) the classification number that occurrence number is maximum is the affiliated classification of state of ground image to be identified.
Claims (6)
1. the state of ground kind recognition methods based on image comprises the steps:
(1) preparation ground image sample set
(x y), writes down each sample image I (x, the kind under y), and try to achieve each sample image I (x, proper vector y) to obtain several sample images I;
(2) obtain state of ground image to be identified from imaging device;
(3) proper vector of the said state of ground image to be identified of acquisition;
(4) similarity measurement calculates, and promptly calculates the proper vector of said state of ground image to be identified and the distance of the proper vector of each sample image respectively, and wherein, the computing formula of said distance is:
Wherein, S, M represent the histogram distribution of sample image and state of ground image to be identified, S
bAnd M
bRepresent b interval probable value in the histogram of sample image and state of ground image to be identified respectively, B representes histogrammic total interval number;
(5) state of ground decision steps, detailed process is:
(5.1) pick out K minimum in said distance distance, and write down the kind of the pairing sample image of each distance, wherein K is a positive integer;
(5.2) add up the number of times that the kind of the state of ground of corresponding sample image in said K the distance occurs respectively, the kind that wherein occurrence number is maximum is promptly confirmed as the kind of state of ground image to be identified.
2. method according to claim 1 is characterized in that, the concrete procurement process of the proper vector of said image is:
(1.1) obtain the color histogram of image;
(1.2) obtain the LBP histogram of image;
(1.3) be combined into a histogram to LBP histogram and color histogram, this histogram is the proper vector of image.
3. method according to claim 2 is characterized in that, the color histogram acquisition process of said image is:
(1.1.1) the color space of image by the RGB color space conversion to the hsv color space;
(1.1.2) extract hsv color space hypograph two color component H (x, y) and V (x, y);
(1.1.3) each color component is divided into a plurality of minizones, each minizone becomes a histogrammic interval;
(1.1.4) color histogram normalization, the number of pixels in promptly trying to achieve between each chromatic zones accounts for the ratio of entire image pixel count;
(1.1.5) normalized color histogram is synthesized a color histogram according to the der group of said two color components.
4. according to claim 2 or 3 described methods, it is characterized in that the said LBP histogram detailed process of obtaining image is:
(1.2.1) image is carried out gray processing, obtain gray-scale map;
(1.2.2) said gray-scale map is carried out the median filter smoothness of image denoising, obtain result images;
(1.2.3) ask for the LBP histogram of result images under
and
operator respectively; The synthetic histogram of three set of histograms, this histogram is the final LBP histogram of result images.
5. method according to claim 4; It is characterized in that, ask for the LBP histogrammic detailed process of result images under operator
and be:
At first; Said result images is asked for the wherein LBP value of all pixels by pixel; Obtain the LBP characteristic pattern LBP (x of
operator of result images; Y), wherein
is illustrated in the circular neighborhood that radius is R and gets the LBP value that P pixel asked for central pixel point;
Secondly, ask for characteristic pattern LBP (x, histogram y);
6. according to the described method of one of claim 1-5, it is characterized in that said K span is [1,4].
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201110175608 CN102339388B (en) | 2011-06-27 | 2011-06-27 | Method for identifying classification of image-based ground state |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201110175608 CN102339388B (en) | 2011-06-27 | 2011-06-27 | Method for identifying classification of image-based ground state |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102339388A true CN102339388A (en) | 2012-02-01 |
CN102339388B CN102339388B (en) | 2012-12-19 |
Family
ID=45515110
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 201110175608 Expired - Fee Related CN102339388B (en) | 2011-06-27 | 2011-06-27 | Method for identifying classification of image-based ground state |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102339388B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105512689A (en) * | 2014-09-23 | 2016-04-20 | 苏州宝时得电动工具有限公司 | Lawn identification method based on images, and lawn maintenance robot |
CN108346136A (en) * | 2018-01-16 | 2018-07-31 | 杭州电子科技大学 | Industrial reaction kettle liquid face separation method based on histogram Yu LBP Fusion Features |
CN109044203A (en) * | 2018-06-28 | 2018-12-21 | 芜湖泰领信息科技有限公司 | Sweeper automatic cleaning method and intelligent sweeping machine |
CN110245683A (en) * | 2019-05-13 | 2019-09-17 | 华中科技大学 | The residual error relational network construction method that sample object identifies a kind of less and application |
CN112488050A (en) * | 2020-12-16 | 2021-03-12 | 安徽大学 | Color and texture combined aerial image scene classification method and system |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101876993A (en) * | 2009-11-26 | 2010-11-03 | 中国气象科学研究院 | Method for extracting and retrieving textural features from ground digital nephograms |
-
2011
- 2011-06-27 CN CN 201110175608 patent/CN102339388B/en not_active Expired - Fee Related
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101876993A (en) * | 2009-11-26 | 2010-11-03 | 中国气象科学研究院 | Method for extracting and retrieving textural features from ground digital nephograms |
Non-Patent Citations (2)
Title |
---|
JITZE P.VAN DER MEULEN EL AT.: "《Present Weather Observing Systems: One year of experience and comparison with human observations》", 《WMO TECHNICAL CONFERENCE ON INSTRUMENTS AND METHODS OF OBSERVATION》, 15 May 1992 (1992-05-15) * |
岳平等: "《纹理分析法识别静止卫星红外云图和监测汛期强对流天气系统》", 《干旱气象》, vol. 23, no. 2, 30 June 2005 (2005-06-30) * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105512689A (en) * | 2014-09-23 | 2016-04-20 | 苏州宝时得电动工具有限公司 | Lawn identification method based on images, and lawn maintenance robot |
CN108346136A (en) * | 2018-01-16 | 2018-07-31 | 杭州电子科技大学 | Industrial reaction kettle liquid face separation method based on histogram Yu LBP Fusion Features |
CN109044203A (en) * | 2018-06-28 | 2018-12-21 | 芜湖泰领信息科技有限公司 | Sweeper automatic cleaning method and intelligent sweeping machine |
CN110245683A (en) * | 2019-05-13 | 2019-09-17 | 华中科技大学 | The residual error relational network construction method that sample object identifies a kind of less and application |
CN110245683B (en) * | 2019-05-13 | 2021-07-27 | 华中科技大学 | Residual error relation network construction method for less-sample target identification and application |
CN112488050A (en) * | 2020-12-16 | 2021-03-12 | 安徽大学 | Color and texture combined aerial image scene classification method and system |
Also Published As
Publication number | Publication date |
---|---|
CN102339388B (en) | 2012-12-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104112370B (en) | Parking lot based on monitoring image intelligent car position recognition methods and system | |
CN102339388B (en) | Method for identifying classification of image-based ground state | |
CN101793501B (en) | Transmission line ice coating status detection method based on image | |
CN100555325C (en) | A kind of image interfusion method based on wave transform of not sub sampled contour | |
US8396293B1 (en) | Recognizing geometrically salient objects from segmented point clouds using strip grid histograms | |
CN101599120B (en) | Identification method of remote sensing image building | |
CN109389163B (en) | Unmanned aerial vehicle image classification system and method based on topographic map | |
Awrangjeb et al. | Improved building detection using texture information | |
CN108680833B (en) | Composite insulator defect detection system based on unmanned aerial vehicle | |
CN101807345A (en) | Traffic jam judging method based on video detection technology | |
CN104134080A (en) | Method and system for automatically detecting roadbed collapse and side slope collapse of road | |
Huang et al. | A top-down strategy for buildings extraction from complex urban scenes using airborne LiDAR point clouds | |
CN110070545B (en) | Method for automatically extracting urban built-up area by urban texture feature density | |
CN107993488A (en) | A kind of parking stall recognition methods, system and medium based on fisheye camera | |
CN109635722B (en) | Automatic identification method for high-resolution remote sensing image intersection | |
CN103528534A (en) | Image monitoring based method for detecting thickness of icing on power transmission line | |
CN111222511B (en) | Infrared unmanned aerial vehicle target detection method and system | |
CN103324958B (en) | Based on the license plate locating method of sciagraphy and SVM under a kind of complex background | |
CN103793709A (en) | Cell recognition method and device, and urine analyzer | |
CN114511770A (en) | Road sign plate identification method | |
CN110321855A (en) | A kind of greasy weather detection prior-warning device | |
CN111008967B (en) | Insulator RTV coating defect identification method | |
CN114662526A (en) | Forest land ecological monitoring method based on remote sensing data and multi-temporal SAR (synthetic aperture radar) images | |
CN103996266B (en) | Tailing reservoir safety monitoring device based on video analysis | |
Bashit et al. | Analysis of built-up land spatial patterns using multitemporal satellite imagery in Pekalongan city |
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 |
Granted publication date: 20121219 Termination date: 20160627 |
|
CF01 | Termination of patent right due to non-payment of annual fee |