CN114491108B - Online classification system and method based on multi-source remote sensing application data - Google Patents
Online classification system and method based on multi-source remote sensing application data Download PDFInfo
- Publication number
- CN114491108B CN114491108B CN202111650756.7A CN202111650756A CN114491108B CN 114491108 B CN114491108 B CN 114491108B CN 202111650756 A CN202111650756 A CN 202111650756A CN 114491108 B CN114491108 B CN 114491108B
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- data
- information
- sensing image
- product
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 52
- 238000012545 processing Methods 0.000 claims abstract description 60
- 230000000007 visual effect Effects 0.000 claims abstract description 35
- 238000007635 classification algorithm Methods 0.000 claims abstract description 28
- 238000012800 visualization Methods 0.000 claims abstract description 24
- 238000004458 analytical method Methods 0.000 claims abstract description 23
- 238000010586 diagram Methods 0.000 claims abstract description 10
- 238000012300 Sequence Analysis Methods 0.000 claims abstract description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 64
- 238000000605 extraction Methods 0.000 claims description 38
- 238000004519 manufacturing process Methods 0.000 claims description 26
- 238000007405 data analysis Methods 0.000 claims description 25
- 238000007476 Maximum Likelihood Methods 0.000 claims description 24
- 230000006870 function Effects 0.000 claims description 21
- 238000004364 calculation method Methods 0.000 claims description 14
- 238000012937 correction Methods 0.000 claims description 12
- 238000005520 cutting process Methods 0.000 claims description 12
- 238000003709 image segmentation Methods 0.000 claims description 12
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000000926 separation method Methods 0.000 claims description 8
- 238000009826 distribution Methods 0.000 claims description 7
- 238000005192 partition Methods 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 5
- 238000011161 development Methods 0.000 claims description 5
- 230000018109 developmental process Effects 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 6
- 238000012706 support-vector machine Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 230000003321 amplification Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000003199 nucleic acid amplification method Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 238000013475 authorization Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/51—Indexing; Data structures therefor; Storage structures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/53—Querying
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/54—Browsing; Visualisation therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/55—Clustering; Classification
-
- 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
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Image Processing (AREA)
Abstract
The invention discloses an online classification system and method based on multi-source remote sensing application data, and belongs to the technical field of satellite remote sensing. The invention comprises the following steps: the method comprises the following steps: constructing a multi-source data retrieval model based on space and elements, directly positioning remote sensing image data meeting requirements according to the model, and quickly and accurately acquiring metadata information; step two: performing two-dimensional visualization on a user online customized product, analyzing product data information of the online customized product in a time sequence analysis and space analysis mode, and performing online analysis and visual display on the product data information by using a scatter diagram, a pie chart and a histogram; step three: the user with the browsing and downloading authority browses and downloads the product data of the customized product meeting the actual requirements of the user in the step two by means of the related view button and the related tool; step four: and calling an online classification algorithm according to the product information downloaded by the user in the third step to perform online processing on the satellite remote sensing image.
Description
Technical Field
The invention relates to the technical field of satellite remote sensing, in particular to an online classification system and method based on multi-source remote sensing application data.
Background
Since the research of the aerospace remote sensing industry in China is developed, the satellite remote sensing technology is developed vigorously, the categories of satellite systems are gradually enriched, a military and civil application system with a certain scale and a wide degree is formed, the satellite application is developed to be deep and comprehensive, and the industrial scale is increased year by year. However, the existing remote sensing, navigation and communication satellite systems in China are systematic, isolated by military and civilian, information separation and service lag, along with the development of informatization, the application range of remote sensing application data is wider and wider, the data types are gradually increased, the data volume is larger and larger, the requirements of users in different fields and layers on the remote sensing application data are more and more intense, and the requirements of users in areas (smart cities), industry informatization, social public and the like on the deep application and industrialization development of full-range near real-time communication remote information are also more and more intense.
The existing online classification method based on multi-source remote sensing application data can produce satisfactory products only by manual authorization and operation of a complex production flow after production data and an algorithm are selected and a production order is submitted, the flow is complex and tedious, the efficiency is low, the algorithm and sample data of the produced products have uniqueness on the data of the produced products, the method is narrow in application range and not easy to expand a system and update the algorithm, and when the remote sensing images are extracted by using an algorithm formula in a ground object classification mode, the data analysis range is wide, the classification result error is large, the classified customized products cannot meet the requirements of users, and the using effect of the system is further reduced.
Disclosure of Invention
The invention aims to provide an online classification system and method based on multi-source remote sensing application data, so as to solve the problems in the background technology.
In order to solve the technical problems, the invention provides the following technical scheme: an online classification method based on multi-source remote sensing application data comprises the following steps:
the method comprises the following steps: constructing a multi-source data retrieval model based on space and elements, directly positioning remote sensing image data meeting requirements according to the model, and quickly and accurately acquiring metadata information;
step two: performing two-dimensional visualization on a user online customized product based on the two-dimensional model constructed in the first step, analyzing product data information of the online customized product in a time sequence analysis and space analysis mode, performing online analysis and visual display on the product data information by using a scatter diagram, a pie chart and a histogram, and performing amplification, reduction, rotation and mapping operation on a two-dimensional view in the two-dimensional model through visual display;
step three: the user with the browsing and downloading authority browses and downloads the product data of the customized product meeting the actual requirements of the user in the step two by means of the related view button and the related tool, and the user can also amplify, reduce and globally display the two-dimensional vector and the raster data by means of the related view button and the related tool and inquire and display the two-dimensional view information by the set conditions;
step four: calling an online classification algorithm according to the product information downloaded by the user in the third step to perform online processing on the satellite remote sensing image;
step five: a user logs in a system according to own user name and password information, selects corresponding columns according to actual requirements of the user, calls an order template to compile related order information and generates a task order;
step six: and based on the task order generated in the fifth step, the user selects one or more pieces of classification customization information to push according to the requirement of the user, and pushes the classification customization information to the user mobile terminal for application.
Further, the specific method for performing query retrieval on the metadata information based on the constructed model in the first step is as follows:
step one (I), selecting a vue front-end framework to build an online classification customization system based on a development mode of front-end and back-end separation, building a system background by using SSM and ZooKeeper, setting attribute information retrieval conditions of satellite remote sensing data on a vue front-end page, wherein the attribute information comprises a satellite, a sensor, a cloud amount, resolution, time and a region coordinate range, the region coordinate range attribute information represents a space in a model in the step one, the satellite, the sensor, the cloud amount, the resolution and the time attribute information represent elements in the model in the step one, the SSM is often used as a framework of a web project with a simpler data source, and the ZooKeeper represents a distributed application program coordination service of an open source code;
step one (II), a relational database is created, indexes are established for attribute fields which are commonly used as retrieval conditions during creation, and the created relational database is subjected to partition processing according to the use management environment and the data volume of the remote sensing images, so that the metadata information management efficiency and the retrieval efficiency of the satellite remote sensing images are improved, the relational database is a database for organizing data by adopting a relational model, each piece of data in the relational database is related attribute information of each scene of the satellite remote sensing images, and the metadata is also called medium data and relay data and refers to information for describing data attributes;
and step one (III), performing data retrieval on satellite remote sensing image metadata information records stored in the established relational database by using a database query language based on the retrieval conditions set in the step one (I), wherein the time reaches the second level when millions of records are retrieved by using single-element conditions.
Further, the specific method for calling the online classification algorithm to perform online processing on the satellite remote sensing image in the fourth step is as follows:
selecting satellite remote sensing image data to be processed, judging the type of the selected satellite remote sensing image data, if the selected data is judged to be corrected, directly selecting data, samples and algorithms to perform remote sensing image ground object classification extraction operation, otherwise, giving a prompt to remind a user to perform data correction processing, and selecting data, samples and algorithms to perform remote sensing image ground object classification extraction operation after the correction processing is finished;
selecting a stack type denoising self-encoder algorithm for classification, setting production algorithm parameters, calling an online classification algorithm to perform ground object classification extraction on the remote sensing image data, if the pixel size of the remote sensing image exceeds a product production parameter threshold set by the algorithm, cutting the to-be-produced data in a row-column matrix mode, and then calling the online classification algorithm to perform ground object classification extraction on the remote sensing image data;
step four (III), dividing the remote sensing image data into image data blocks under corresponding row numbers and column numbers based on the step four (II), then respectively carrying out classification extraction on the image data blocks, and after the processing is finished, splicing the image data blocks to finally obtain a complete ground object classification result product;
and step four (IV), performing visual display on the product image and the result analysis data file which are obtained in the step four (II) and the step four (III) and are generated according to the classification result through a visual tool in a chart mode, and facilitating the selection of the classified and customized product by a user.
Further, the specific steps of selecting data, samples and algorithms to perform the remote sensing image surface feature classification extraction operation in the fourth step (I) are as follows:
(1) selecting satellite remote sensing image data to be processed, and correcting the satellite remote sensing image data;
(2) selecting sample types according to user requirements, drawing corresponding sample points through a map control of a web end according to different sample types, and forming a sample set file by the sample points according to a format agreed with an algorithm;
(3) carrying out ground object classification extraction on the remote sensing image, wherein the specific method comprises the following steps:
1) Organizing the submitted sample set, and acquiring pixels under the corresponding coordinates of the samples;
2) Based on the data obtained in 1), selecting an algorithm model, carrying out model training and obtaining a classifier;
the selected algorithm model is a maximum likelihood classification algorithm, a part of the remote sensing image data with higher probability for generating observation data is searched through the maximum likelihood classification algorithm, and the method specifically comprises the following steps:
construction of a probability density function f D If f is D Is D, the distribution parameter is theta, a sample X having e values is extracted from the distribution 1 ，X 2 \8230;, xe, using function f D Calculating the probability, wherein a specific probability calculation formula P is as follows:
P(X 1 ，X 2 ，…，Xe)＝fD(X 1 ，X 2 ，…，Xe|θ)；
wherein Xe | θ represents a probability magnitude of an e-th sample value;
the maximum likelihood value of θ is calculated, and the specific likelihood calculation formula lik (θ) is:
lik(θ)＝f D (X 1 ，X 2 ，…，Xe|θ)；
finding all values of theta to maximize the function, wherein the value which maximizes the function is called as the maximum likelihood estimation of theta;
carrying out error solving on the obtained maximum likelihood estimation value theta, and adjusting the maximum likelihood estimation value theta according to an error result, wherein a specific error solving formula H (theta) is as follows:
wherein,
represents->
Is desired value of->
When H (theta) =0 or H (theta) = theta, the solved maximum likelihood estimation value theta is equal to the true value, otherwise, the solved maximum likelihood estimation value theta is adjusted according to the difference value of the two values;
3) And screening and extracting pixel points of the remote sensing image data based on a classifier, wherein the classifier represents a sample pixel set obtained based on the sample and coordinate information.
Further, the specific method for finally obtaining the complete product of the ground feature classification result in the step four (III) is as follows:
(1) Building a judgment model to judge whether the remote sensing image needs to be cut, wherein the specifically built judgment model Q is as follows:
a1 and A2 respectively represent the transverse distance and the longitudinal distance of a satellite remote sensing image to be processed, A1A 2 represents the size of the satellite remote sensing image to be processed, R1 and R2 respectively represent the transverse slice threshold and the longitudinal slice threshold of a remote sensing image slice, R1R 2 represents the remote sensing image slice threshold, when Q is less than or equal to 1, slicing processing on the remote sensing image is not needed, and when Q is more than or equal to 1, slicing processing on the remote sensing image is needed;
(2) Calculating the number of slices needed by the remote sensing image based on the judgment result in the step (1), wherein a specific calculation formula is as follows:
the number of transverse slices W is:
the number of longitudinal slices C is:
(3) Building a coordinate system, placing the satellite remote sensing images to be processed in the coordinate system, and calculating the longitude and latitude coordinates corresponding to each small remote sensing image slice based on the number of the slices calculated in the step (2), wherein the specific calculation formula is as follows:
the section abscissa is:
the slice ordinate is:
wherein i =1, 2, 3, \ 8230;, W represents the i-th transversely segmented small remote sensing image, j =1, 2, 3, \ 8230;, C represents the z-th transversely segmented small remote sensing image, andthe jth longitudinally-segmented small remote sensing image (x ', y') represents the longitude and latitude coordinates of the point of the satellite remote sensing image to be processed which is farthest from the origin, (x 1, y 1) represents the longitude and latitude coordinates of the satellite remote sensing image which is positioned on the same horizontal line with the point (x ', y'),
represents the lateral distance, based on the same ordinate, of each small remote sensing image>
The longitudinal distance of each small remote sensing image is shown when the horizontal coordinates are the same, xj (i-1) shows the horizontal coordinate corresponding to the i-1 th small remote sensing image, yj (i-1) shows the vertical coordinate corresponding to the i-1 th small remote sensing image, and the longitude and latitude coordinate corresponding to a certain point on each slice is (xji, yji);
(4) Marking each small image block by using a plurality of rows and columns while slicing, wherein the specific marking format is as follows:
name_m_n；
wherein m represents the number of rows and n represents the number of columns;
(5) Placing the cut remote sensing image slices under a temporary folder to be classified, placing classification results of the remote sensing image slices under a result temporary folder, traversing each piece of slice data to be classified for classified production after the remote sensing image slicing operation is completed, traversing the classification results in the result temporary folder after each remote sensing image slice completes ground feature classified extraction production, splicing the results according to longitude and latitude coordinate information of the slice data and mark information remained during cutting, and completing splicing of one scene of image data after all the results are traversed to generate a scene of remote sensing image data classification products.
An online classification system based on multi-source remote sensing application data comprises an information retrieval module, an information visualization display module, an information browsing and downloading module, a product classification module, an order management module and an information pushing module;
the information retrieval module is used for establishing a multi-source data retrieval model based on space and elements and transmitting the established two-dimensional model and a data retrieval result to the information visualization display module;
the information visualization display module is used for receiving the two-dimensional model and the data retrieval result transmitted by the information retrieval module, supporting multi-source and multi-scale display of product information of a server end and a PC end based on the constructed two-dimensional model, providing an online data analysis function, and transmitting the data analysis result and the displayed product information to the information browsing and downloading module;
the information browsing and downloading module is used for receiving the data analysis result transmitted by the information visualization display module and the displayed product information, browsing the product information by using a related view button and a related tool, downloading the required product information by combining with the actual requirement, and transmitting the downloaded product information and the data analysis result to the product classification module;
the product classification module receives the product information and the data analysis result transmitted by the information browsing and downloading module, provides matched classified customized products for the user according to the received content and the user requirements, and transmits the classified customized products to the order management module;
the order management module receives the classified customized products transmitted by the product classification module, selects corresponding columns according to the classified customized products, calls an order template to compile related order information, generates a task order and transmits the generated task order to the information pushing module;
the information pushing module receives the task order transmitted by the order management module, and the user selects one or more pieces of classified customized information to push according to the requirements of the user and confirms the generated task order.
Furthermore, the information retrieval module comprises a multi-source data retrieval model construction unit, a relational database creation unit and a data retrieval unit;
the multi-source data retrieval model building unit builds an online classification customization system on a vue front-end frame, sets attribute information retrieval conditions of satellite remote sensing data on a vue front-end page, builds a multi-source data retrieval model by utilizing an SSM (simple service modeling) and zookeeper building system background in a front-and-back end separation mode, and transmits the built multi-source data retrieval model and the set attribute information retrieval conditions of the satellite remote sensing data to a relational database building unit;
the relational database creating unit receives the multi-source data retrieval model transmitted by the multi-source data retrieval model creating unit and the set attribute information retrieval condition of the satellite remote sensing data, constructs a relational database by the related attribute information of each satellite remote sensing image, creates an index for the attribute field which is commonly used as the retrieval condition in the relational database, performs partition processing on the created relational database according to the use management environment and the data size of the remote sensing image, and transmits the partitioned relational database and the multi-source data retrieval model to the data retrieval unit;
the data retrieval unit receives the relational database and the multi-source data retrieval model transmitted by the relational database creation unit, uses a database query language to perform data retrieval on satellite remote sensing image pixel data information required by a user in the created relational database, and transmits a data retrieval result and the multi-source data retrieval model to the information visualization display module.
Further, the information visualization display module receives the two-dimensional model and the data retrieval result transmitted by the information retrieval module, the constructed two-dimensional model supports multi-source and multi-scale display of product information at a server end and a PC end, product data information is analyzed in a time sequence analysis and space analysis mode, the analysis result is displayed through a scatter diagram, a pie chart and a histogram, and the data analysis result and the displayed product information are transmitted to the information browsing and downloading module;
the information browsing and downloading module receives the data analysis result transmitted by the information visualization display module and the displayed product information, a user browses the product information by using a related view button and a related tool, two-dimensional vector and raster data in a two-dimensional view of the product are amplified, reduced and displayed globally, meanwhile, the product information can be inquired and displayed through set conditions, the user can download required product information according to authority and related configuration and in combination with actual requirements, and the downloaded product information and the data analysis result are transmitted to the product classification module.
Furthermore, the product classification module comprises a remote sensing image surface feature classification extraction unit, a remote sensing image segmentation unit, a surface feature classification result product acquisition unit and a visual display unit;
the remote sensing image ground object classification extraction unit receives the product information and the data analysis result transmitted by the information browsing and downloading module, judges the type of satellite remote sensing image data of the received product, transmits the satellite remote sensing image data of the product to the remote sensing image segmentation unit if the selected data is judged to be corrected, otherwise gives a prompt to remind a user of performing data correction, and performs type judgment again after the correction is completed;
the remote sensing image segmentation unit receives the product satellite remote sensing image data transmitted by the remote sensing image ground object classification extraction unit, selects a classification processing algorithm, sets production algorithm parameters, then calls an online classification algorithm to perform ground object classification extraction on the remote sensing image data, if the pixel size of the remote sensing image exceeds a product production parameter threshold set by the algorithm, cuts the data to be produced in a row-column matrix mode, transmits an image data block after cutting processing and the selected classification processing algorithm to the ground object classification result product acquisition unit, and if the production parameter set by the algorithm is lower than the threshold, directly transmits the remote sensing image and the selected classification processing algorithm to the ground object classification result product acquisition unit;
the ground object classification result product acquisition unit receives the remote sensing image transmitted by the remote sensing image segmentation unit, the cut image data block and the selected classification processing algorithm, classifies and extracts the received remote sensing image and the cut image data block through the selected classification processing algorithm, splices the image data blocks after the processing is finished, finally obtains a complete ground object classification result product, and transmits the obtained ground object classification result product to the visual display unit;
the visual display unit receives the ground feature classification result products transmitted by the ground feature classification result product acquisition unit, generates product images and result analysis data files according to the received contents, visually displays the result analysis data files in a chart mode through a visual tool, and transmits the visual display results to the order management module.
Further, the order management module receives a visual display result transmitted by the visual display unit, a user logs in the system according to own user name and password information, selects corresponding columns according to own actual requirements based on the visual display result, calls an order template to compile related order information, generates a task order, and transmits the generated task order to the information pushing module;
the information pushing module receives the task order transmitted by the order management module, the user selects one or more pieces of classified customization information in the generated task order according to the self requirement and pushes the one or more pieces of classified customization information to the system terminal, and the system terminal pushes the classified customization information to the mobile platform at the first time of data updating to confirm the generated task order.
Compared with the prior art, the invention has the following beneficial effects:
1. the remote sensing application data online customization classification system integrates remote sensing processing algorithms and processing tool resources, provides online data analysis Service for users in a unified mode by using a standard Web Service calling mode and sharing according to needs, enables the users to process and analyze data online without installing complex processing software, meets the requirement of multi-level users on diversification of remote sensing application data products, and is large in remote sensing application data type, large in data volume and wide in application range.
2. The method aims at the multi-source remote sensing application data, does not need to install any third-party plug-in, online self-defines and selects the sample and various classification algorithms, and online classification of the multi-source remote sensing application data is realized by online calling and obtaining production data.
3. When the remote sensing image is classified and extracted by using the algorithm formula, the part of the remote sensing image with higher probability generating observation data is searched by using the online algorithm, then the classification error in the classification extraction process is calculated, the problem that the result error after classification is inconsistent with the actual condition is avoided, the customized product obtained by the remote sensing image after classification extraction is ensured to meet the user requirement, and the use effect of the system is further improved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
FIG. 1 is a schematic view of a business process flow of an online classification system and method based on multi-source remote sensing application data of the present invention;
FIG. 2 is a system technical service flow diagram of an online classification system and method based on multi-source remote sensing application data of the present invention;
FIG. 3 is a system workflow of an online classification system and method based on multisource remote sensing application data of the present invention;
FIG. 4 is a schematic diagram of a general route of an online customized classification system for multi-source remote sensing application data based on the online classification system and method for multi-source remote sensing application data of the present invention;
FIG. 5 is a general architecture of an online customized classification system for multi-source remote sensing application data based on the online classification system and method for multi-source remote sensing application data of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
Referring to fig. 1-5, the present invention provides a technical solution: an online classification method based on multi-source remote sensing application data comprises the following steps:
the method comprises the following steps: the method comprises the following steps of constructing a multi-source data retrieval model based on space and elements, directly positioning remote sensing image data meeting requirements according to the model, and quickly and accurately acquiring metadata information, wherein the specific method for inquiring and retrieving the metadata information by constructing the model comprises the following steps:
step one (I), selecting a vue front-end framework to build an online classification customization system based on a development mode of front-end and back-end separation, building a system background by using SSM and ZooKeeper, setting attribute information retrieval conditions of satellite remote sensing data on a vue front-end page, wherein the attribute information comprises a satellite, a sensor, a cloud amount, resolution, time and a region coordinate range, the region coordinate range attribute information represents a space in a model in the step one, the satellite, the sensor, the cloud amount, the resolution and the time attribute information represent elements in the model in the step one, the SSM is often used as a framework of a web project with a simpler data source, and the ZooKeeper represents a distributed application program coordination service with an open source code;
step one (II), a relational database is created, indexes are established for attribute fields which are commonly used as retrieval conditions during creation, and the created relational database is subjected to partition processing according to the use management environment and the data volume of the remote sensing images, so that the metadata information management efficiency and the retrieval efficiency of the satellite remote sensing images are improved, the relational database is a database which organizes data by adopting a relational model, each piece of data in the relational database is related attribute information of each satellite remote sensing image, and the metadata is also called intermediary data and relay data and is information describing data attributes;
step one (III), performing data retrieval on satellite remote sensing image metadata information records stored in a created relational database by using a database query language based on retrieval conditions set in the step one (I), wherein the time reaches the second level when millions of records are retrieved by using single-element conditions;
step two: performing two-dimensional visualization on a user online customized product based on the two-dimensional model constructed in the first step, analyzing product data information of the online customized product in a time sequence analysis and space analysis mode, performing online analysis and visual display on the product data information by using a scatter diagram, a pie chart and a histogram, and performing amplification, reduction, rotation and mapping operation on a two-dimensional view in the two-dimensional model through visual display;
step three: the user with the browsing and downloading authority browses and downloads the product data of the customized product meeting the actual requirements of the user in the step two by means of the related view button and the related tool, and the user can also amplify, reduce and globally display the two-dimensional vector and the raster data by means of the related view button and the related tool and inquire and display the two-dimensional view information by the set conditions;
step four: calling an online classification algorithm according to product information downloaded by a user in the third step to perform online processing on the satellite remote sensing image, wherein the specific method comprises the following steps:
selecting satellite remote sensing image data to be processed, judging the type of the selected satellite remote sensing image data, if the selected data is judged to be corrected, directly selecting data, samples and algorithms to perform remote sensing image ground feature classification extraction operation, otherwise giving a prompt to remind a user to perform data correction processing, and selecting data, samples and algorithms to perform remote sensing image ground feature classification extraction operation after the correction processing is finished, wherein the specific steps of selecting data, samples and algorithms to perform remote sensing image ground feature classification extraction operation are as follows:
(1) selecting satellite remote sensing image data to be processed, and correcting the satellite remote sensing image data;
(2) selecting sample types according to user requirements, drawing corresponding sample points through a map control of a web end according to different sample types, and forming a sample set file by the sample points according to a format agreed with an algorithm;
(3) carrying out ground object classification extraction on the remote sensing image, wherein the specific method comprises the following steps:
1) Organizing the submitted sample set, and acquiring pixels under the corresponding coordinates of the samples;
2) Based on the data obtained in 1), selecting an algorithm model, carrying out model training and obtaining a classifier;
the selected algorithm model is a maximum likelihood classification algorithm, a part of the remote sensing image data with higher probability for generating observation data is searched through the maximum likelihood classification algorithm, and the method specifically comprises the following steps:
construction of a probability density function f D Is provided with f D Is D, the distribution parameter is theta, a sample X with e values is extracted from the distribution 1 ，X 2 \8230;, xe, using function f D Calculating the probability, wherein a specific probability calculation formula P is as follows:
P(X 1 ，X 2 ，…，Xe)＝fD(X 1 ，X 2 ，…，Xe|θ)；
wherein Xe | θ represents the probability size of the e-th sampling value;
the maximum likelihood value of θ is calculated, and the specific likelihood calculation formula lik (θ) is:
lik(θ)＝f D (X 1 ，X 2 ，…，Xe|θ)；
finding all values of theta to maximize the function, wherein the value which maximizes the function is called maximum likelihood estimation of theta;
carrying out error solving on the obtained maximum likelihood estimation value theta, and adjusting the maximum likelihood estimation value theta according to an error result, wherein a specific error solving formula H (theta) is as follows:
wherein,
represents->
Is desired value of->
When H (theta) =0 or H (theta) = theta, the solved maximum likelihood estimation value theta is equal to the true value, otherwise, the solved maximum likelihood estimation value theta is adjusted according to the difference value of the two values;
3) Based on a classifier, screening and extracting pixel points of remote sensing image data, wherein the classifier represents a sample pixel set obtained based on sample and coordinate information;
selecting a stack type denoising self-encoder algorithm for classification, setting production algorithm parameters, calling an online classification algorithm to perform ground object classification extraction on the remote sensing image data, if the pixel size of the remote sensing image exceeds a product production parameter threshold set by the algorithm, cutting the to-be-produced data in a row-column matrix mode, and then calling the online classification algorithm to perform ground object classification extraction on the remote sensing image data;
step four (III), dividing the remote sensing image data into image data blocks under corresponding row and column numbers based on the step four (II), then respectively classifying and extracting the image data blocks, and after the processing is finished, splicing the image data blocks to finally obtain a complete ground object classification result product, wherein the specific method comprises the following steps:
(1) Building a judgment model to judge whether the remote sensing image needs to be cut, wherein the specifically built judgment model Q is as follows:
a1 and A2 respectively represent the transverse distance and the longitudinal distance of a satellite remote sensing image to be processed, A1A 2 represents the size of the satellite remote sensing image to be processed, R1 and R2 respectively represent the transverse slice threshold and the longitudinal slice threshold of a remote sensing image slice, R1R 2 represents the remote sensing image slice threshold, when Q is less than or equal to 1, slicing processing on the remote sensing image is not needed, and when Q is more than or equal to 1, slicing processing on the remote sensing image is needed;
(2) Calculating the number of slices needed by the remote sensing image based on the judgment result in the step (1), wherein the specific calculation formula is as follows:
the number of transverse slices W is:
the number of longitudinal slices C is:
(3) Building a coordinate system, placing the satellite remote sensing images to be processed in the coordinate system, and calculating the longitude and latitude coordinates corresponding to each small remote sensing image slice based on the number of the slices calculated in the step (2), wherein the specific calculation formula is as follows:
the section abscissa is:
the slice ordinate is:
wherein, i =1, 2, 3, \ 8230, W represents the ith transversely-divided small remote sensing image, j =1, 2, 3, \8230, C represents the jth longitudinally-divided small remote sensing image, (x ', y') represents the longitude and latitude coordinates of the point which is farthest from the origin of the satellite remote sensing image to be processed, (x 1, y 1) represents the longitude and latitude coordinates of the satellite remote sensing image which is positioned on the same horizontal line with the point (x ', y'),
represents the transverse distance of each small remote sensing image when the ordinate is the same>
The longitudinal distance of each small remote sensing image is shown when the horizontal coordinates are the same, xj (i-1) shows the horizontal coordinate corresponding to the i-1 th small remote sensing image, yj (i-1) shows the vertical coordinate corresponding to the i-1 th small remote sensing image, and the longitude and latitude coordinate corresponding to a certain point on each slice is (xji, yji);
(4) Marking each small image block by using a plurality of rows and columns while slicing, wherein the specific marking format is as follows:
name_m_n；
wherein m represents the number of rows and n represents the number of columns;
(5) Placing the segmented remote sensing image slices under a temporary folder to be classified, placing the classification result of the remote sensing image slices under a result temporary folder, traversing each data of the slices to be classified for classified production after the remote sensing image slicing operation is finished, traversing the classification result in the result temporary folder after each remote sensing image slice finishes ground feature classification extraction production, splicing the results according to longitude and latitude coordinate information of the slice data and mark information reserved during segmentation, completing the splicing of one scene of image data after all the results are traversed, and generating a scene of remote sensing image data classification product;
step four (IV), performing visual display on the product image and the result analysis data file which are obtained in the step four (II) and the step four (III) and are generated according to the classification result through a visual tool in a chart mode, and facilitating selection of the classified and customized product by a user;
step five: a user logs in a system according to own user name and password information, selects corresponding columns according to actual requirements of the user, calls an order template to compile related order information and generates a task order;
step six: and based on the task order generated in the fifth step, the user selects one or more pieces of classified customization information to push according to the requirement of the user, and pushes the classified customization information to the mobile terminal of the user for application.
An online classification system based on multi-source remote sensing application data comprises an information retrieval module S1, an information visualization display module S2, an information browsing and downloading module S3, a product classification module S4, an order management module S5 and an information pushing module S6;
the information retrieval module S1 is used for establishing a multi-source data retrieval model based on space and elements, transmitting the established two-dimensional model and a data retrieval result to the information visualization display module S2, and the information retrieval module S1 comprises a multi-source data retrieval model establishing unit S11, a relational database establishing unit S12 and a data retrieval unit S13;
the multi-source data retrieval model building unit S11 is used for building an online classification customization system on a vue front-end frame, setting attribute information retrieval conditions of satellite remote sensing data on a vue front-end page, building a multi-source data retrieval model by utilizing an SSM (simple service manager) and a zookeeper building system background in a front-and-back end separation mode, and transmitting the built multi-source data retrieval model and the set attribute information retrieval conditions of the satellite remote sensing data to the relational database building unit S12;
the relational database creating unit S12 receives the multi-source data retrieval model transmitted by the multi-source data retrieval model creating unit S11 and the set attribute information retrieval conditions of the satellite remote sensing data, creates a relational database from the related attribute information of each satellite remote sensing image, creates an index for the attribute field commonly used as the retrieval conditions in the relational database, performs partition processing on the created relational database according to the use management environment and the data size of the remote sensing image, and transmits the partitioned relational database and the multi-source data retrieval model to the data retrieval unit S13;
the data retrieval unit S13 receives the relational database and the multi-source data retrieval model transmitted by the relational database creation unit S12, uses a database query language to perform data retrieval on satellite remote sensing image pixel data information required by a user in the created relational database, and transmits a data retrieval result and the multi-source data retrieval model to the information visualization display module S2;
the information visualization display module S2 receives the two-dimensional model and the data retrieval result transmitted by the information retrieval module S1, the constructed two-dimensional model supports multi-source and multi-scale display of product information at a server end and a PC end, the product data information is analyzed in a time sequence analysis and space analysis mode, the analysis result is displayed through a scatter diagram, a pie chart and a histogram, and the data analysis result and the displayed product information are transmitted to the information browsing and downloading module S3;
the information browsing and downloading module S3 receives the data analysis result and the displayed product information transmitted by the information visualization display module S2, a user browses the product information by using a related view button and a related tool, two-dimensional vector and raster data in a two-dimensional view of the product are amplified, reduced and displayed globally, meanwhile, the product information can be inquired and displayed through set conditions, the user can download required product information according to authority and related configuration and in combination with actual requirements, and the downloaded product information and the data analysis result are transmitted to the product classification module S4;
the product classification module S4 receives the product information and the data analysis result transmitted by the information browsing and downloading module S3, provides a matched classified customized product for the user according to the received content and the user requirement, and transmits the classified customized product to the order management module S5, wherein the product classification module S4 comprises a remote sensing image surface feature classification extraction unit S41, a remote sensing image segmentation unit S42, a surface feature classification result product acquisition unit S43 and a visual display unit S44;
the remote sensing image ground object classification extraction unit S41 receives the product information and the data analysis result transmitted by the information browsing and downloading module S3, judges the type of the satellite remote sensing image data of the received product, transmits the satellite remote sensing image data of the product to the remote sensing image segmentation unit S42 if the selected data is judged to be corrected, otherwise gives a prompt to remind a user of performing data correction processing, and performs type judgment again after the correction processing is completed;
the remote sensing image segmentation unit S42 receives the product satellite remote sensing image data transmitted by the remote sensing image ground object classification extraction unit S41, selects a classification processing algorithm, sets production algorithm parameters, then calls an online classification algorithm to perform ground object classification extraction on the remote sensing image data, if the pixel size of the remote sensing image exceeds a product production parameter threshold set by the algorithm, cuts the data to be produced in a row-column matrix mode, transmits an image data block after cutting processing and the selected classification processing algorithm to the ground object classification result product acquisition unit S43, and if the production parameter set by the algorithm is lower than the threshold, directly transmits the remote sensing image and the selected classification processing algorithm to the ground object classification result product acquisition unit S43;
the surface feature classification result product acquisition unit S43 receives the remote sensing image transmitted by the remote sensing image segmentation unit S42, the image data block after cutting processing and the selected classification processing algorithm, performs classification extraction on the received remote sensing image and the image data block after cutting processing through the selected classification processing algorithm, splices the image data blocks after processing, finally obtains a complete surface feature classification result product, and transmits the obtained surface feature classification result product to the visual display unit S44;
the visual display unit S44 receives the surface feature classification result product transmitted by the surface feature classification result product acquisition unit S43, generates a product image and a result analysis data file according to the received content, visually displays the result analysis data file in a chart manner by using a visualization tool, and transmits the visual display result to the order management module S5;
the order management module S5 receives the visual display result transmitted by the visual display unit S44, the user logs in the system according to own user name and password information, selects corresponding columns according to own actual requirements based on the visual display result, calls an order template to compile related order information, generates a task order, and transmits the generated task order to the information pushing module S6;
the information pushing module S6 receives the task order transmitted by the order management module S5, the user selects one or more pieces of classified customization information in the generated task order according to the self requirement and pushes the one or more pieces of classified customization information to the system terminal, and the system terminal pushes the classified customization information to the mobile platform at the first time of data updating to confirm the generated task order.
The first embodiment is as follows: setting a remote sensing image slice threshold value to be 1000 x 1000, setting the size of a satellite remote sensing image to be processed to be 2000 x 3000, and setting four-point longitude and latitude coordinates of the satellite remote sensing image to be processed to be (4, 4), (8, 0) and (4, 0);
slicing the remote sensing image;
the number of transverse slices W is:
the number of longitudinal slices C is:
the section abscissa is:
x11＝x1＝4；
the slice ordinate is:
y11＝y1＝4；
then the corresponding transverse slice coordinates of the remote sensing image are respectively: (4,4), (6,4), (8,4).
Example two: the selected algorithm model further comprises a stack type denoising self-encoder, a BP neural network classification algorithm, a minimum distance classification algorithm and a support vector machine algorithm;
the stack type denoising self-encoder comprises:
the method comprises the following steps of constructing a stacked denoising autoencoder model, wherein the model is formed by stacking a plurality of basic composition units (DAEs), a shallow network is built into a deep network in a stacking mode, and the stacked denoising autoencoder comprises two processes: encoding and decoding, the encoder mapping input data to a hidden layer to obtain a new feature representation, and the decoder mapping the hidden layer mapping data back to the original input data
The specific method for automatically analyzing the spectral statistical measurement parameters of each ground object type by the stacked denoising autoencoder model and identifying the ground object type of each pixel in the image to be processed comprises the following steps:
the coding formula is as follows:
the decoding formula is:
where, x represents the original input data,
mapping data representing hidden layers of the model, b (1) 、b (2) All represent the input bias of the hidden layer, W represents the weight of the hidden layer input, and s represents the activation function;
the error L of the reconstruction of the image data to be processed is as follows:
L＝||x-g(h(x))|| 2 ；
g (h (x)) represents original input data obtained by substituting mapping data of a model hidden layer into a decoding formula, x-g (h (x)) represents the difference between actual original input data and the original input data obtained by calculation, a sparse matrix is added into the hidden layer to reduce model parameters, few dimensions are used for representing the input data, the average value output by hidden nodes is 0 as much as possible, most of hidden layer nodes are set to be in an inactive state, the sparse matrix represents that the number of elements with the value of 0 in the matrix is far more than the number of elements with the value of 0, and the distribution of the elements with the value of 0 is irregular;
the activation value of the hidden layer is represented as:
wherein,
representing the activation value of the jth neuron in the k-th hidden layer, and defining a loss function J of a stacked denoising self-encoder model in order to enable more neural nodes to be in a suppression state sparse (W, b) is:
wherein x is (i) Represents the ith sample; n represents the number of samples; s l Expressing the number of the units of the l layer, wherein lambda is a regularization coefficient, beta is the weight of a penalty factor, and min J is solved through a gradient descent algorithm sparse (W, b), obtaining a local optimal solution;
BP neural network classification algorithm:
after a sample is input, obtaining a characteristic vector of the sample, obtaining an input value of a sensor according to a weight vector, calculating the output of each sensor by using a sigmoid function, taking the output as the input of the sensor of the next layer, and repeating the steps until the output layer, wherein the derivation process of the algorithm mainly utilizes a gradient descent algorithm to minimize a loss function, and the loss function is as follows:
the method comprises the following steps that Tu represents expected output of a decoder, u represents the number of network layers of a stacked denoising self-encoder model, x represents original input data output by the decoder after mapping processing, E represents an error value between the expected output and actual output, and the typical spectral parameters of the surface features are analyzed according to the E value, so that the type of the surface features to which each pixel in an image to be processed belongs is identified;
for each weight wij in the network, its derivative is calculated:
the weight changes towards the negative gradient of the loss function, and there is thus a weight change Δ w ij Comprises the following steps:
minimum distance classification algorithm:
marking typical objects in the image to be processed by a minimum distance classification algorithm, wherein the specific method comprises the following steps:
in the two-dimensional model, the mean value of two dimensions of known pixels belonging to the same ground object class is calculated, and the distance between the identified ground object class and each known pixel belonging to the same ground object class is calculated, so that a typical ground object in the image to be processed is marked, wherein the specific distance calculation formula is as follows:
d(f，h i )＝|f-h i | 2 ＝(f-h i ) T (f-h i )＝f T f-(f T h i +fh i T -h i T h i )；
wherein d (f, h) i ) Indicating the distance between the recognized feature class and each known feature class belonging to the same feature class, i indicating the feature class, f indicating the recognized feature class, h i Represents the known mean value of the terrain belonging to class i when d (f, h) i ) When the minimum value is taken, judging the type of the recognized ground object according to the value of i at the moment;
support vector machine algorithm:
the remote sensing image is segmented through a support vector machine classification algorithm, and the specific method comprises the following steps:
and (3) solving a separation hyperplane which enables the remote sensing image interval to be maximum:
W * ·x′+b * ＝0；
a corresponding classification decision function f (x) when the remote sensing image is separated:
f(x)＝sign(W * ·x′+b * )；
wherein, b * Represents a shift, W * Denotes coefficient when sample point (x 'i, y' i ) Is hyperplane (W) * ，b * ) Correctly classified, point x' i And the hyperplane (W) * ，b * ) The distance of (a) is:
in order to solve the optimization problem of the linear branching support vector machine, the optimization problem is used as an original optimization problem, lagrangian duality is applied, the optimal solution of the original problem is obtained by solving a duality problem, and the optimal solution is as follows:
it is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that changes may be made in the embodiments and/or equivalents thereof without departing from the spirit and scope of the invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (6)
1. An online classification method based on multi-source remote sensing application data is characterized by comprising the following steps: the method comprises the following steps:
the method comprises the following steps: constructing a multi-source data retrieval model based on space and elements, directly positioning remote sensing image data meeting requirements according to the model, and quickly and accurately acquiring metadata information;
step two: performing two-dimensional visualization on the online customized product of the user based on the two-dimensional model constructed in the first step, analyzing the product data information of the online customized product in a time sequence analysis and space analysis mode, and performing online analysis and visual display on the product data information by using a scatter diagram, a pie chart and a histogram;
step three: the user with the browsing and downloading authority browses and downloads the product data of the customized product meeting the actual requirements of the user in the step two by means of the related view button and the related tool, and the user can also amplify, reduce and globally display the two-dimensional vector and the raster data by means of the related view button and the related tool and inquire and display the two-dimensional view information by the set conditions;
step four: calling an online classification algorithm to perform online processing on the satellite remote sensing image according to the product information downloaded by the user in the third step;
the specific method for calling the online classification algorithm to perform online processing on the satellite remote sensing image in the fourth step is as follows:
selecting satellite remote sensing image data to be processed, judging the type of the selected satellite remote sensing image data, if the selected data is judged to be corrected, directly selecting data, samples and algorithms to perform remote sensing image ground object classification extraction operation, otherwise, giving a prompt to remind a user to perform data correction processing, and selecting data, samples and algorithms to perform remote sensing image ground object classification extraction operation after the correction processing is finished;
the specific steps of selecting data, samples and algorithms to carry out remote sensing image surface feature classification extraction operation in the step four (I) are as follows:
(1) selecting satellite remote sensing image data to be processed, and correcting the satellite remote sensing image data;
(2) selecting sample categories according to user requirements, drawing corresponding sample points through a map control of a web end according to different sample categories, and forming sample set files by the sample points according to a format agreed with an algorithm;
(3) carrying out ground object classification extraction on the remote sensing image, wherein the specific method comprises the following steps:
1) Organizing the submitted sample set, and acquiring pixels under the corresponding coordinates of the samples;
2) Based on the data obtained in 1), selecting an algorithm model, carrying out model training and obtaining a classifier;
the selected algorithm model is a maximum likelihood classification algorithm, a part of the remote sensing image data with higher probability for generating observation data is searched through the maximum likelihood classification algorithm, and the method specifically comprises the following steps:
construction of a probability density function f D Is provided with f D Has a probability distribution of d, distributionWith a parameter theta, a sample X having e values is extracted from the distribution 1 ，X 2 \8230;, xe, using function f D Calculating the probability, wherein a specific probability calculation formula P is as follows:
P(X 1 ，X 2 ，…，Xe)＝fD(X 1 ，X 2 ，…，Xe|θ)；
wherein Xe | θ represents a probability magnitude of an e-th sample value;
the maximum likelihood value of θ is calculated, and the specific likelihood calculation formula lik (θ) is:
lik(θ)＝f D (X 1 ，X 2 ，…，Xe|θ)；
finding all values of theta to maximize the function, wherein the value which maximizes the function is called as the maximum likelihood estimation of theta;
carrying out error solving on the obtained maximum likelihood estimation value theta, and adjusting the maximum likelihood estimation value theta according to an error result, wherein a specific error solving formula H (theta) is as follows:
wherein,
to represent
The expected value of (c) is,
when H (theta) =0 or H (theta) = theta, the solved maximum likelihood estimation value theta is equal to the true value, otherwise, the solved maximum likelihood estimation value theta is adjusted according to the difference value of the two values;
3) Based on the classifier, screening and extracting pixel points of the remote sensing image data;
selecting a stack type denoising self-encoder algorithm for classification processing, setting production algorithm parameters, calling an online classification algorithm to perform ground object classification extraction on the remote sensing image data, if the pixel size of the remote sensing image exceeds a product production parameter threshold value set by the algorithm, cutting the to-be-produced data in a row-column matrix mode, and then calling the online classification algorithm to perform ground object classification extraction on the remote sensing image data;
step four (III) based on the step four (II), the remote sensing image data are divided into image data blocks under corresponding row and column numbers, then the image data blocks are respectively classified and extracted, after the processing is finished, the image data blocks are spliced, and finally a complete ground object classification result product is obtained;
the concrete method for finally obtaining the complete ground object classification result product in the step four (III) comprises the following steps:
(1) Building a judgment model to judge whether the remote sensing image needs to be cut, wherein the specifically built judgment model Q is as follows:
the method comprises the following steps that A1 and A2 respectively represent the transverse distance and the longitudinal distance of a satellite remote sensing image to be processed, A1A 2 represents the size of the satellite remote sensing image to be processed, R1 and R2 respectively represent the transverse slice threshold and the longitudinal slice threshold of a remote sensing image slice, R1R 2 represents the remote sensing image slice threshold, when Q is less than or equal to 1, slicing processing on the remote sensing image is not needed, and when Q is more than or equal to 1, slicing processing on the remote sensing image is needed;
(2) Calculating the number of slices needed by the remote sensing image based on the judgment result in the step (1), wherein the specific calculation formula is as follows:
the number of transverse slices W is:
the number of longitudinal slices C is:
(3) Building a coordinate system, placing the satellite remote sensing images to be processed in the coordinate system, and calculating the longitude and latitude coordinates corresponding to each small remote sensing image slice based on the number of the slices calculated in the step (2), wherein the specific calculation formula is as follows:
the section abscissa is:
the slice ordinate is:
wherein, i =1, 2, 3, \ 8230, W represents the ith transversely-divided small remote sensing image, j =1, 2, 3, \8230, C represents the jth longitudinally-divided small remote sensing image, (x ', y') represents the longitude and latitude coordinates of the point which is farthest from the origin of the satellite remote sensing image to be processed, (x 1, y 1) represents the longitude and latitude coordinates of the satellite remote sensing image which is positioned on the same horizontal line with the point (x ', y'),
which represents the lateral distance of each small remote sensing image when the ordinate is the same,
the longitudinal distance of each small remote sensing image is shown when the horizontal coordinates are the same, xj (i-1) shows the horizontal coordinate corresponding to the i-1 th small remote sensing image, yj (i-1) shows the vertical coordinate corresponding to the i-1 th small remote sensing image, and the longitude and latitude coordinate corresponding to a certain point on each slice is (xji, yji);
(4) Marking each small image block by using a plurality of rows and columns while slicing, wherein the specific marking format is as follows:
name_m_n；
wherein m represents the number of rows and n represents the number of columns;
(5) Placing the cut remote sensing image slices under a temporary folder to be classified, placing classification results of the remote sensing image slices under a result temporary folder, traversing each piece of slice data to be classified for classified production after the remote sensing image slicing operation is finished, traversing the classification results in the result temporary folder after each remote sensing image slice finishes ground feature classified extraction production, splicing the results according to longitude and latitude coordinate information of the slice data and mark information remained during cutting, and completing splicing of one scene of image data after all the results are traversed to generate a scene of remote sensing image data classification products;
step four (IV), performing visual display on the product image and the result analysis data file which are obtained in the step four (II) and the step four (III) and are generated according to the classification result through a visual tool in a chart mode;
step five: a user logs in a system according to own user name and password information, selects corresponding columns according to actual requirements of the user, calls an order template to compile related order information and generates a task order;
step six: and based on the task order generated in the fifth step, the user selects one or more pieces of classification customization information to push according to the requirement of the user, and pushes the classification customization information to the user mobile terminal for application.
2. The on-line classification method based on multi-source remote sensing application data according to claim 1, characterized in that: the specific method for querying and retrieving the metadata information based on the constructed model in the first step is as follows:
selecting a vue front-end framework to build an online classification customization system based on a development mode of front-end and back-end separation, building a system background by using SSM and zookeeper, and setting attribute information retrieval conditions of satellite remote sensing data on a vue front-end page;
creating a relational database, creating an index for an attribute field which is commonly used as a retrieval condition during creation, and performing partition processing on the created relational database according to the use management environment and the data volume of the remote sensing image;
and step one (III), performing data retrieval on the satellite remote sensing image pixel data information records stored in the created relational database by using a database query language based on the retrieval conditions set in the step one (I).
3. An online classification system based on multi-source remote sensing application data is characterized in that: the system comprises an information retrieval module (S1), an information visualization display module (S2), an information browsing and downloading module (S3), a product classification module (S4), an order management module (S5) and an information pushing module (S6);
the information retrieval module (S1) is used for establishing a multi-source data retrieval model based on space and elements and transmitting the established two-dimensional model and a data retrieval result to the information visualization display module (S2);
the information visualization display module (S2) is used for receiving the two-dimensional model and the data retrieval result transmitted by the information retrieval module (S1), supporting multi-source and multi-scale display of product information of a server side and a PC side based on the constructed two-dimensional model, providing an online data analysis function, and transmitting the data analysis result and the displayed product information to the information browsing and downloading module (S3);
the information browsing and downloading module (S3) is used for receiving the data analysis result transmitted by the information visualization display module (S2) and the displayed product information, browsing the product information by using a related view button and a related tool, downloading the required product information by combining with the actual requirement, and transmitting the downloaded product information and the data analysis result to the product classification module (S4);
the product classification module (S4) receives the product information and the data analysis result transmitted by the information browsing and downloading module (S3), provides a matched classified customized product for the user according to the received content and the user requirement, and transmits the classified customized product to the order management module (S5);
the product classification module (S4) comprises a remote sensing image surface feature classification extraction unit (S41), a remote sensing image segmentation unit (S42), a surface feature classification result product acquisition unit (S43) and a visual display unit (S44);
the remote sensing image ground object classification extraction unit (S41) receives the product information and the data analysis result transmitted by the information browsing download module (S3), judges the type of satellite remote sensing image data of a received product, transmits the satellite remote sensing image data of the product to the remote sensing image segmentation unit (S42) if the selected data is judged to be corrected, otherwise gives a prompt to remind a user of performing data correction processing, and performs type judgment again after the correction processing is finished;
the remote sensing image segmentation unit (S42) receives the product satellite remote sensing image data transmitted by the remote sensing image ground object classification extraction unit (S41), selects a classification processing algorithm, sets production algorithm parameters, then calls an online classification algorithm to perform ground object classification extraction on the remote sensing image data, if the pixel size of the remote sensing image exceeds a product production parameter threshold set by the algorithm, cuts the production data by adopting a row-column matrix mode, transmits an image data block after cutting processing and the selected classification processing algorithm to the ground object classification result product acquisition unit (S43), and if the production parameter set by the algorithm is lower than the threshold, directly transmits the remote sensing image and the selected classification processing algorithm to the ground object classification result product acquisition unit (S43);
the land feature classification result product acquisition unit (S43) receives the remote sensing image transmitted by the remote sensing image segmentation unit (S42), the image data block after cutting processing and the selected classification processing algorithm, performs classification extraction on the received remote sensing image and the image data block after cutting processing through the selected classification processing algorithm, splices the image data blocks after processing, finally obtains a complete land feature classification result product, and transmits the obtained land feature classification result product to the visual display unit (S44);
the visual display unit (S44) receives the surface feature classification result products transmitted by the surface feature classification result product acquisition unit (S43), generates a product image and a result analysis data file according to the received content, visually displays the result analysis data file in a chart mode through a visual tool, and transmits the visual display result to the order management module (S5);
the order management module (S5) receives the classified customized products transmitted by the product classification module (S4), selects corresponding columns according to the classified customized products, calls an order template to compile related order information to generate a task order, and transmits the generated task order to the information pushing module (S6);
the information pushing module (S6) receives the task order transmitted by the order management module (S5), and the user selects one or more pieces of classified customization information to push according to the requirement of the user so as to confirm the generated task order.
4. The on-line classification system based on multi-source remote sensing application data according to claim 3, characterized in that: the information retrieval module (S1) comprises a multi-source data retrieval model construction unit (S11), a relational database creation unit (S12) and a data retrieval unit (S13);
the multi-source data retrieval model building unit (S11) builds an online classification customization system on a vue front-end frame, sets attribute information retrieval conditions of satellite remote sensing data on a vue front-end page, builds a multi-source data retrieval model by utilizing an SSM (simple service modeling) and zookeeper building system background in a front-end and back-end separation mode, and transmits the built multi-source data retrieval model and the set attribute information retrieval conditions of the satellite remote sensing data to a relational database building unit (S12);
the relational database creating unit (S12) receives the multi-source data retrieval model transmitted by the multi-source data retrieval model building unit (S11) and the set attribute information retrieval condition of the satellite remote sensing data, builds a relational database by the related attribute information of each scene satellite remote sensing image, builds an index for the attribute field which is commonly used as the retrieval condition in the relational database, performs partition processing on the created relational database according to the use management environment and the data size of the remote sensing image, and transmits the partitioned relational database and the multi-source data retrieval model to the data retrieval unit (S13);
the data retrieval unit (S13) receives the relational database and the multi-source data retrieval model transmitted by the relational database creation unit (S12), uses a database query language to perform data retrieval on satellite remote sensing image metadata information required by a user in the created relational database, and transmits a data retrieval result and the multi-source data retrieval model to the information visualization display module (S2).
5. The on-line classification system based on multi-source remote sensing application data according to claim 4, characterized in that: the information visualization display module (S2) receives the two-dimensional model and the data retrieval result transmitted by the information retrieval module (S1), the constructed two-dimensional model supports multi-source and multi-scale display of product information of a server side and a PC side, product data information is analyzed in a time sequence analysis and space analysis mode, the analysis result is displayed through a scatter diagram, a pie chart and a column diagram, and the data analysis result and the displayed product information are transmitted to the information browsing and downloading module (S3);
the information browsing and downloading module (S3) receives the data analysis result transmitted by the information visualization display module (S2) and the displayed product information, a user browses the product information by using a related view button and a related tool, two-dimensional vector and raster data in a two-dimensional view of the product are amplified, reduced and displayed globally, meanwhile, the product information can be inquired and displayed through set conditions, the user can download required product information according to authority and related configuration in combination with actual requirements, and the downloaded product information and the data analysis result are transmitted to the product classification module (S4).
6. The on-line classification system based on multi-source remote sensing application data according to claim 5, characterized in that: the order management module (S5) receives the visual display result transmitted by the visual display unit (S44), a user logs in the system according to own user name and password information, selects corresponding columns according to own actual requirements based on the visual display result, calls an order template to compile related order information, generates a task order, and transmits the generated task order to the information push module (S6);
the information pushing module (S6) receives the task order transmitted by the order management module (S5), the user selects one or more pieces of classified customization information in the generated task order according to the self requirement and pushes the one or more pieces of classified customization information to the system terminal, and the system terminal pushes the classified customization information to the mobile platform at the first time of data updating to confirm the generated task order.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111650756.7A CN114491108B (en) | 2021-12-30 | 2021-12-30 | Online classification system and method based on multi-source remote sensing application data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111650756.7A CN114491108B (en) | 2021-12-30 | 2021-12-30 | Online classification system and method based on multi-source remote sensing application data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114491108A CN114491108A (en) | 2022-05-13 |
| CN114491108B true CN114491108B (en) | 2023-03-24 |
Family
ID=81507679
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111650756.7A Active CN114491108B (en) | 2021-12-30 | 2021-12-30 | Online classification system and method based on multi-source remote sensing application data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114491108B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115934759B (en) * | 2022-11-30 | 2023-12-22 | 二十一世纪空间技术应用股份有限公司 | Acceleration calculation method for massive multi-source heterogeneous satellite data query |
| CN116028566B (en) * | 2023-02-21 | 2023-08-22 | 中国科学院空间应用工程与技术中心 | Spacecraft data analysis system, method, medium and equipment |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| ATE443846T1 (en) * | 2004-09-15 | 2009-10-15 | Deutsch Zentr Luft & Raumfahrt | PROCESSING OF REMOTE SENSING DATA |
| CN111126482B (en) * | 2019-12-23 | 2020-11-13 | 自然资源部国土卫星遥感应用中心 | Remote sensing image automatic classification method based on multi-classifier cascade model |
| CN111125080B (en) * | 2019-12-23 | 2023-09-26 | 北京航天世景信息技术有限公司 | Multi-source remote sensing image integrated management system and method based on pattern spot model |
| CN111666313B (en) * | 2020-05-25 | 2023-02-07 | 中科星图股份有限公司 | Correlation construction and multi-user data matching method based on multi-source heterogeneous remote sensing data |
-
2021
- 2021-12-30 CN CN202111650756.7A patent/CN114491108B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN114491108A (en) | 2022-05-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111666313B (en) | Correlation construction and multi-user data matching method based on multi-source heterogeneous remote sensing data | |
| Zhong et al. | Landslide mapping with remote sensing: challenges and opportunities | |
| Zhao et al. | Classifying airborne LiDAR point clouds via deep features learned by a multi-scale convolutional neural network | |
| WO2023168781A1 (en) | Soil cadmium risk prediction method based on spatial-temporal interaction relationship | |
| CN114491108B (en) | Online classification system and method based on multi-source remote sensing application data | |
| Buyukdemircioglu et al. | Semi-automatic 3D city model generation from large-format aerial images | |
| Su et al. | A new hierarchical moving curve-fitting algorithm for filtering lidar data for automatic DTM generation | |
| CN104679827A (en) | Big data-based public information association method and mining engine | |
| CN113656647B (en) | Intelligent operation and maintenance oriented engineering archive data management platform, method and system | |
| Wan et al. | A landslide expert system: image classification through integration of data mining approaches for multi-category analysis | |
| CN111598460A (en) | Method, device and equipment for monitoring heavy metal content in soil and storage medium | |
| CN117036939A (en) | Urban functional area identification method based on multi-source data collaboration of graph rolling network | |
| CN113408663B (en) | Fusion model construction method, fusion model using device and electronic equipment | |
| Sen et al. | Unsupervised extraction of urban features from airborne lidar data by using self-organizing maps | |
| Zhou et al. | Pairs autogeo: an automated machine learning framework for massive geospatial data | |
| CN112015937B (en) | Picture geographic positioning method and system | |
| Hui et al. | Automatic morphological filtering algorithm for airborne lidar data in urban areas | |
| CN117093890A (en) | Comprehensive evaluation method for ecological environment of energy resource development area | |
| Itakura et al. | Automatic method for segmenting leaves by combining 2D and 3D image-processing techniques | |
| Wang et al. | Automatic registration framework for multi-platform point cloud data in natural forests | |
| CN115393731A (en) | Method and system for generating virtual cloud picture based on interactive scenario and deep learning | |
| CN114936279A (en) | Unstructured chart data analysis method for collaborative manufacturing enterprise | |
| Doneus et al. | Human‐in‐the‐loop development of spatially adaptive ground point filtering pipelines—An archaeological case study | |
| Tang et al. | Accuracy test of point-based and object-based urban building feature classification and extraction applying airborne LiDAR data | |
| CN111435537B (en) | Model training method and device and pose optimization method and device based on mosaic |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |