CN104966102A - Typhoon detection method based on satellite image - Google Patents
Typhoon detection method based on satellite image Download PDFInfo
- Publication number
- CN104966102A CN104966102A CN201510348983.2A CN201510348983A CN104966102A CN 104966102 A CN104966102 A CN 104966102A CN 201510348983 A CN201510348983 A CN 201510348983A CN 104966102 A CN104966102 A CN 104966102A
- Authority
- CN
- China
- Prior art keywords
- satellite image
- typhoon
- hog feature
- detected
- detection method
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to the technical field of weather forecasting, and provides a typhoon detection method based on a satellite image. The method comprises steps: in a training state, training is carried out via a support vector machine according to Hog (Histograms of Oriented Gradients) features of a positive sample and Hog (Histograms of Oriented Gradients) features of a negative sample so as to generate a template for judging and telling the positive sample and the negative sample according to Hog features; in a typhoon detection stage, Hog features of a to-be-detected satellite image are extracted, and the template trained by the support vector machine is compared with the Hog features of the to-be-detected satellite image so as to detect whether a typhoon exists in the to-be-detected satellite image. According to the typhoon detection method based on the satellite image, the typhoon detection process is simplified, the workload of a weather staff is reduced, and the method is of great importance for analyzing and forecasting the typhoon path in advance.
Description
Technical field
The invention belongs to the technical field of weather prognosis, be specifically related to a kind of typhoon detection method based on satellite image.
Background technology
Strictly speaking, typhoon is a kind of intensity of tropical cyclone, and table 1 is the classification of tropical cyclone according to its intensity,
Table 1
| Intensity title | Central wind (level) | Bottom center wind speed (meter per second) |
| Tropical depression | 6～7 | 10.8～17.1 |
| Tropical storm | 8～9 | 17.2～24.4 |
| Severe tropical storm | 10～11 | 24.5～32.6 |
| Typhoon | 12～13 | 32.7～41.4 |
| Violent typhoon | 14～15 | 41.5～50.9 |
| Super Typhoon | 16 | >51 |
As can be seen from Table 1, when the central wind of tropical cyclone continues to reach 12-13 level, when bottom center wind speed is 32.7 meters-41.4 meters, meteorology is called typhoon.But usually also the tropical cyclone of varying strength is referred to as typhoon, namely the typhoon in the technical program refers to the tropical cyclone of varying strength.
In the whole world, have two typhoon growth centres, lay respectively at Western Pacific and the Atlantic, China is positioned at west side, Western Pacific, is that global typhoon generation quantity is maximum here, the sea area that intensity is the strongest, the whole world have an appointment more than 36% typhoon all concentrate on this.The every annual of China is subject to the attack of 9 ~ 10 typhoons, occupies first place in the world, and is to endanger one of country the most serious by typhoon in the world.Typhoon causes serious threat to the people's life of southeastern coast provinces and cities of China and property safety, brings about great losses to industrial and agricultural production and communications and transportation.
Typhoon track is forecast with unerring accuracy and can make the arrival of southeastern coast provinces and cities of China in advance pre-typhoon protection, to reduce the loss because typhoon causes.Be the precondition of accurate forecast typhoon track to the accurate location of center of typhoon, and whether detection have typhoon existence to be the precondition of carrying out Typhoon center location.
Due to typhoon between its active stage by way of marine site widely, scope traditional is at sea rare, therefore satellite becomes the Main Means estimating typhoon position and intensity, the remote sensing images in the wide marine site that it can provide other direct measurement means to relate to, and typhoon can be monitored from the overall process being generated to extinction.Therefore, how to detect the typhoon region in satellite image, for typhoon forecast and analyze tool be of great significance.
Typhoon in satellite image has following feature: first, and the yardstick of typhoon is different, and large typhoon diameter can reach 1200 kms, and little diameter may less than 400 kms, and the size that they are reflected on satellite image is different; Secondly, typhoon comes in every shape, and different typhoons has different forms, even if same typhoon also has different forms in different developing periods; Finally, typhoon identification difficulty, for the image of a pedestrian, even if only see nose or eyes, also can differentiate it is the part of face, and in nascent phase of typhoon and the phase of extinction, is difficult on individual satellite image, find out the very weak typhoon of intensity.
Whether current detection satellite image there is typhoon cloud system mainly still by artificial method, meteorologist is needed rule of thumb to analyze on satellite image whether have typhoon cloud system one by one, and the generation detecting new typhoon then needs weather scientist to analyze the satellite image of continuous several days just can reach a conclusion, analytic process is loaded down with trivial details, waste time and energy.If can automatically detect typhoon cloud system on satellite image, for simplifying the process detecting typhoon, the workload reducing meteorologist, carrying out Typhoon Analysis and accurate forecast typhoon track tool is of great significance ahead of time.
Summary of the invention
In order to solve the problem detecting at present and whether satellite image has the method for typhoon cloud system to waste time and energy, the present invention proposes a kind of typhoon detection method based on satellite image, can automatically detect typhoon cloud system on satellite image, reach the process, the minimizing workload of meteorologist and the object of accurate forecast typhoon track that simplify and detect typhoon.
The typhoon detection method that the present invention is based on satellite image comprises the following steps:
One, the training stage
(1) there is the picture region of typhoon as positive sample using what choose on satellite image, extract the Hog feature of this positive sample;
(2) using the picture region without typhoon chosen on satellite image as negative sample, the Hog feature of this negative sample is extracted;
(3) the Hog feature of support vector machine negative sample according to the Hog characteristic sum of described positive sample is trained, and can judge to distinguish according to Hog feature the template of positive sample and negative sample to generate;
Two, typhoon detection-phase
(1) satellite image to be detected is inputted;
(2) scaling is carried out to described satellite image to be detected;
(3) the Hog feature of described satellite image to be detected is extracted;
(4) the Hog feature of the template of training out by described support vector machine and described satellite image to be detected contrasts, when the Hog feature of described satellite image to be detected is consistent with the Hog feature of described positive sample, then there is typhoon cloud system in described satellite image to be detected; When the Hog feature of described satellite image to be detected and the Hog feature of described negative sample consistent time, then there is no typhoon cloud system in described satellite image to be detected.
The step (1) of described typhoon detection-phase also comprises: check in satellite image to be detected whether lose sweep trace, when losing a sweep trace in satellite image to be detected, made up by the average of the sweep trace before and after the sweep trace of this loss.
In the step (1) of described training stage and step (2), the size of described positive sample and described negative sample is 320*320 pixel.
When extracting Hog feature in the step (3) of the step (1) of described training stage, step (2) and described typhoon detection-phase, the size of window is 320*320 pixel.
When extracting Hog feature in the step (3) of the step (1) of described training stage, step (2) and described typhoon detection-phase, the length of side of block is 2 times of the cell factory length of side.
When extracting Hog feature in the step (3) of the step (1) of described training stage, step (2) and described typhoon detection-phase, the resolution of satellite image is 16 kms.
When extracting Hog feature in the step (3) of the step (1) of described training stage, step (2) and described typhoon detection-phase, the length of side of block is 1/5th of the window length of side.
When extracting Hog feature in the step (3) of the step (1) of described training stage, step (2) and described typhoon detection-phase, sliding step equals the length of side of cell factory.
The typhoon detection method that the present invention is based on satellite image has following beneficial effect:
When using the typhoon detection method that the present invention is based on satellite image to detect whether satellite image have typhoon cloud system, only need input satellite image to be detected, the present invention detects satellite to be detected, to detect whether there is typhoon cloud system from view picture satellite image fast and accurately, if there is typhoon cloud system, then be partitioned into the picture region of typhoon cloud system, the process of whole detection is automatically run, decrease meteorologist finds typhoon labour intensity from a large amount of satellite cloud pictures to a great extent, add work efficiency, be conducive to the typhoon finding as early as possible to affect China and other area.The typhoon detection method that the present invention is based on satellite image can detect in satellite image whether have typhoon cloud system automatically, simplify detect typhoon process, decrease the workload of meteorologist, for carrying out Typhoon Analysis ahead of time and accurate forecast typhoon track tool is of great significance.
Accompanying drawing explanation
Fig. 1 is the principle schematic of the Hog feature extracting satellite image.
Embodiment
Technical scheme of the present invention is introduced below in conjunction with accompanying drawing.
The typhoon detection method that the present invention is based on satellite image comprises the following steps:
One, the training stage
(1) there is the picture region of typhoon as positive sample using what choose on satellite image, extract gradient orientation histogram (Histogram of Oriented Gradients, hereinafter referred to as the Hog) feature of this positive sample;
(2) using the picture region without typhoon chosen on satellite image as negative sample, the Hog feature of this negative sample is extracted;
(3) support vector machine (Support Vector Machine, hereinafter referred to as SVM) according to the Hog characteristic sum of described positive sample, the Hog feature of negative sample is trained, and can judge to distinguish according to Hog feature the template of positive sample and negative sample to generate;
Two, typhoon detection-phase
(1) situation whether having in satellite image to be detected and lose sweep trace is detected;
(2) satellite image to be detected is inputted;
(3) scaling is carried out to satellite image to be detected;
(4) the Hog feature of satellite image to be detected is extracted;
(5) the Hog feature of the template of training out by support vector machine and satellite image to be detected contrasts, when the Hog feature of described satellite image to be detected is consistent with the Hog feature of described positive sample, then there is typhoon cloud system in described satellite image to be detected; When the Hog feature of described satellite image to be detected and the Hog feature of described negative sample consistent time, then there is no typhoon cloud system in described satellite image to be detected.
Introduce the typhoon detection method that the present invention is based on satellite image below in detail.
One, the training stage
(1) there is the picture region of typhoon as positive sample using what choose on satellite image, extract the Hog feature of this positive sample;
Satellite image in this step can be the satellite image having typhoon location information former years, on this satellite image centered by center of typhoon, chooses the picture region that size is 320*320 pixel, using this picture region as positive sample.Because the resolution of the satellite image used is up to 4 kms, the maximum gauge of typhoon is about 1000 kms, and size is that the picture region of 320*320 pixel can cover whole typhoon cloud system.Description is had below the method for extraction Hog feature.
(2) using the picture region without typhoon chosen on satellite image as negative sample, the Hog feature of this negative sample is extracted;
The satellite image used in this step can be identical with the satellite image in step (2), but this step in satellite image, chooses picture region without typhoon as negative sample, and the size of negative sample is also 320*320 pixel.Description is had below the method for extraction Hog feature.
(3) support vector machine is trained according to the Hog feature of the Hog characteristic sum negative sample of positive sample, can judge to distinguish according to Hog feature the template of positive sample and negative sample to generate.
Two, typhoon detection-phase
(1) situation whether having in satellite image to be detected and lose sweep trace is detected, for the satellite image losing a sweep trace, can be made up by the average of the sweep trace before and after the position of the sweep trace of loss, can not typhoon detection be carried out for the satellite image losing multi-strip scanning line owing to lacking data.Whether have sweep trace loss can pass through detect in satellite image whether have 0 of full line or the extreme point of 255, because the bright temperature scope of normal satellite image can not reach extreme value, the extreme point of full line must cause owing to losing line if detecting satellite image;
(2) satellite image to be detected is inputted;
(3) scaling is carried out to satellite image to be detected, to adjust the resolution of satellite image, the resolution of satellite image can be 4 kms, 8 kms, 16 kms, 32 kms and 64 kms, preferably, is 16 kms by the resolution adjustment of satellite image to be detected;
(4) the Hog feature of satellite image to be detected is extracted;
When extracting Hog feature, first fix the window that a size is 320*320 pixel, extract the Hog feature of this window, then this window slides on satellite image to be detected, to travel through whole satellite image to be detected, obtain multiple window like this, extract the Hog feature of each window respectively.
As shown in Figure 1, when extracting the feature of each window 10, the first step, carries out the standardization of color space to satellite image; Second step, the length of side of the cell factory (cell) 12 chosen, the length of side of the cell factory (cell) 12 such as chosen is 40 pixels, four cell factory 12 form a block 11 (block), and the length of side of block 11 is 2 times of cell factory 12 length of side, in cell factory 12, evenly choose 3*3 point, extract the Hog feature of this 3*3 point, wherein, the direction of extracting gradient during Hog feature is that an interval (bin) divides with 20 degree; 3rd step, couples together the Hog feature obtaining block 11 by the Hog feature of four cell factory 12; 4th step, by block 11 with 40 pixels for sliding step (block stride) carries out horizontal and vertical slip in window 10, to travel through whole window 10, block 11 often slides once, all produce a new block, the Hog feature of these several blocks 11 is coupled together the Hog feature just obtaining window 10.Wherein, sliding step equals the length of side of cell factory 12, can improve the efficiency extracting Hog feature like this, if sliding step is less than the length of side of cell factory 12, impact can extract the efficiency of Hog feature because repetition rate is too high.
The step (1) of training stage is identical with the method that this step extracts Hog feature with the method extracting Hog feature in step (2);
(5) the Hog feature of each window of the template of training out by support vector machine and satellite image to be detected contrasts, when the Hog feature of certain window of satellite image to be detected and the Hog feature of positive sample consistent time, then have typhoon cloud system in this window; When the Hog feature of certain window of satellite image to be detected and the Hog feature of negative sample consistent time, then there is no typhoon cloud system in this window, successively the template that each window in satellite image to be detected and support vector machine train out is contrasted, to determine whether there is typhoon cloud system in window, thus determine in whole satellite image, whether there is the window comprising typhoon cloud system, namely whether there is typhoon cloud system in whole satellite image.
Here is the experiment utilizing the typhoon detection method that the present invention is based on satellite image to carry out typhoon detection.
The typhoon that northwest Pacific or the South Sea generate, full-fledgedly probably can have influence on China's Coastal Areas afterwards, and therefore will pay close attention in the nascent phase of typhoon, nascent phase typhoon is commonly referred to tropical depression or tropical storm on meteorology; For the typhoon that other areas generate, owing to substantially can not impact China, therefore only have and develop into the prosperous and powerful phase to it, and may cause other areas and just to need during major disaster to pay close attention to, prosperous and powerful phase typhoon is commonly referred to severe tropical storm, typhoon, violent typhoon or Super Typhoon on meteorology.
Detection to nascent phase typhoon:
Training sample set: training sample takes from the satellite image that 2011 have locating information, intensity is totally 395 of tropical depression and tropical storm, wherein can download on the net from Chinese satellite remote sensing date and errorless totally 370 of the quality of data, obtain 370 positive samples of training that size is 320*320 pixel thus.Had other positions of the satellite image of locating information in 2011, random selecting size is the image totally 3909 of 320*320 pixel simultaneously, as training negative sample.
Test sample book collection: test sample book takes from the satellite image of locating information in 2012, intensity is totally 473 of tropical depression and tropical storm, wherein can download on the net from Chinese satellite remote sensing date and errorless totally 468 of the quality of data, obtain 468 test sample books of size 320*320 pixel thus.There were other positions of the satellite image of locating information simultaneously in 2012, get the image totally 1000 that size is 320*320 pixel at random, also as detecting sample.
Test with above-mentioned training sample set and test sample book collection, table 2 is select different window sizes and cell factory size, when the block length of side is different ratio from the length of side of cell factory, and the loss of testing result:
Table 2
Test with above-mentioned training sample set and test sample book collection, table 3 is select different window sizes and cell factory size, when the block length of side is different ratio from the length of side of cell factory, and the false drop rate of testing result:
Table 3
As can be seen from Table 2, when selecting different window sizes and cell factory size, when the ratio of the length of side of the block length of side and cell factory is 2, loss is minimum, and when the ratio of the length of side of the block length of side and cell factory is 1,3 or 4, loss is all higher.As can be seen from Table 3, when selecting different window sizes and cell factory size, the ratio of the length of side of the block length of side and cell factory affects very little on false drop rate.So the ratio selecting the length of side of the block length of side and cell factory is 2, when not making false drop rate raise, can obtain lower loss.
Test with above-mentioned training sample set and test sample book collection, the resolution that table 4 is satellite image to be detected is respectively 4 kms, 8 kms, 16 kms, 32 kms and 64 kms, when the ratio of the block length of side and the window length of side is respectively 0.1,0.2,0.3,0.4 and 0.5, the loss of testing result:
Table 4
The resolution that table 5 is satellite image to be detected is respectively 4 kms, 8 kms, 16 kms, 32 kms and 64 kms, when the ratio of the block length of side and the window length of side is respectively 0.1,0.2,0.3,0.4 and 0.5, and the false drop rate of experimental result:
Table 5
As can be seen from table 4 and table 5, along with the reduction of resolution, false drop rate presents the trend of increase, but loss presents the trend of reduction substantially, can also find out that the ratio along with the block length of side and the window length of side increases, loss presents the trend of growth, the target detected for nascent phase typhoon reduces loss as far as possible but allows suitable false drop rate, therefore choosing resolution is 16 kms, the ratio of the block length of side and the window length of side is 0.2, and resolution is 16 kms, when the ratio of the block length of side and the window length of side is 0.2, false drop rate 1.6%, loss 15.81%.
Detection to prosperous and powerful phase typhoon:
Training sample set: training sample takes from the satellite image that 2011 have locating information, prosperous and powerful phase typhoon totally 271, wherein can download on the net from Chinese satellite remote sensing date and errorless totally 264 of the quality of data, obtain 264 positive samples of training that size is 320*320 pixel thus.Had other positions of the satellite image of locating information in 2011, random selecting size is the image totally 3909 of 320*320 pixel simultaneously, as training negative sample.
Test sample book collection: test sample book takes from the satellite image that 2012 have locating information, intensity is higher than totally 375 of severe tropical storm, wherein can download on the net from Chinese satellite remote sensing date and errorless totally 370 of the quality of data, obtain 370 test sample books that size is 320*320 pixel thus.There were other positions of the satellite image of locating information simultaneously in 2012, get the image totally 1000 that size is 320*320 pixel at random, also as detecting sample.
Extract the major parameter of Hog feature: four cell factory form a block 11, and the block length of side is 2 times of the cell factory length of side, the ratio of the block length of side and the window length of side is 0.2, and the resolution of satellite image is 16 kms, testing result: loss is 14.32%, false drop rate is 1.50%.
The typhoon detection method that the present invention is based on satellite image can detect in satellite image whether have typhoon cloud system automatically, detects that the process speed of typhoon is fast, efficiency is high, precision is high, practical and time saving and energy saving.
Claims (8)
1., based on a typhoon detection method for satellite image, it is characterized in that, the method comprises the following steps:
One, the training stage
(1) there is the picture region of typhoon as positive sample using what choose on satellite image, extract the Hog feature of this positive sample;
(2) using the picture region without typhoon chosen on satellite image as negative sample, the Hog feature of this negative sample is extracted;
(3) the Hog feature of support vector machine negative sample according to the Hog characteristic sum of described positive sample is trained, and can judge to distinguish according to Hog feature the template of positive sample and negative sample to generate;
Two, typhoon detection-phase
(1) satellite image to be detected is inputted;
(2) scaling is carried out to described satellite image to be detected;
(3) the Hog feature of described satellite image to be detected is extracted;
(4) the Hog feature of the template of training out by described support vector machine and described satellite image to be detected contrasts, when the Hog feature of described satellite image to be detected is consistent with the Hog feature of described positive sample, then there is typhoon cloud system in described satellite image to be detected; When the Hog feature of described satellite image to be detected and the Hog feature of described negative sample consistent time, then there is no typhoon cloud system in described satellite image to be detected.
2. the typhoon detection method based on satellite image according to claim 1, it is characterized in that, the step (1) of described typhoon detection-phase also comprises: check in satellite image to be detected whether lose sweep trace, when losing a sweep trace in satellite image to be detected, made up by the average of the sweep trace before and after the sweep trace of this loss.
3. the typhoon detection method based on satellite image according to claim 1 and 2, is characterized in that, in the step (1) of described training stage and step (2), the size of described positive sample and described negative sample is 320*320 pixel.
4. the typhoon detection method based on satellite image according to claim 1 and 2, it is characterized in that, when extracting Hog feature in the step (3) of the step (1) of described training stage, step (2) and described typhoon detection-phase, the size of window is 320*320 pixel.
5. the typhoon detection method based on satellite image according to claim 1 and 2, it is characterized in that, when extracting Hog feature in the step (3) of the step (1) of described training stage, step (2) and described typhoon detection-phase, the length of side of block is 2 times of the cell factory length of side.
6. the typhoon detection method based on satellite image according to claim 1 and 2, it is characterized in that, when extracting Hog feature in the step (3) of the step (1) of described training stage, step (2) and described typhoon detection-phase, the resolution of satellite image is 16 kms.
7. the typhoon detection method based on satellite image according to claim 1 and 2, it is characterized in that, when extracting Hog feature in the step (3) of the step (1) of described training stage, step (2) and described typhoon detection-phase, the length of side of block is 1/5th of the window length of side.
8. the typhoon detection method based on satellite image according to claim 1 and 2, it is characterized in that, when extracting Hog feature in the step (3) of the step (1) of described training stage, step (2) and described typhoon detection-phase, sliding step equals the length of side of cell factory.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510348983.2A CN104966102A (en) | 2015-06-23 | 2015-06-23 | Typhoon detection method based on satellite image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510348983.2A CN104966102A (en) | 2015-06-23 | 2015-06-23 | Typhoon detection method based on satellite image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104966102A true CN104966102A (en) | 2015-10-07 |
Family
ID=54220137
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510348983.2A Pending CN104966102A (en) | 2015-06-23 | 2015-06-23 | Typhoon detection method based on satellite image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104966102A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108364097A (en) * | 2018-02-07 | 2018-08-03 | 国家海洋局北海预报中心 | Based on the typhoon cloud system prediction technique for generating confrontation network |
| GB2559687A (en) * | 2017-02-08 | 2018-08-15 | Ford Global Tech Llc | Tornado detection systems and methods |
| CN108919384A (en) * | 2018-03-26 | 2018-11-30 | 宁波市水利水电规划设计研究院 | It is a kind of based on the typhoon track DATA PROCESSING IN ENSEMBLE PREDICTION SYSTEM method for estimating deviation |
| CN112801227A (en) * | 2021-04-06 | 2021-05-14 | 航天宏图信息技术股份有限公司 | Typhoon identification model generation method, device, equipment and storage medium |
| CN114740550A (en) * | 2022-06-14 | 2022-07-12 | 广东海洋大学 | Intelligent recognition early warning method and system for continuous storm events |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102902983A (en) * | 2012-09-21 | 2013-01-30 | 博康智能网络科技股份有限公司 | Taxi recognition method based on support vector machine |
-
2015
- 2015-06-23 CN CN201510348983.2A patent/CN104966102A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102902983A (en) * | 2012-09-21 | 2013-01-30 | 博康智能网络科技股份有限公司 | Taxi recognition method based on support vector machine |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2559687A (en) * | 2017-02-08 | 2018-08-15 | Ford Global Tech Llc | Tornado detection systems and methods |
| CN108364097A (en) * | 2018-02-07 | 2018-08-03 | 国家海洋局北海预报中心 | Based on the typhoon cloud system prediction technique for generating confrontation network |
| CN108919384A (en) * | 2018-03-26 | 2018-11-30 | 宁波市水利水电规划设计研究院 | It is a kind of based on the typhoon track DATA PROCESSING IN ENSEMBLE PREDICTION SYSTEM method for estimating deviation |
| CN112801227A (en) * | 2021-04-06 | 2021-05-14 | 航天宏图信息技术股份有限公司 | Typhoon identification model generation method, device, equipment and storage medium |
| CN112801227B (en) * | 2021-04-06 | 2021-09-28 | 航天宏图信息技术股份有限公司 | Typhoon identification model generation method, device, equipment and storage medium |
| CN114740550A (en) * | 2022-06-14 | 2022-07-12 | 广东海洋大学 | Intelligent recognition early warning method and system for continuous storm events |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104036239B (en) | Fast high-resolution SAR (synthetic aperture radar) image ship detection method based on feature fusion and clustering | |
| CN104966102A (en) | Typhoon detection method based on satellite image | |
| CN102252623B (en) | Measurement method for lead/ground wire icing thickness of transmission line based on video variation analysis | |
| CN105785411B (en) | A kind of abnormal track-detecting method based on region division | |
| CN103544296B (en) | Adaptive intelligent integration detection method of radar range extension target | |
| CN102279973B (en) | Sea-sky-line detection method based on high gradient key points | |
| CN110645961B (en) | Forest resource dynamic change detection method based on remote sensing and NDVI | |
| CN103984862A (en) | Multielement remote sensing information coordinated snow cover parameter inversion method | |
| CN105719283A (en) | Road surface crack image detection method based on Hessian matrix multi-scale filtering | |
| CN109635713A (en) | The shadow region glacier recognition methods of plateau mountainous region | |
| CN102567726B (en) | Technology for automatically extracting floating ice in polar ice sheet edge regions | |
| CN102129566A (en) | Method for identifying rainstorm cloud cluster based on stationary meteorological satellite | |
| CN108154271A (en) | A kind of surface air temperature method of quality control based on spatial coherence and surface fitting | |
| Li et al. | Enhanced automatic root recognition and localization in GPR images through a YOLOv4-based deep learning approach | |
| CN104199009B (en) | A kind of radar image clutter suppression method based on time domain specification | |
| CN103487844B (en) | Intelligent coronal mass emission (CME) event observation procedure | |
| CN113902792A (en) | Building height detection method and system based on improved RetinaNet network and electronic equipment | |
| CN109269987A (en) | A kind of qualitative estimating and measuring method of salt lake water body salinity based on SPOT5 remotely-sensed data | |
| CN102445712A (en) | Character window weighting related spectrum matching method facing rocks and minerals | |
| CN109405809B (en) | Method and system for detecting flood water depth of transformer substation | |
| CN116503750A (en) | Large-range remote sensing image rural block type residential area extraction method and system integrating target detection and visual attention mechanisms | |
| Fang | Prediction and analysis of urban heat island effect in Dangshan by remote sensing | |
| CN111639623B (en) | Multi-scale ship-enterprise scene recognition and extraction method combined with space distance constraint | |
| CN102109456B (en) | Method for calculating high-suspension sediment concentration of Landsat satellite sensors | |
| Zhang et al. | Detection of land use change in urban agglomeration using Sentinel-1 SAR data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20151007 |