CN108520231A - A kind of analysis system and method for intelligence wetland landscape evolution process - Google Patents
A kind of analysis system and method for intelligence wetland landscape evolution process Download PDFInfo
- Publication number
- CN108520231A CN108520231A CN201810301492.6A CN201810301492A CN108520231A CN 108520231 A CN108520231 A CN 108520231A CN 201810301492 A CN201810301492 A CN 201810301492A CN 108520231 A CN108520231 A CN 108520231A
- Authority
- CN
- China
- Prior art keywords
- wetland
- module
- data
- acquisition module
- ordinate
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/188—Vegetation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
Abstract
The invention belongs to wetland landscape evolution analysis technical fields, disclose a kind of analysis system and method for intelligent wetland landscape evolution process, and the analysis system of the intelligence wetland landscape evolution process includes：Land-scape picture acquisition module, environmental factor acquisition module, data acquisition module, data processing module, database storage module, develops drafting module, cold and wet climate analog module, display module at animals and plants physiological acquisition module.The present invention can carry out visual in image analysis by developing drafting module to complicated landscape data, contribute to the advantageous protection to landscape；The spatial distribution of microclimatic elements can be accurately obtained by cold and wet climate analog module simultaneously, overcome it is traditional based on weather station data can not corresponsively spot interior details feature the shortcomings that, radiometric resolution and gross information content are enhanced about more than once, spatial simulation effect is inherently improved.
Description
Technical field
The invention belongs to wetland landscape evolution analysis technical field more particularly to a kind of intelligent wetland landscape evolution process
Analysis system and method.
Background technology
Wetland landscape (wetland landscape) refers to paying no attention to it as natural or artificial, often long or temporary marsh
Ground, moor, bog or waters area, carry or static or flowing or be fresh water, brackish water or salt water water body, including low tide
Shi Shuishen is no more than six meters of waters landscape.However, it is complicated to the process data analysis of Landscape in existing landscape system,
It is not intuitive enough, impact analysis result；The weather in existing landscape system is obtained by weather station simultaneously, however in difference
Under the conditions of landform and different landscape, the range that a weather station can represent has very big difference, even if passing through spatial interpolation methods
Also high-precision temperature spatial distribution can not be obtained.
To sum up, problem of the existing technology is：It is multiple to the process data analysis of Landscape in existing landscape system
It is miscellaneous, not enough intuitively, impact analysis result；Simultaneously the weather in existing landscape system be obtained by weather station, however
Under the conditions of different terrain and different landscape, the range that a weather station can represent has very big difference, even if passing through space interpolation
Method can not also obtain high-precision temperature spatial distribution.
Invention content
In view of the problems of the existing technology, the present invention provides a kind of analysis systems of intelligent wetland landscape evolution process
And method.
The invention is realized in this way a kind of analysis system of intelligence wetland landscape evolution process includes：
Land-scape picture acquisition module, connect with data acquisition module, for passing through satellite and live camera to wetland scape
See comprehensive shot；
The Land-scape picture acquisition module clusters feature space using sampling spectral clustering:(1) each pixel pair is utilized
The euclidean distance metric similarity for answering feature description vector constructs corresponding similarity matrix W, is degree according to L=D-W, wherein D
Matrix carries out the normalization of Laplacian Matrix using normalized cutFeature point is carried out to Laplacian Matrix
Solution exports its feature vector f；(2) the feature vector f exported is clustered, and obtains pixel cluster as a result, being corresponded to according to disparity map I
Pixel puts in order, and reverts to the size of original image, obtains final variation testing result figure, exports result；
Animals and plants physiological acquisition module, connect with data acquisition module, is adopted for the upgrowth situation to animal, plant
Collection；
Environmental factor acquisition module, connect with data acquisition module, for obtain the temperature of wetland, humidity, intensity of illumination,
Water quality, soil data are measured in real time acquisition；
Data acquisition module, with Land-scape picture acquisition module, animals and plants physiological acquisition module, environmental factor acquisition module,
Data processing module connects, and is used for Land-scape picture acquisition module, animals and plants physiological acquisition module and environmental factor acquisition module
The analog electric signal of acquisition is converted to digital quantity signal, and is sent to data processing module；
Data processing module is simulated with data acquisition module, database storage module, differentiation drafting module, cold and wet climate
Module, display module connection, are used for the data of the wetland according to data collecting module collected, carry out analyzing processing, and be saved in
In database storage module；
The data processing module carries out rectangle partitioning algorithm specific method such as in a scanning area to region of variation
Under：
Step 1, wetland image transmitting terminal obtain the resolution ratio of screen first, and the 0~C of range and row for obtaining column scan are swept
0~the R of range retouched；
The data of present frame wetland image conservation zone are saved in former frame wetland image buffer by step 2, transmitting terminal；
It intercepts and captures current screen bitmaps data and is stored in present frame wetland image buffer；
Step 3, transmitting terminal initializes variation rectangular area top left co-ordinate first and bottom right angular coordinate is (0,0), next time
Sweep starting point coordinate is (0,0), and row is unchanged to be identified as true, updates the range of the range and row scanning of column scan；
Step 4 judges whether to be expert in scanning range, not exist, jumps to step 10；
Step 5 judges whether within the scope of column scan, does not exist, and jumps to step 8；Within the scope of column scan using every
Row direct comparison method is detected current sampling point；Value is different, sets the unchanged mark of row to false first, then sentences
Whether disconnected be the first variation sampled point detected, be using sample point coordinate as the top left co-ordinate for changing rectangular area,
It is not first variation sampled point, the coordinate of the coordinate in the rectangle lower right corner and the point relatively and is maximized as new rectangle
Bottom right angular coordinate, then judge whether the sampled point is first variation sampled point of one's own profession, it is that the ordinate just by the sampled point is same
The ordinate in the rectangle upper left corner is compared and is minimized the top left co-ordinate of more new change rectangular area；It is worth identical, needs
Judge that row is unchanged and identifies whether, for false, if it is false, starting point of the record coordinate as scanning next time detects it is most
Latter row sampled point jumps to step 7 using last row sample point coordinate as the starting point of scanning next time；
Row coordinate is moved to right N row, jumps to step 5 and detect next sampled point by step 6；
Step 7, one's own profession detection finish, and the next time of the next sweep starting point coordinate of one's own profession and lastrow record is scanned
Point coordinates compares, and is maximized as new next sweep starting point coordinate, and line number adds 1, jump to step 4 from next line from
Head starts from left to right to detect；
Step 8, judge go it is unchanged identify whether as true and variation rectangular area top left co-ordinate be not (0,0), no
It is true, line number adds 1, jumps to step 4；It is true, then shows that full line without different pixels, has obtained the square of a variation
Shape region unit；Obtained variation rectangular area block upper left corner ordinate be moved to the left N row, lower right corner ordinate move right N row
To include wetland image boundary information；
Step 9 records the variation rectangular area coordinate detected and corresponding next sweep starting point coordinate, judges to work as
The range of preceding column scan whether 0~C and row scanning range whether 0~R, be, setting mark show the variation that current detection goes out
Rectangular area mark detects that then line number adds 1 to jump to step 4 to detect next change since next line for the first time
The rectangular area block of change；Until detecting the range beyond row scanning；
Step 10 after this detection, handles next sweep starting point all in this detection, calculates down
The set of secondary scanning range；The ordinate for first next sweep starting point that this is detected is first checked for whether than last row
The ordinate of sampled point is small, is not, which completes, and detects the ordinate of next next sweep starting point；It is, with first
The abscissa in the secondary variation rectangular area upper left corner detected is abscissa, is scanned relevant next time with currently changing rectangular area
The ordinate of starting point coordinate is ordinate, generates the top left co-ordinate of a next scanning range；With the change detected for the first time
The abscissa for changing the rectangular area lower right corner is abscissa, and a scanning next time model is generated by ordinate of the maximum number of column C of screen
The bottom right angular coordinate enclosed；Then handle second next sweep starting point, until next sweep starting point all in this detection all
It is treated as stopping；
Step 11 detects scanning area all in next scanning range set, is primarily based on next scanning range collection
The width and height of first scanning area in conjunction, the range of raw row scan and column scan repeat step 3 and are examined to step 10
The rectangular area block changed in first scanning area is surveyed, second scanning area is then handled, until next scanning range collection
Until all scanning areas are all detected in conjunction；
Step 12 repeats step 10 to step 11, obtains the variation rectangular area block of scanning range next time, until
The ordinate of all next sweep starting points is greater than or equal to the ordinate of last row sampled point, and entire screen detection finishes；
Step 13, the area for having obtained all frame wetland images relative to former frame wetland image change are minimum not
The set in overlapping rectangles region checks the rectangular area in the set, the vertical seat of two its upper left corner ordinates of rectangle and the lower right corner
Mark identical, and the lower right corner abscissa of a rectangle is adjacent with another rectangle upper left corner abscissa, merges into a rectangle, so
The wetland image data and respective coordinates that the set for recompressing and sending rectangular area afterwards is included are to client；
Step 14, wetland image receiving terminal will be based on each rectangular area wetland image data after the data decompression of reception
And respective coordinates are integrated into former frame wetland image and show；
Step 15 repeated step 2 every T seconds and arrives step 14, according to difference and the requirements of bandwidth of application scenarios,
It adjusts to interval time T；
Database storage module is connect with data processing module, for being protected to the data that data processing module is handled
It deposits；
The database storage module is respectively to the anisotropy value in the region there are significant difference, average diffusion
Rate value, radial diffusivity value are averaged, and are obtained the average anisotropy value in the region there are significant difference, are averagely put down
Equal diffusivity number, average radial diffusivity value；The average anisotropy value, average Mean diffusivity value, average radial are more
It dissipates rate value to be input in linear SVM as feature, linear SVM is trained by leaving-one method, finally
The region where feature is obtained, to obtain region related with lesion；
The specific implementation step being trained to linear SVM by leaving-one method is as follows：
Step 1 indicates the individual sum in data sample with n, and each individual has a m characteristic quantity, and the class of each individual
Attribute is all known, i.e. patient or normal person；Obtained data sample is divided into two groups, one group is test set, including one
Individual, one group is training set, includes except contained external owner in test set, total n-1 individual；
Step 2 trains the linear SVM with the training set, obtains support vector machines after training：According to
Weight vector is calculated in lower formula, is a m dimensional vector, and each element therein corresponds to a characteristic quantity；
yi(wTxi+b)-1+ξi≥0
s.t.ξi≥0；
Wherein, γ is punishment parameter, for realizing algorithm complexity and the wrong compromise for dividing sample number；ξiIt measures mistake and divides journey
Degree；yiFor everyone generic attribute；xiFor the feature vector of each individual；B is constant；
Step 3 assesses the performance of support vector machines after the training with the test set of known generic attribute：With institute
Support vector machines judges the generic attribute of the test set after stating training, and supporting vector chance provides attribute tags after the training
1 or -1, wherein 1 is patient, -1 is normal person, the judging result obtained by support vector machines after the training and the test
The practical generic attribute of collection compares, if the two is consistent, support vector cassification is correct after the training, otherwise, then mistake of classifying
Accidentally；
N individual is divided into test set and training set by step 4 again, and the test set includes an individual, and this
Body is differed with the individual in the preceding test set once tested, and remaining all individuals are used as training set, then according to step 2
Method train the linear SVM, support vector machines after training is obtained, then according still further to the method for the step 3
The performance of support vector machines after the training that assessment obtains；Repeat n-1 rear stopping of step 4；
The n times weights of each feature are averaging weights by step 5, and according to average weight by the descending progress of feature
Sequence, the minimum characteristic quantity of removal sequence；
Step 6 repeats step 1 to step i, then executes step 7；
Step 7, just according to the classification repeated in step 6 in the wheel n times obtained after step 1 to step 4 test
True rate and the classification accuracy rate result of the comparison of last round of n times test judge whether to stop：If the classification of wheel n times test is just
True rate is greater than or equal to the classification accuracy rate of last round of n times test, then returns to step five to step 6, otherwise stop；
Drafting module is developed, is connect with data processing module, graphic plotting is carried out for being developed to wetland landscape；
Cold and wet climate analog module, connect with data processing module, for close to wetland landscape by satellite remote sensing date
Stratum cold and wet climate element GIS spatial simulations；
Display module is connect with data processing module, for showing wetland landscape evolution.
Include following step another object of the present invention is to provide a kind of analysis method of intelligent wetland landscape evolution process
Suddenly：
Step 1, Land-scape picture acquisition module, animals and plants physiological acquisition module, environmental factor acquisition module are by the wet of detection
Ground landscape information data is converted to digital quantity signal by data acquisition module, and is sent to data processing module；
Step 2, data processing module carries out analyzing processing according to the wetland data of data collecting module collected, and preserves
Into database storage module；
Step 3 develops progress graphic plotting by developing drafting module to wetland landscape；Mould is simulated by cold and wet climate
Block is to wetland landscape surface layer cold and wet climate element GIS spatial simulations；
Step 4, finally by display module to wetland landscape evolution into display.
Further, the analogy method of the cold and wet climate analog module is as follows：
First, obtain research area's MODIS remote sensing images vegetation index data set NDVI, surface temperature data set LST with
And precipitable water data set Pw, and data processing is carried out, it is built using vegetation index data set NDVI and surface temperature data set LST
Vertical surface layer temperature inverse model, obtains the spatial distribution of temperature inside wetland patch and nonirrigated farmland patch；
Secondly, surface layer relative humidity inverting mould is established using surface temperature data set LST and precipitable water data set Pw
Type obtains the spatial distribution of surface layer relative humidity inside wetland patch and nonirrigated farmland patch；
Then, according to the spatial distribution of temperature and relative humidity inside wetland patch and nonirrigated farmland patch, use space polymerization
Method obtains the average value of wetland patch and nonirrigated farmland patch surface layer cold and wet climate element, builds cold and wet climate element edge effect water
Flat variation model；Cold and wet climate element is temperature and relative humidity；
Finally, according to the horizontal variation model analog result of cold and wet climate element edge effect of acquisition, GIS technology pair is utilized
Surface layer temperature and humidity under wetland landscape scale carry out spatial simulation.
Advantages of the present invention and good effect are：The present invention by develop drafting module can to complicated landscape data into
The visual in image analysis of row, contributes to the advantageous protection to landscape；It can accurately be obtained by cold and wet climate analog module simultaneously
The spatial distribution for taking microclimatic elements, overcome it is traditional based on weather station data can not corresponsively spot interior details feature lack
Point, radiometric resolution and gross information content are enhanced about more than once, and inherently improve spatial simulation effect.
Description of the drawings
Fig. 1 is the analysis method flow chart of intelligent wetland landscape evolution process provided in an embodiment of the present invention.
Fig. 2 is the analysis system structure diagram of intelligent wetland landscape evolution process provided in an embodiment of the present invention.
In figure：1, Land-scape picture acquisition module；2, animals and plants physiological acquisition module；3, environmental factor acquisition module；4, number
According to acquisition module；5, data processing module；6, database storage module；7, drafting module is developed；8, cold and wet climate analog module；
9, display module.
Specific implementation mode
In order to further understand the content, features and effects of the present invention, the following examples are hereby given, and coordinate attached drawing
Detailed description are as follows.
The structure of the present invention is explained in detail below in conjunction with the accompanying drawings.
As shown in Figure 1, the analysis method of intelligence wetland landscape evolution process provided in an embodiment of the present invention includes following step
Suddenly：
Step S101, Land-scape picture acquisition module, animals and plants physiological acquisition module, environmental factor acquisition module are by detection
Wetland landscape information data is converted to digital quantity signal by data acquisition module, and is sent to data processing module；
Step S102, data processing module carries out analyzing processing according to the wetland data of data collecting module collected, and protects
It is stored in database storage module；
Step S103 develops progress graphic plotting by developing drafting module to wetland landscape；It is simulated by cold and wet climate
Module is to wetland landscape surface layer cold and wet climate element GIS spatial simulations；
Step S104, finally by display module to wetland landscape evolution into display.
As shown in Fig. 2, the analysis system of intelligence wetland landscape evolution process provided in an embodiment of the present invention includes：Landscape map
As acquisition module 1, animals and plants physiological acquisition module 2, environmental factor acquisition module 3, data acquisition module 4, data processing module
5, database storage module 6, differentiation drafting module 7, cold and wet climate analog module 8, display module 9.
Land-scape picture acquisition module 1 is connect with data acquisition module 4, for passing through satellite and live camera to wetland
Landscape is comprehensive to be shot；
Animals and plants physiological acquisition module 2, connect with data acquisition module 4, is carried out for the upgrowth situation to animal, plant
Acquisition；
Environmental factor acquisition module 3 is connect with data acquisition module 4, and the temperature, humidity, illumination for obtaining wetland are strong
The data such as degree, water quality, soil are measured in real time acquisition；
Data acquisition module 4 acquires mould with Land-scape picture acquisition module 1, animals and plants physiological acquisition module 2, environmental factor
Block 3, data processing module 5 connect, for adopting Land-scape picture acquisition module 1, animals and plants physiological acquisition module 2 and environmental factor
The analog electric signal that collection module 3 obtains is converted to digital quantity signal, and is sent to data processing module 5；
Data processing module 5, with data acquisition module 4, database storage module 6, differentiation drafting module 7, cold and wet climate
Analog module 8, display module 9 connect, the data of the wetland for being acquired according to data acquisition module 4, carry out analyzing processing, and
It is saved in database storage module 6；
Database storage module 6 is connect with data processing module 5, and the data for handling data processing module 5 carry out
It preserves；
Drafting module 7 is developed, is connect with data processing module 5, graphic plotting is carried out for being developed to wetland landscape；
Cold and wet climate analog module 8 is connect with data processing module 5, for passing through satellite remote sensing date to wetland landscape
Surface layer cold and wet climate element GIS spatial simulations；
Display module 9 is connect with data processing module 5, for showing wetland landscape evolution.
The Land-scape picture acquisition module clusters feature space using sampling spectral clustering:(1) each pixel pair is utilized
The euclidean distance metric similarity for answering feature description vector constructs corresponding similarity matrix W, is degree according to L=D-W, wherein D
Matrix carries out the normalization of Laplacian Matrix using normalized cutFeature point is carried out to Laplacian Matrix
Solution exports its feature vector f；(2) the feature vector f exported is clustered, and obtains pixel cluster as a result, being corresponded to according to disparity map I
Pixel puts in order, and reverts to the size of original image, obtains final variation testing result figure, exports result；
The data processing module carries out rectangle partitioning algorithm specific method such as in a scanning area to region of variation
Under：
Step 1, wetland image transmitting terminal obtain the resolution ratio of screen first, and the 0~C of range and row for obtaining column scan are swept
0~the R of range retouched；
The data of present frame wetland image conservation zone are saved in former frame wetland image buffer by step 2, transmitting terminal；
It intercepts and captures current screen bitmaps data and is stored in present frame wetland image buffer；
Step 3, transmitting terminal initializes variation rectangular area top left co-ordinate first and bottom right angular coordinate is (0,0), next time
Sweep starting point coordinate is (0,0), and row is unchanged to be identified as true, updates the range of the range and row scanning of column scan；
Step 4 judges whether to be expert in scanning range, not exist, jumps to step 10；
Step 5 judges whether within the scope of column scan, does not exist, and jumps to step 8；Within the scope of column scan using every
Row direct comparison method is detected current sampling point；Value is different, sets the unchanged mark of row to false first, then sentences
Whether disconnected be the first variation sampled point detected, be using sample point coordinate as the top left co-ordinate for changing rectangular area,
It is not first variation sampled point, the coordinate of the coordinate in the rectangle lower right corner and the point relatively and is maximized as new rectangle
Bottom right angular coordinate, then judge whether the sampled point is first variation sampled point of one's own profession, it is that the ordinate just by the sampled point is same
The ordinate in the rectangle upper left corner is compared and is minimized the top left co-ordinate of more new change rectangular area；It is worth identical, needs
Judge that row is unchanged and identifies whether, for false, if it is false, starting point of the record coordinate as scanning next time detects it is most
Latter row sampled point jumps to step 7 using last row sample point coordinate as the starting point of scanning next time；
Row coordinate is moved to right N row, jumps to step 5 and detect next sampled point by step 6；
Step 7, one's own profession detection finish, and the next time of the next sweep starting point coordinate of one's own profession and lastrow record is scanned
Point coordinates compares, and is maximized as new next sweep starting point coordinate, and line number adds 1, jump to step 4 from next line from
Head starts from left to right to detect；
Step 8, judge go it is unchanged identify whether as true and variation rectangular area top left co-ordinate be not (0,0), no
It is true, line number adds 1, jumps to step 4；It is true, then shows that full line without different pixels, has obtained the square of a variation
Shape region unit；Obtained variation rectangular area block upper left corner ordinate be moved to the left N row, lower right corner ordinate move right N row
To include wetland image boundary information；
Step 9 records the variation rectangular area coordinate detected and corresponding next sweep starting point coordinate, judges to work as
The range of preceding column scan whether 0~C and row scanning range whether 0~R, be, setting mark show the variation that current detection goes out
Rectangular area mark detects that then line number adds 1 to jump to step 4 to detect next change since next line for the first time
The rectangular area block of change；Until detecting the range beyond row scanning；
Step 10 after this detection, handles next sweep starting point all in this detection, calculates down
The set of secondary scanning range；The ordinate for first next sweep starting point that this is detected is first checked for whether than last row
The ordinate of sampled point is small, is not, which completes, and detects the ordinate of next next sweep starting point；It is, with first
The abscissa in the secondary variation rectangular area upper left corner detected is abscissa, is scanned relevant next time with currently changing rectangular area
The ordinate of starting point coordinate is ordinate, generates the top left co-ordinate of a next scanning range；With the change detected for the first time
The abscissa for changing the rectangular area lower right corner is abscissa, and a scanning next time model is generated by ordinate of the maximum number of column C of screen
The bottom right angular coordinate enclosed；Then handle second next sweep starting point, until next sweep starting point all in this detection all
It is treated as stopping；
Step 11 detects scanning area all in next scanning range set, is primarily based on next scanning range collection
The width and height of first scanning area in conjunction, the range of raw row scan and column scan repeat step 3 and are examined to step 10
The rectangular area block changed in first scanning area is surveyed, second scanning area is then handled, until next scanning range collection
Until all scanning areas are all detected in conjunction；
Step 12 repeats step 10 to step 11, obtains the variation rectangular area block of scanning range next time, until
The ordinate of all next sweep starting points is greater than or equal to the ordinate of last row sampled point, and entire screen detection finishes；
Step 13, the area for having obtained all frame wetland images relative to former frame wetland image change are minimum not
The set in overlapping rectangles region checks the rectangular area in the set, the vertical seat of two its upper left corner ordinates of rectangle and the lower right corner
Mark identical, and the lower right corner abscissa of a rectangle is adjacent with another rectangle upper left corner abscissa, merges into a rectangle, so
The wetland image data and respective coordinates that the set for recompressing and sending rectangular area afterwards is included are to client；
Step 14, wetland image receiving terminal will be based on each rectangular area wetland image data after the data decompression of reception
And respective coordinates are integrated into former frame wetland image and show；
Step 15 repeated step 2 every T seconds and arrives step 14, according to difference and the requirements of bandwidth of application scenarios,
It adjusts to interval time T；
The database storage module is respectively to the anisotropy value in the region there are significant difference, average diffusion
Rate value, radial diffusivity value are averaged, and are obtained the average anisotropy value in the region there are significant difference, are averagely put down
Equal diffusivity number, average radial diffusivity value；The average anisotropy value, average Mean diffusivity value, average radial are more
It dissipates rate value to be input in linear SVM as feature, linear SVM is trained by leaving-one method, finally
The region where feature is obtained, to obtain region related with lesion；
The specific implementation step being trained to linear SVM by leaving-one method is as follows：
Step 1 indicates the individual sum in data sample with n, and each individual has a m characteristic quantity, and the class of each individual
Attribute is all known, i.e. patient or normal person；Obtained data sample is divided into two groups, one group is test set, including one
Individual, one group is training set, includes except contained external owner in test set, total n-1 individual；
Step 2 trains the linear SVM with the training set, obtains support vector machines after training：According to
Weight vector is calculated in lower formula, is a m dimensional vector, and each element therein corresponds to a characteristic quantity；
yi(wTxi+b)-1+ξi≥0
s.t.ξi≥0；
Wherein, γ is punishment parameter, for realizing algorithm complexity and the wrong compromise for dividing sample number；ξiIt measures mistake and divides journey
Degree；yiFor everyone generic attribute；xiFor the feature vector of each individual；B is constant；
Step 3 assesses the performance of support vector machines after the training with the test set of known generic attribute：With institute
Support vector machines judges the generic attribute of the test set after stating training, and supporting vector chance provides attribute tags after the training
1 or -1, wherein 1 is patient, -1 is normal person, the judging result obtained by support vector machines after the training and the test
The practical generic attribute of collection compares, if the two is consistent, support vector cassification is correct after the training, otherwise, then mistake of classifying
Accidentally；
N individual is divided into test set and training set by step 4 again, and the test set includes an individual, and this
Body is differed with the individual in the preceding test set once tested, and remaining all individuals are used as training set, then according to step 2
Method train the linear SVM, support vector machines after training is obtained, then according still further to the method for the step 3
The performance of support vector machines after the training that assessment obtains；Repeat n-1 rear stopping of step 4；
The n times weights of each feature are averaging weights by step 5, and according to average weight by the descending progress of feature
Sequence, the minimum characteristic quantity of removal sequence；
Step 6 repeats step 1 to step i, then executes step 7；
Step 7, just according to the classification repeated in step 6 in the wheel n times obtained after step 1 to step 4 test
True rate and the classification accuracy rate result of the comparison of last round of n times test judge whether to stop：If the classification of wheel n times test is just
True rate is greater than or equal to the classification accuracy rate of last round of n times test, then returns to step five to step 6, otherwise stop.
The analogy method of cold and wet climate analog module 8 provided by the invention is as follows：
First, obtain research area's MODIS remote sensing images vegetation index data set NDVI, surface temperature data set LST with
And precipitable water data set Pw, and data processing is carried out, it is built using vegetation index data set NDVI and surface temperature data set LST
Vertical surface layer temperature inverse model, obtains the spatial distribution of temperature inside wetland patch and nonirrigated farmland patch；
Secondly, surface layer relative humidity inverting mould is established using surface temperature data set LST and precipitable water data set Pw
Type obtains the spatial distribution of surface layer relative humidity inside wetland patch and nonirrigated farmland patch；
Then, according to the spatial distribution of temperature and relative humidity inside wetland patch and nonirrigated farmland patch, use space polymerization
Method obtains the average value of wetland patch and nonirrigated farmland patch surface layer cold and wet climate element, builds cold and wet climate element edge effect water
Flat variation model；Cold and wet climate element is temperature and relative humidity；
Finally, according to the horizontal variation model analog result of cold and wet climate element edge effect of acquisition, GIS technology pair is utilized
Surface layer temperature and humidity under wetland landscape scale carry out spatial simulation.The above is only to presently preferred embodiments of the present invention
, it is not intended to limit the present invention in any form, it is every that above example is done according to the technical essence of the invention
Any simple modification, equivalent variations with modification, belong in the range of technical solution of the present invention.
Claims (3)
1. a kind of analysis system of intelligence wetland landscape evolution process, which is characterized in that the intelligence wetland landscape evolution process
Analysis system include：
Land-scape picture acquisition module, connect with data acquisition module, for complete to wetland landscape by satellite and live camera
Orientation is shot；
The Land-scape picture acquisition module clusters feature space using sampling spectral clustering:(1) it is corresponded to using each pixel special
The euclidean distance metric similarity for levying description vectors, constructs corresponding similarity matrix W, is degree square according to L=D-W, wherein D
Battle array carries out the normalization of Laplacian Matrix using normalized cutFeature decomposition is carried out to Laplacian Matrix
Export its feature vector f；(2) the feature vector f exported is clustered, and obtains pixel cluster as a result, corresponding to picture according to disparity map I
Putting in order for element, reverts to the size of original image, obtains final variation testing result figure, exports result；
Animals and plants physiological acquisition module, connect with data acquisition module, is acquired for the upgrowth situation to animal, plant；
Environmental factor acquisition module, connect with data acquisition module, for obtaining the temperature of wetland, humidity, intensity of illumination, water
Matter, soil data are measured in real time acquisition；
Data acquisition module, with Land-scape picture acquisition module, animals and plants physiological acquisition module, environmental factor acquisition module, data
Processing module connects, for obtaining Land-scape picture acquisition module, animals and plants physiological acquisition module and environmental factor acquisition module
Analog electric signal be converted to digital quantity signal, and be sent to data processing module；
Data processing module, with data acquisition module, database storage module, develop drafting module, cold and wet climate analog module,
Display module connects, and is used for the data of the wetland according to data collecting module collected, carries out analyzing processing, and be saved in database
In memory module；
The data processing module carries out rectangle partitioning algorithm in a scanning area to region of variation, and the specific method is as follows：
Step 1, wetland image transmitting terminal obtain the resolution ratio of screen first, obtain 0~C of range and the row scanning of column scan
0~R of range；
The data of present frame wetland image conservation zone are saved in former frame wetland image buffer by step 2, transmitting terminal；It intercepts and captures
Current screen bitmaps data are simultaneously stored in present frame wetland image buffer；
Step 3, transmitting terminal initializes variation rectangular area top left co-ordinate first and bottom right angular coordinate is (0,0), scanning next time
Starting point coordinate is (0,0), and row is unchanged to be identified as true, updates the range of the range and row scanning of column scan；
Step 4 judges whether to be expert in scanning range, not exist, jumps to step 10；
Step 5 judges whether within the scope of column scan, does not exist, and jumps to step 8；Using straight every row within the scope of column scan
Comparison method is connect to be detected current sampling point；Value is different, sets the unchanged mark of row to false first, and then judgement is
No is the first variation sampled point detected, is not to be using sample point coordinate as the top left co-ordinate of variation rectangular area
The coordinate of the coordinate in the rectangle lower right corner and the point relatively and is maximized as new rectangle bottom right by first variation sampled point
Angular coordinate, then judge whether the sampled point is first variation sampled point of one's own profession, it is the same rectangle of ordinate just by the sampled point
The ordinate in the upper left corner is compared and is minimized the top left co-ordinate of more new change rectangular area；It is worth identical, needs to judge
Row is unchanged to identify whether detect it is last if it is the starting point that false, record coordinate are scanned as next time for false
Row sampled point jumps to step 7 using last row sample point coordinate as the starting point of scanning next time；
Row coordinate is moved to right N row, jumps to step 5 and detect next sampled point by step 6；
Step 7, one's own profession detection finish, and the next sweep starting point of the next sweep starting point coordinate of one's own profession and lastrow record is sat
Mark compares, and is maximized as new next sweep starting point coordinate, and line number adds 1, jumps to step 4 and is from the beginning opened from next line
Beginning is from left to right detected；
Step 8, judge go it is unchanged identify whether as true and variation rectangular area top left co-ordinate be not (0,0), be not
True, line number add 1, jump to step 4；It is true, then shows that full line without different pixels, has obtained the rectangle of a variation
Region unit；Obtained variation rectangular area block upper left corner ordinate be moved to the left N row, lower right corner ordinate move right N arrange with
Including wetland image boundary information；
Step 9, records the variation rectangular area coordinate detected and corresponding next sweep starting point coordinate, and forefront is worked as in judgement
The range of scanning whether 0~C and row scanning range whether 0~R, be, setting mark show the variation rectangle that current detection goes out
Area identification is to detect for the first time, and then line number adds 1 to jump to step 4 to detect next variation since next line
Rectangular area block；Until detecting the range beyond row scanning；
Step 10 after this detection, handles next sweep starting point all in this detection, calculates and sweep next time
Retouch the set of range；The ordinate for first next sweep starting point that this is detected is first checked for whether than last row sampling
The ordinate of point is small, is not, which completes, and detects the ordinate of next next sweep starting point；It is, to examine for the first time
The abscissa in the variation rectangular area upper left corner measured is abscissa, currently to change the relevant next sweep starting point in rectangular area
The ordinate of coordinate is ordinate, generates the top left co-ordinate of a next scanning range；With the variation square detected for the first time
The abscissa in the shape region lower right corner is abscissa, using the maximum number of column C of screen as one next scanning range of ordinate generation
Bottom right angular coordinate；Then second next sweep starting point is handled, until next sweep starting point all in this detection is all located
Until reason；
Step 11 detects scanning area all in next scanning range set, is primarily based in next scanning range set
The width and height of first scanning area, the range of raw row scan and column scan repeat step 3 and detect the to step 10
The rectangular area block changed in one scanning area then handles second scanning area, until in next scanning range set
Until all scanning areas are all detected；
Step 12 repeats step 10 to step 11, obtains the variation rectangular area block of scanning range next time, until all
The ordinate of next sweep starting point be greater than or equal to the ordinate of last row sampled point, the detection of entire screen finishes；
Step 13, the area for having obtained all frame wetland images relative to former frame wetland image change minimum are not overlapped
The set of rectangular area checks the rectangular area in the set, two its upper left corner ordinates of rectangle and lower right corner ordinate phase
Together, and the lower right corner abscissa of a rectangle is adjacent with another rectangle upper left corner abscissa, merges into a rectangle, then again
The wetland image data and respective coordinates that the set for compressing and sending rectangular area is included are to client；
Step 14, wetland image receiving terminal will be after the data decompressions of reception based on each rectangular area wetland image data and right
It answers coordinate to be integrated into former frame wetland image and shows；
Step 15 repeated step 2 every T seconds to step 14, according to the difference of application scenarios and the requirement of bandwidth, between pair
It adjusts every time T；
Database storage module is connect with data processing module, for being preserved to the data that data processing module is handled；
The database storage module is respectively to anisotropy value, the Mean diffusivity in the region there are significant difference
Value, radial diffusivity value are averaged, and are obtained the average anisotropy value in the region there are significant difference, are averaged
Diffusivity number, average radial diffusivity value；The average anisotropy value, average Mean diffusivity value, average radial disperse
Rate value is input to as feature in linear SVM, is trained to linear SVM by leaving-one method, final
Go out the region where feature, to obtain region related with lesion；
The specific implementation step being trained to linear SVM by leaving-one method is as follows：
Step 1 indicates the individual sum in data sample with n, and each individual has a m characteristic quantity, and the generic attribute of each individual
All it is known, i.e. patient or normal person；Obtained data sample is divided into two groups, one group is test set, including an individual,
One group is training set, includes except contained external owner in test set, total n-1 individual；
Step 2 trains the linear SVM with the training set, obtains support vector machines after training：According to following public affairs
Weight vector is calculated in formula, is a m dimensional vector, and each element therein corresponds to a characteristic quantity；
yi(wTxi+b)-1+ξi≥0
s.t.ξi≥0；
Wherein, γ is punishment parameter, for realizing algorithm complexity and the wrong compromise for dividing sample number；ξiIt measures mistake and divides degree；yiFor
Everyone generic attribute；xiFor the feature vector of each individual；B is constant；
Step 3 assesses the performance of support vector machines after the training with the test set of known generic attribute：With the instruction
Support vector machines judges the generic attribute of the test set after white silk, after the training supporting vector chance provide attribute tags 1 or-
1, wherein 1 is patient, -1 is normal person, the judging result obtained by support vector machines after the training and the test set
Practical generic attribute compares, if the two is consistent, support vector cassification is correct after the training, otherwise, then classification error；
N individual is divided into test set and training set by step 4 again, and the test set includes an individual, and the individual and
Individual in the preceding test set once tested differs, and remaining all individuals are used as training set, then according to the side of step 2
Method trains the linear SVM, obtains support vector machines after training, is then assessed according still further to the method for the step 3
The performance of support vector machines after the training obtained；Repeat n-1 rear stopping of step 4；
The n times weights of each feature are averaging weights by step 5, and are arranged according to average weight by feature is descending
Sequence, the minimum characteristic quantity of removal sequence；
Step 6 repeats step 1 to step i, then executes step 7；
Step 7, according to the classification accuracy rate repeated in step 6 in the wheel n times obtained after step 1 to step 4 test
Judge whether to stop with the classification accuracy rate result of the comparison of last round of n times test：If the classification accuracy rate of wheel n times test
More than or equal to the classification accuracy rate of last round of n times test, then five are returned to step to step 6, is otherwise stopped；
Drafting module is developed, is connect with data processing module, graphic plotting is carried out for being developed to wetland landscape；
Cold and wet climate analog module, connect with data processing module, for passing through satellite remote sensing date to wetland landscape surface layer
Cold and wet climate element GIS spatial simulations；
Display module is connect with data processing module, for showing wetland landscape evolution.
2. a kind of intelligent wetland landscape evolution process of the analysis system of intelligent wetland landscape evolution process as described in claim 1
Analysis method, which is characterized in that the analysis method of the intelligence wetland landscape evolution process includes the following steps：
Step 1, Land-scape picture acquisition module, animals and plants physiological acquisition module, environmental factor acquisition module are by the wetland scape of detection
It sees information data and digital quantity signal is converted to by data acquisition module, and be sent to data processing module；
Step 2, data processing module carries out analyzing processing according to the wetland data of data collecting module collected, and is saved in number
According in library storage module；
Step 3 develops progress graphic plotting by developing drafting module to wetland landscape；Pass through cold and wet climate analog module pair
Wetland landscape surface layer cold and wet climate element GIS spatial simulations；
Step 4, finally by display module to wetland landscape evolution into display.
3. the analysis method of intelligence wetland landscape evolution process as claimed in claim 2, which is characterized in that the cold and wet climate
The analogy method of analog module is as follows：
First, obtain research area's MODIS remote sensing images vegetation index data set NDVI, surface temperature data set LST and can
Precipitation data collection Pw, and data processing is carried out, it is established using vegetation index data set NDVI and surface temperature data set LST close
Stratum temperature inverse model obtains the spatial distribution of temperature inside wetland patch and nonirrigated farmland patch；
Secondly, surface layer relative humidity inverse model is established using surface temperature data set LST and precipitable water data set Pw, is obtained
Obtain the spatial distribution of surface layer relative humidity inside wetland patch and nonirrigated farmland patch；
Then, according to the spatial distribution of temperature and relative humidity inside wetland patch and nonirrigated farmland patch, use space polymerization obtains
The average value of wetland patch and nonirrigated farmland patch surface layer cold and wet climate element is obtained, structure cold and wet climate element edge effect level becomes
Change model；Cold and wet climate element is temperature and relative humidity；
Finally, according to the horizontal variation model analog result of cold and wet climate element edge effect of acquisition, using GIS technology to wetland
Surface layer temperature and humidity under landscape scale carry out spatial simulation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810301492.6A CN108520231A (en) | 2018-04-04 | 2018-04-04 | A kind of analysis system and method for intelligence wetland landscape evolution process |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810301492.6A CN108520231A (en) | 2018-04-04 | 2018-04-04 | A kind of analysis system and method for intelligence wetland landscape evolution process |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108520231A true CN108520231A (en) | 2018-09-11 |
Family
ID=63432118
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810301492.6A Pending CN108520231A (en) | 2018-04-04 | 2018-04-04 | A kind of analysis system and method for intelligence wetland landscape evolution process |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108520231A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103258126A (en) * | 2013-05-03 | 2013-08-21 | 中国科学院东北地理与农业生态研究所 | Wetland landscape surface layer cold and wet climatic element GIS space simulation method based on remote sensing data |
| CN103996196A (en) * | 2014-05-28 | 2014-08-20 | 西安电子科技大学 | DTI image analytical method based on multiple variables |
| CN104735449A (en) * | 2015-02-27 | 2015-06-24 | 成都信息工程学院 | Image transmission method and system based on rectangular segmentation and interlaced scanning |
-
2018
- 2018-04-04 CN CN201810301492.6A patent/CN108520231A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103258126A (en) * | 2013-05-03 | 2013-08-21 | 中国科学院东北地理与农业生态研究所 | Wetland landscape surface layer cold and wet climatic element GIS space simulation method based on remote sensing data |
| CN103996196A (en) * | 2014-05-28 | 2014-08-20 | 西安电子科技大学 | DTI image analytical method based on multiple variables |
| CN104735449A (en) * | 2015-02-27 | 2015-06-24 | 成都信息工程学院 | Image transmission method and system based on rectangular segmentation and interlaced scanning |
Non-Patent Citations (2)
| Title |
|---|
| 刘振宇 等: "应用区域统计特征的PolSAR影像分割", 《中国图象图形学报》 * |
| 符米静: "多光谱遥感图像变化检测方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dai et al. | The effects of image misregistration on the accuracy of remotely sensed change detection | |
| CN104881865B (en) | Forest pest and disease monitoring method for early warning and its system based on unmanned plane graphical analysis | |
| Wang et al. | An automatic region-based image segmentation algorithm for remote sensing applications | |
| Thomlinson et al. | Coordinating methodologies for scaling landcover classifications from site-specific to global: Steps toward validating global map products | |
| CN109635875A (en) | A kind of end-to-end network interface detection method based on deep learning | |
| Reis et al. | Identification of hazelnut fields using spectral and Gabor textural features | |
| Fang | Rice crop area estimation of an administrative division in China using remote sensing data | |
| CN107705309A (en) | Forest parameter evaluation method in laser point cloud | |
| Zhang et al. | Mapping freshwater marsh species in the wetlands of Lake Okeechobee using very high-resolution aerial photography and lidar data | |
| CN107767364A (en) | The method that trees canopy surface temperature is precisely extracted based on thermal-induced imagery | |
| Bai et al. | Quantifying tree cover in the forest–grassland ecotone of British Columbia using crown delineation and pattern detection | |
| CN116561509A (en) | Urban vegetation overground biomass accurate inversion method and system considering vegetation types | |
| Chen et al. | 3D model construction and ecological environment investigation on a regional scale using UAV remote sensing | |
| Solaimani et al. | Land use/cover change detection based on remote sensing data (A case study; Neka Basin) | |
| Haines-Young | The use of remotely-sensed satellite imagery for landscape classification in Wales (UK) | |
| Loos et al. | Shoreline feature extraction from remotely-sensed imagery | |
| CN108520231A (en) | A kind of analysis system and method for intelligence wetland landscape evolution process | |
| Babykalpana et al. | Classification of land use land cover change detection using remotely sensed data | |
| Wilkinson et al. | Change detection techniques for use in a statewide forest inventory program | |
| Bruce | Object oriented classification: case studies using different image types with different spatial resolutions | |
| László et al. | Object-based image analysis of pasture with trees and red mud spill | |
| Verma et al. | A comparative study of land cover classification techniques for “farmscapes” using very high resolution remotely sensed data | |
| Congalton | 21 How to Assess the Accuracy of Maps Generated from Remotely Sensed Data | |
| LU500262B1 (en) | Method for estimating vegetation coverage rate change in desert area | |
| Edwards et al. | A cognitive view of spatial uncertainty |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180911 |
|
| RJ01 | Rejection of invention patent application after publication |