A Grid Division Management and Storage Method for Remote Sensing Change Detection Samples

Abstract

The present invention relates to a method for grid division management and storage of remote sensing change detection samples. The method includes two parts: grid division management of remote sensing change detection areas and grid storage of remote sensing change detection samples. The grid coding is used to index and manage the remote sensing change detection sample data, and through the association of grids, the real-time and efficient sharing of samples and learning results is realized. At the same time, it is convenient for location-based monitoring services and rapid information statistics on a large scale, and is also suitable for sensitive remote sensing application services for the public and individuals.

Description

Technical Field

[0001] The invention relates to a grid division management and storage method for remote sensing change detection samples. Background Art

[0002] The existing remote sensing change detection sample storage and management methods lack a unified domain space identification and cannot meet the sample sharing based on domain space. In addition, as the amount of remote sensing change detection sample data will increase in the future, the retrieval efficiency of conventional spatial databases will be low. Summary of the invention

[0003] The purpose of the present invention is to overcome the above-mentioned defects and provide a grid division management and storage method for remote sensing change detection samples. The grid coding can be used to index the remote sensing change detection sample data, and the grid can be associated to achieve real-time and efficient sharing of samples and learning results. At the same time, it is convenient for location-based monitoring services and rapid statistics of large-scale information, and is also suitable for sensitized remote sensing application services for the public and individuals.

[0004] To achieve the above-mentioned purpose, the technical solution of the present invention is: a method for grid division management and storage of remote sensing change detection samples, including grid division management of remote sensing change detection areas and grid storage of remote sensing change detection samples;

[0005] The remote sensing change detection area grid division management is achieved by establishing a secondary grid remote sensing change detection area division mechanism. Specifically,

[0006] The first level of GeoSOT grid division: establish the study area level grid division unit, and assign the study area grid code ID Area at the study area subdivision level n (0≤n<32) GeoSOT , establish multi-level grid storage units including national, provincial, municipal, district, county, and street levels;

[0007] The second level of GeoSOT grid division: establish a sample level grid division unit, and assign the sample grid code ID Sample at the sample subdivision level m (n<m≤32) GeoSOT ;

[0008] The remote sensing change detection samples are stored in a grid, and the spatial location information of the remote sensing change detection grid sample RS_CDGS is converted into a GeoSOT partition code through a GeoSOT spatial coding generation operation, and a grid sample partition index table is established.

[0009] In one embodiment of the present invention, the first level division of the GeoSOT grid is specifically implemented as follows:

[0010] The grid storage units are classified according to administrative divisions. The grid storage units are divided into 5 levels, namely the national level (the 1st level), the provincial level (the 2nd level), the municipal level (the 3rd level), the district / county level (the 4th level), and the sub-district level (the 5th level), which are 5 administrative division levels. A multi-level grid storage unit set of national-level cou, provincial-level pro, municipal-level cit, district / county-level dis, and sub-district-level str is established:

[0011] GridStorageUnitSet j , j ∈ {cou, pro, cit, dis, str}

[0012] Each grid storage unit set GridStorageUnitSet j Each grid storage unit in it corresponds to a grid storage unit space; for an administrative division A at a certain level, according to the multi-scale subdivision algorithm for generating the grid storage unit space, a grid storage unit space corresponding to the corresponding grid storage unit is generated, where, from left to right and from top to bottom, they are the provincial grid storage unit space, the municipal grid storage unit space, the district / county grid storage unit space, and the sub-district grid storage unit space;

[0013] Each grid storage unit space is expressed by grid sets at different levels, that is, a grid storage unit space corresponds to a multi-level grid code set. Let m = {m cou , m pro , m cit , m dis , m str} be the number of grid storage unit spaces corresponding to different grid storage unit levels, be the number of grid codes in the i-th grid storage unit space under the corresponding grid storage unit level.

[0014] In an embodiment of the present invention, the specific implementation of the multi-scale subdivision algorithm for generating the grid storage unit space is as follows:

[0015] Step1. According to the area size of the vector layer of administrative division A, first select a relatively coarse grid storage unit space subdivision level S, and obtain its corresponding grid set G = {G i |G0, G1, G3,..., G n}; Initialize the final grid subdivision set corresponding to administrative division A and the intermediate storage set

[0016] Step2. Traverse the grid set G:

[0017] Step 2.1. Calculate G iThe proportion overlap(A) of the vector layer area of administrative region A;

[0018] Step 2.2: If overlap(A) = 0, then skip Step 2.3 and Step 2.4;

[0019] Step 2.3: If overlap(A) > 0.95, then G final = G final ∪ G i , and skip Step 2.4;

[0020] Step 2.4: Calculate the next-level grid G' of the i corresponding dissection level of the th-level grid G i , G temp = G temp ∪ G' i ;

[0021] Step 3: The dissection level S = S + 1 (S < 26), the dissection grid set G = G temp , the intermediate storage set

[0022] Step 4: Repeat Step 2 and Step 3 until S > 8, or

[0023] Step 5: The final grid dissection set G final = G final ∪ G temp ;

[0024] Step 6: Output G final , as the final grid dissection set of administrative region A, that is, the grid code set corresponding to the grid storage unit space.

[0025] In an embodiment of the present invention, the remote sensing change detection grid sample RS_CDGS includes grid sample data of sample attributes, positions, grid levels, and image resolutions.

[0026] In an embodiment of the present invention, the grid-based storage of remote sensing change detection samples is specifically implemented as follows:

[0027] Associate the grid samples with spatial position information based on the grid coding columns of the study area and the sample grid coding columns of GeoSOT, and use them as the query primary key QPK to obtain the metadata of RS_CDGS; the specific formula expression of QPK is as follows:

[0028]

[0029] Among them, k nm is the number of encodings of RS_CDGS in the database under the secondary grid remote sensing change detection area division mechanism, is the encoding of the i-th research area at the n-th level, is the encoding of the i-th sample at the m-th level;

[0030] Put the JSON file into the database, and save the association relationship (Code, label n ) between the GeoSOT encoding and the grid sample into the large sample dissection index table. RS_CDGS locates to the GeoSOT system according to its header file parameters, obtains the grid sample image through the combination of the RS_CDGS parent path and the sample name, and retrieves the sample Label in the dissection index large table to obtain the complete remote sensing sample; the attribute storage expression formula of the sample dissection index large table is as follows:

[0031]

[0032] Among them, Attribute(QPK) is the attribute information corresponding to QPK, c is the number of attribute columns, imagePath QPK is the main path of RS_CDGS, and INCLUDE() is the defined attribute inclusion operation.

[0033] Compared with the prior art, the present invention has the following beneficial effects: The global dissection grid encoding has spatio-temporal uniqueness, and in any regional range, it can very conveniently automatically associate various types and sizes of entity objects, as well as various events related to the object, with the spatial position where the object is located. The method of the present invention can use the grid encoding to index and manage the remote sensing change detection sample data, associate through the grid, realize the real-time and efficient sharing of samples and learning results, and at the same time facilitate location-based monitoring services and large-scale information rapid statistics, and is also suitable for sensitive remote sensing application services for the public and individuals. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 is a schematic flow chart of the method of the present invention.

[0035] Figure 2 is composed of remote sensing change detection grid samples. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] The technical solution of the present invention will be specifically described below with reference to the accompanying drawings.

[0037] As Figure 1As shown in the figure, the present invention provides a method for grid division management and storage of remote sensing change detection samples, including grid division management of remote sensing change detection areas and grid storage of remote sensing change detection samples. The specific implementation of the method of the present invention is as follows:

[0038] 1. Grid division management of remote sensing change detection areas

[0039] The GeoSOT grid coding provides data index support for the iteration of the grid deep learning model. The sub-divided remote sensing samples within the same grid correspond to a unique coding ID, realizing the query statistics and real-time sharing of samples and learning results. Moreover, the GeoSOT binary extended coding can provide more efficient services relying on its advantage of fast retrieval.

[0040] We establish a two-level grid division mechanism for remote sensing change detection areas.

[0041] (1) The first-level division of GeoSOT grids: Establish grid division units at the research area level. Under the hierarchical level n (0 ≤ n < 32) of the research area, assign the grid coding ID Area to the research area grid GeoSOT . That is, establish multi-level grid storage units at multiple levels such as national, provincial, municipal, district / county, and street levels. Samples based on the research area grid coding are directly accessible, realizing efficient retrieval; at the same time, the grid location expression avoids the ambiguity of one address with multiple names.

[0042] Each grid storage unit GridStorageUnit corresponds to a grid storage unit space S_GridStorageUnit, and each grid storage unit space S_GridStorageUnit corresponds to a grid code set G_CodeSet = {G_Code1, G_Code2, G_Code3,...}. Therefore, there is a mapping relationship between the grid storage unit and the grid code set G_CodeSet. The grid storage unit space S_GridStorageUnit serves as a spatial location middleware to undertake the functions of spatial identification and coding identification. There is the following mapping relationship between the grid storage unit and the grid code set.

[0043]

[0044] We classify the grid storage units according to the administrative divisions. The grid storage units are divided into 5 levels, namely national level (the 1st level), provincial level (the 2nd level), municipal level (the 3rd level), district / county level (the 4th level), and street level (the 5th level), 5 administrative division levels, and establish multi-level grid storage unit sets such as national level (cou), provincial level (pro), municipal level (cit), district / county level (dis), and street level (str)

[0045] Grid Storage Unit Set j where \(j\in\{cou, pro, cit, dis, str\}\)

[0046] Each level of the Grid Storage Unit Set j Each grid storage unit corresponds to a grid storage unit space. Table 1 shows the multi-scale dissection algorithm for generating the grid storage unit space of the grid storage units in administrative division A at a certain level, based on the grid storage unit space division units established according to administrative divisions.

[0047] Table 1 Multi-scale dissection algorithm for generating grid storage unit space

[0048]

[0049] According to the multi-scale dissection algorithm for generating grid storage unit space, we generate the grid storage unit space corresponding to this grid storage unit; among them, from left to right and from top to bottom are the provincial grid storage unit space, the municipal grid storage unit space, the district / county-level grid storage unit space, and the street-level grid storage unit space respectively.

[0050] Each grid storage unit space is expressed by grid sets at different levels, that is to say, a grid storage unit space corresponds to a multi-level grid code set. Table 2 shows the corresponding relationship between the grid storage unit levels, the grid storage unit space sets, and the grid code sets, where \(m = \{m cou , m pro , m cit , m dis , m str \}\) is the number of grid storage unit spaces corresponding to different grid storage unit levels, where is the number of grid codes in the \(i\)-th grid storage unit space at the corresponding grid storage unit level.

[0051] Table 2 Corresponding table of grid storage unit levels, grid storage unit space sets, and grid code sets

[0052]

[0053] (2) Second-level division of GeoSOT grid: Establish sample-level grid division units. At the sample dissection level \(m\) (\(n \lt m\leq32\)), assign the sample grid code ID Sample GeoSOT .

[0054] 2. Grid-based storage of remote sensing change detection samples

[0055] Through the operation of generating GeoSOT spatial coding, the spatial location information of RS_CDGS is converted into GeoSOT subdivision coding, and a large table of grid sample subdivision indexes is established. The remote sensing change detection grid sample (RS_CDGS) includes grid sample data such as sample attributes, locations, grid levels, image resolutions, etc., as Figure 2 shown.

[0056] Based on the grid coding column of the study area and the sample grid coding column of GeoSOT, the grid samples with spatial location information are associated and used as the query primary key (QPK) to obtain the metadata of RS_CDGS. The specific formula expression of QPK is as follows.

[0057]

[0058] Among them, k nm is the number of encodings of RS_CDGS in the database under the secondary grid remote sensing change detection area division mechanism, is the i-th study area code at the n-th level, is the i-th sample code at the m-th level.

[0059] JSON (JavaScript Object Notation) is a lightweight data exchange format with good readability and scalability, and has great advantages in processing spatial data such as RS CDGS. We store the JSON file in the database and save the association relationship (Code, label n ) between the GeoSOT coding and the grid samples in the large table of sample subdivision indexes. RS_CDGS is located in the GeoSOT system according to its header file parameters, and the grid sample image is obtained through the combination of the RS_CDGS parent path and the sample name, and the sample Label is retrieved in the subdivision index large table to obtain the complete remote sensing sample. The attribute storage expression formula of the sample subdivision index large table is as follows.

[0060]

[0061] Among them, Attribute(QPK) is the attribute information corresponding to QPK, c is the number of attribute columns, and imagePath QPK is the main path of RS_CDGS, and INCLUDE() is the defined attribute inclusion operation.

[0062] The above are the preferred embodiments of the present invention. All changes made according to the technical solution of the present invention, when the functions and effects generated do not exceed the scope of the technical solution of the present invention, shall fall within the protection scope of the present invention.

Claims

1. A grid division management and storage method for remote sensing change detection samples, characterized in that, It includes grid-based management of remote sensing change detection areas and grid-based storage of remote sensing change detection samples; The grid-based management of remote sensing change detection areas is achieved by establishing a two-level grid-based remote sensing change detection area division mechanism. Specifically, The first-level division of the GeoSOT grid: Establish grid division units at the research area level. Under the hierarchical level n of the research area, 0 ≤ n < 32, and assign the grid code ID Area to the research area grid GeoSOT , and establish multi-level grid storage units including multiple levels of country, province, city, county, district, and street Second-level division of the GeoSOT grid: Establish sample-level grid division units. Under the sample section level m, where n < m ≤ 32, assign the sample grid code ID Sample GeoSOT ; For the grid-based storage of remote sensing change detection samples, through the GeoSOT spatial coding generation operation, the spatial location information of the remote sensing change detection grid samples RS_CDGS is converted into GeoSOT dissection codes, and a large table of grid sample dissection indexes is established; The specific implementation method of the first-level division of the GeoSOT grid is as follows: The grid storage units are classified according to the administrative divisions. The grid storage units are divided into 5 levels, namely the national level (the 1st level), the provincial level (the 2nd level), the municipal level (the 3rd level), the district / county level (the 4th level), and the street level (the 5th level), a total of 5 administrative division levels. A multi-level set of grid storage units of national cou, provincial pro, municipal cit, district / county dis, and street str is established: Grid Storage Unit Set j , j ∈ {cou, pro, cit, dis, str} Each level of grid storage unit set GridStorageUnitSet j Each grid storage unit therein corresponds to a grid storage unit space; For an administrative division A at a certain level, according to the multi-scale dissection algorithm for generating the grid storage unit space, the grid storage unit space corresponding to the corresponding grid storage unit is generated. Among them, from left to right and from top to bottom are the provincial grid storage unit space, the municipal grid storage unit space, the district / county grid storage unit space, and the street grid storage unit space; Each grid storage unit space is represented by a set of grids at different levels, that is, a grid storage unit space corresponds to a multi-level grid code set. Let \(m = \{m cou , m pro , m cit , m dis , m str}\) be the number of grid storage unit spaces corresponding to different grid storage unit levels, and \(n_{i}\) be the number of grid codes in the \(i\)-th grid storage unit space at the corresponding grid storage unit level. The specific implementation of the multi-scale dissection algorithm for generating the grid storage unit space is as follows: Step1. According to the area size of the vector layer of administrative region A, first select a relatively coarse grid storage unit spatial subdivision level S, and obtain its corresponding grid set G = {G i | G0, G1, G3,..., G n}; Initialize the final grid subdivision set corresponding to administrative region A and the intermediate storage set Step2. Traverse the grid set G: Step 2.

1. Calculate G i The proportion overlap(A) of the vector layer area of administrative region A in i ; Step 2.

2. If overlap(A) = 0, then skip Step 2.3 and Step 2.4; Step 2.

3. If overlap(A) > 0.95, then G final = G final ∪ G i , and skip Step 2.4; Step 2.

4. Calculate G i The corresponding hierarchical level of dissection The next level of the Level i grid G' temp where G temp = G i ∪ G' Step 3, the subdivision level S = S + 1, S < 26, the subdivision grid set G = G temp , intermediate storage collection Step 4. Repeat Step 2 and Step 3 until S > 8, or Step5, Final mesh division set G final = G final ∪G temp ; Step6. Output G final , as the final grid division set of administrative division A, that is, the grid code set corresponding to the grid storage unit space.

2. The grid division management and storage method for remote sensing change detection samples according to claim 1, characterized in that, The remote sensing change detection grid sample RS_CDGS includes grid sample data of sample attributes, positions, grid levels, and image resolutions.

3. The grid division management and storage method for remote sensing change detection samples according to claim 1, characterized in that, The specific implementation of the grid-based storage of remote sensing change detection samples is as follows: Based on the grid coding column of the study area and the sample grid coding column of GeoSOT, the grid samples with spatial location information are associated, and used as the query primary key QPK to obtain the metadata of RS_CDGS. The specific formula expression of QPK is as follows: where k nm is the number of encodings of RS_CDGS in the database under the second-level grid remote sensing change detection area division mechanism, is the encoding of the i-th study area at the n-th level, is the encoding of the i-th sample at the m-th level; Store the JSON file in the database, and save the association relationship (Code, label n ) between the GeoSOT code and the grid sample into the large table of sample dissection index. RS_CDGS locates to the GeoSOT system according to its header file parameters, obtains the grid sample image by combining the RS_CDGS parent path and the sample name, and retrieves the sample Label in the large table of dissection index to obtain the complete remote sensing sample. The attribute storage expression formula of the large table of sample dissection index is as follows: Among them, Attribute(QPK) is the attribute information corresponding to QPK, c is the number of attribute columns, and imagePath QPK is the main path of RS_CDGS, and INCLUDE() is the defined attribute inclusion operation.

Images