Compact tile cutting method suitable for multi-source heterogeneous global remote sensing image

Abstract

The invention relates to a compact tile cutting method suitable for a multi-source heterogeneous global remote sensing image. The compact tile cutting method comprises the following steps: step 1, tile task division; 2, data preprocessing; step 3, tile cutting; step 4, tile merging; and step 5, data storage and archiving. According to the compact tile cutting method suitable for the multi-source heterogeneous global remote sensing image, the multi-source heterogeneous global remote sensing image data is cut into the tile blocks, and the tile blocks are stored as compact cache format files, so that efficient processing and storage management of the remote sensing image data are realized. According to the cutting technology, the cutting speed can be greatly increased, the occupied storage space is reduced, the data retrieval speed is increased, and the remote sensing image processing and using efficiency is improved.

Description

Technical Field

[0001] The invention relates to the technical field of remote sensing information technology applications, and in particular to a compact tile cutting method suitable for multi-source heterogeneous global remote sensing images. Background Art

[0002] With the rapid development of space-based earth observation technology, my country has gradually formed an all-weather, global earth observation capability, and established a standard and complete space remote sensing and flexible and diverse aviation remote sensing data acquisition system. The implementation of high-resolution earth observation systems and the commissioning of commercial remote sensing satellite constellations have greatly enriched remote sensing data resources. Since 2014, the amount of remote sensing data has increased rapidly, with the amount of data growing at the PB level and expanding at a faster rate, especially the multi-phase global one-picture image data taken by the Jilin-1 constellation, which includes multiple models of satellites and has huge data volume. Although massive remote sensing data can reflect ground object information in multiple granularities, multiple phases, multiple directions and multiple levels, its data volume is increasing day by day and its types are increasing day by day. Downstream users are limited by network transmission bandwidth, local storage space and parallel computing capabilities, and it is difficult to quickly browse and interact with diverse geographic spatial data, which greatly limits the mining and application of high-value information from massive remote sensing data.

[0003] Limited by the size and volume of remote sensing image data files, in order to realize the browsing and use of remote sensing image data across various computing terminals, a storage management and publishing method different from the traditional image file has been derived. Among them, the product-customized data service model is based on the cyberspace service standard, provides a unified data publishing, access and processing interface, and migrates data processing to the remote server for execution, solving the problems of interactive operations and limited local resources. This OGC standard is the WMS (Web Map Service) service. Although the product-customized service model alleviates the pressure of local computing through computing migration, with the increase of different versions, batches, and categories of data sets, more service processes consume server CPU and memory resources sharply, causing process congestion and service jams. Therefore, this model still does not solve the problem of management and use of massive data.

[0004] On the basis of WMS service, OGC has launched WMTS protocol, which is the slice publishing method. Slice publishing is mainly to solve the key problems of services like WMS when forming maps, such as serious time-consuming program IO and network transmission, and large CPU resources occupied by data rendering map images. According to the coordinate system and projection, the cache is rendered in advance on the server, and then the entire remote sensing image data is traversed and cut into images indexed by x, y (row and column numbers), z / l (tile level) according to different levels (pyramid principle), and stored in the server storage medium. Each time the user visits, the current geographic range is mapped to the sequence index of the xyz image. After receiving the returned image from the server, the client can organize and display it in a predefined way. This processing greatly improves the user's experience of browsing remote sensing images. For the client, the server publishes the remote sensing image data as a service, and the client only needs to request the corresponding image tile, which is the best technical solution at this stage. However, the slicing solution is extremely dependent on the processing speed and data storage method under the current situation of massive data. How to improve the slicing speed and storage efficiency has become the key. Summary of the invention

[0005] The present invention aims to solve the technical problems in the prior art and provide a compact tile cutting method suitable for multi-source heterogeneous global remote sensing images.

[0006] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0007] A compact tile cutting method applicable to multi-source heterogeneous global remote sensing images includes the following steps:

[0008] Step 1: Tile task division;

[0009] The slicing task strategy first divides the world into grids according to the projection, and divides it into a certain number of levels according to the level of the data. Each grid corresponds to a final compact cache file. After obtaining the original TIFF data files of multi-source heterogeneous global remote sensing images, each TIFF data file corresponds to a task, and the TIFF data file is cut into one or more grids according to the range.

[0010] Step 2: Data preprocessing;

[0011] Standardize the input multi-source heterogeneous global remote sensing image data;

[0012] Step 3: Tile cutting;

[0013] After data preprocessing, the input multi-source heterogeneous global remote sensing image data is sliced ​​and cut into intermediate files according to the format, coordinate system and projection information of the image and the corresponding vector. The information corresponding to the image in the vector file is read and saved in the database to build a spatiotemporal information index.

[0014] Step 4: Tile merging;

[0015] According to predefined rules and algorithms, the overlapping areas between adjacent tiles are matched and spliced ​​to generate a seamless large image; through the tile merging algorithm, according to the predefined grid division, the tiles under the same grid are stored in the same compact file;

[0016] Step 5: Data storage and archiving;

[0017] Data information is stored in the database or storage system in a specific organization and format by defining data storage structure and rules, and unique identifiers and metadata information are assigned to the data.

[0018] In the above technical solution, the standardization processing in step 2 includes: image format processing, coordinate system correction, image band verification, and image resolution verification.

[0019] In the above technical solution, step three is specifically as follows:

[0020] Cutting TIFF data files into tile images one by one involves the level and encoding of tiles. Through the correspondence between projection and pixels, pixels are sampled from the TIFF data file to form a tile image.

[0021] In the above technical solution, the tile image includes:

[0022] Tile layer numbering: The tile pyramid is set to 18-22 layers, each layer has a number, and the layer number is represented by l. The value range of l is an integer in the interval [1, 22]. Level 1 is the top layer and level 22 is the bottom layer.

[0023] Tile column number: given level l, the total number of columns in this layer is represented by Cl, with Cl = 2 l , the tile column number starts from 0, and the range of the level l column number is [0, 2 l –1];

[0024] Tile row numbering: Given a level l, the total number of rows in that layer is represented by Rl, and Rl = 2 l–1 The row number of the tile starts from 0, and the range of the row number of level l is [0, 2 l–1 -1];

[0025] Tile encoding: The encoding of a tile is a triple consisting of level and row and column number (l, col, row), where l represents the level, col represents the column number, and row represents the row number;

[0026] Longitude and latitude bounding box: For a certain geographic space range, a rectangular longitude and latitude bounding box is used to represent it. The bounding box is a four-tuple (Xmin, Ymin, Xmax, Ymax), where Xmin is the longitude of the point at the lower left corner of the bounding box, Ymin is the latitude of the point at the lower left corner of the bounding box, Xmax is the longitude of the point at the upper right corner of the bounding box, and Ymax is the latitude of the point at the upper right corner of the bounding box;

[0027] Conversion of tile encoding to latitude and longitude bounding box: Given a tile encoding (l, col, row), calculate its corresponding latitude and longitude bounding box (Xmin, Ymin, Xmax, Ymax), where:

[0028]

[0029] Conversion of geographic coordinates to the code of the tile to which they belong: Given a certain latitude and longitude coordinate (x, y) and layer number l, get the code of the tile to which the geographic location belongs (l, col, row), where:

[0030]

[0031] Tile pixel size: The pixel size of each tile is the same, with a width and height of 256 pixels. The total number of pixels contained in each tile is represented by P, and P = 256*256;

[0032] The image data contains N×N pixels, where N=2k, corresponding to level k, and the missing pixels are filled with null values; assuming that every 2×2 pixels in the k layer are synthesized into one pixel, a k-1 level image is generated, and so on, until a level 0 image is generated, which is 1 pixel; the sizes of the k-1 pixel matrices are N / 2×N / 2, N / 4×N / 4, and 1×1, respectively; in addition to the original data, the number of pixels M in the resampled pyramid is:

[0033]

[0034] According to the corresponding position of the tile pixels and the longitude and latitude projection, the pixels in the image are written into the tile. When cutting, the image resolution of the TIFF data file is correspondingly cut to the adaptation level;

[0035] Sample the previous level of the adaptation level and save the two levels of images to an intermediate file named after the grid to complete the image-to-tile cutting.

[0036] In the above technical solution, step 4 is specifically as follows:

[0037] The sampling process is as follows: first, obtain the 3-level original remote sensing image, divide it into 3 blocks to obtain 32 tiles in total, then resample the 3-level tiles in groups of four to generate 2-level tiles, and finally resample the 2-level tiles in groups of four to generate 1-level tiles;

[0038] The tile resampling process is as follows: Assuming that the code of the output tile is (l-1, col, row), and the codes of the four input tiles are T00, T10, T01, and T11, the codes of the four input tiles are calculated as follows:

[0039] T00 = (l, 2*col, 2*row)

[0040] T10 = (l, 2*col+1, 2*row)

[0041] T01 = (l, 2*col, 2*row+1)

[0042] T11 = (l, 2*col+1, 2*row+1)

[0043] According to the code (Li, Ci, Ri) of a tile in layer l, l>=2, the code (Li-1, Ci-1, Ri-1) of the tile generated when it is used as one of the four input tiles of the tile resampling operation is calculated, where:

[0044] L i-1 =L i -1,

[0045] Merge the data in the same grid into one grid and perform sampling to complete the merging of tile data.

[0046] The present invention has the following beneficial effects:

[0047] The compact tile cutting method for multi-source heterogeneous global remote sensing images of the present invention cuts multi-source heterogeneous global remote sensing image data into tile blocks and saves them as compact cache format files, thereby realizing efficient processing and storage management of remote sensing image data. This cutting technology can greatly improve the cutting speed, reduce the storage space occupied, speed up the data retrieval speed, and improve the efficiency of remote sensing image processing and use. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0049] Figure 1 Schematic diagram of task division.

[0050] Figure 2Schematic diagram of the latitude and longitude bounding box.

[0051] Figure 3 Schematic diagram of the secondary decomposition strategy for data.

[0052] Figure 4 Schematic diagram of the resampling process built for a remote sensing image tile pyramid.

[0053] Figure 5 Schematic diagram of the resampling operation.

[0054] Figure 6 A schematic diagram of the merger process.

[0055] Figure 7 The present invention is a schematic flow chart of the steps of a compact tile cutting method applicable to multi-source heterogeneous global remote sensing images.

[0056] Figure 8 and Fig. 9 The division of the global grid when cutting using the compact tile cutting method applicable to multi-source heterogeneous global remote sensing images of the present invention is respectively: Figure 8 It is a pyramid design. Fig. 9 It is a grid design solution that can generate compact files at different levels and cut accordingly.

[0057] Fig.10 and Fig.11 The compact tile cutting method for multi-source heterogeneous global remote sensing images of the present invention is used to cut images and save them as intermediate files, wherein Fig.10 It is a slice server that distributes tasks to remote sensing images and collaborates with multiple terminals. Fig.11 It is the slicing client that receives tasks and performs slicing processing.

[0058] Fig.12 and Fig.13 The compact tile cutting method for multi-source heterogeneous global remote sensing images of the present invention is used to merge the slice intermediate files, and whether to perform image infill is selected according to the settings, wherein Fig.12 Merge the server, distribute tasks to slice intermediate files, and collaborate with multiple terminals. Fig.13 It is a merge client that accepts tasks and performs merge processing.

[0059] Fig.14 and Fig.15 are respectively compact files generated by merging the compact tile cutting method applicable to multi-source heterogeneous global remote sensing images of the present invention, wherein Fig.14 It is a Mercator projection file with 10-level grid division. Fig.15 is the file content.

[0060] Fig.16 , Fig.17 , Fig.18 and Fig.19 These are the effects of loading slice data on different GIS platforms.

[0061] Fig. 20 This is the flow chart of the image correction subsystem.

[0062] Fig.21 This is the flow chart of the data slicing subsystem.

[0063] Fig. 22 The flowchart of the data merging subsystem is shown in the figure.

[0064] Fig.23 Flowchart of the data warehousing and archiving subsystem. DETAILED DESCRIPTION

[0065] The inventive concept of the present invention is:

[0066] In view of the above-mentioned processing problems of massive high-resolution remote sensing satellite spatiotemporal data, especially the processing of global remote sensing data, the present invention proposes an economical, practical, horizontally scalable, high-performance compact tile cutting method. Aiming at the rapid information service needs of multi-source heterogeneous global high-resolution remote sensing satellite data, based on the technical main line of "parallel slicing-dynamic storage", a set of standardized remote sensing data compact tile cutting technology is formed. Through tile cutting, merging, and storage strategies, it can support the rapid cutting of multi-source heterogeneous remote sensing data, further improve the efficiency of data cutting and use, effectively solve the bottleneck problem of remote sensing data application caused by the contradiction between massive data and limited network transmission speed, greatly improve the utilization rate of remote sensing data service resources and processing concurrency efficiency, and realize the rapid cutting and efficient storage of multi-source heterogeneous global high-resolution remote sensing satellite data.

[0067] The present invention is described in detail below with reference to the accompanying drawings.

[0068] The compact tile cutting method applicable to multi-source heterogeneous global remote sensing images of the present invention comprises the following steps:

[0069] Step 1: Tile task division.

[0070] The slicing task strategy first divides the world into grids according to the projection. According to the level of the data, it can be divided into 10, 12, and 14 levels. Each grid corresponds to a final compact cache file. After obtaining the original TIFF data files of multi-source heterogeneous global remote sensing images, each TIFF data file corresponds to a task. This TIFF data file will be cut into one or more grids according to the range, such as Figure 1 shown.

[0071] Taking the 10-level grid and longitude and latitude projection as an example, the total number of grids in the world is about 500,000, which corresponds to about 130,000 on land. Compared with the number of tiles, it is greatly reduced and the task division is faster.

[0072] Step 2: Data preprocessing.

[0073] Standardize the input multi-source heterogeneous global remote sensing image data, including image format processing, coordinate system correction, image band verification, and image resolution verification.

[0074] Step 3: Tile cutting.

[0075] After data preprocessing, the input multi-source heterogeneous global remote sensing image data is sliced. According to the format, coordinate system, projection and other information of the image and the corresponding vector, the image is cut into intermediate files, and the information corresponding to the image in the vector file is read and saved in the database to build a spatiotemporal information index.

[0076] Tile cutting is to cut TIFF data files into tile images one by one, which involves the level and encoding of tiles. Through the correspondence between projection and pixels, pixels are sampled from TIFF data files to form tile images. The details are as follows (taking longitude and latitude projection as an example):

[0077] Tile layer numbering: The tile pyramid can be divided into 18-22 layers, each layer has a number, and the layer number is represented by l. The value range of l is an integer in the interval [1, 22]. Level 1 is the top layer (the least number of tiles, 2 for longitude and latitude projection, and 1 for Mercator projection), and level 22 is the bottom layer (the most tiles).

[0078] Tile column number: Given a level l, the total number of columns in this layer is denoted by Cl, we have Cl = 2 l , the column number of the tile starts from 0, so the range of the column number of level l is [0, 2 l –1].

[0079] Tile row numbering: Given a level l, the total number of rows in that layer is denoted by Rl, and we have Rl = 2 l–1 Similarly, the row number of the tile also starts from 0, so the range of the row number of level l is [0, 2 l–1 -1].

[0080] Tile encoding: The tile encoding is a triple consisting of level plus row and column number (l, col, row), where l represents the level, col represents the column number, and row represents the row number.

[0081] Longitude and latitude bounding box: For a certain geographic space range, a rectangular longitude and latitude bounding box is used to represent it, such as Figure 2As shown, the bounding box is a four-tuple (Xmin, Ymin, Xmax, Ymax),

[0082] Among them, Xmin is the longitude of the point in the lower left corner of the bounding box, Ymin is the latitude of the point in the lower left corner of the bounding box, Xmax is the longitude of the point in the upper right corner of the bounding box, and Ymax is the latitude of the point in the upper right corner of the bounding box.

[0083] Conversion of tile encoding to latitude and longitude bounding box: Given a tile encoding (l, col, row), we can easily calculate its corresponding latitude and longitude bounding box (Xmin, Ymin, Xmax, Ymax), where:

[0084]

[0085] Conversion of geographic coordinates to the code of the tile to which they belong: Given a certain latitude and longitude coordinate (x, y) and layer number l, get the code of the tile to which the geographic location belongs (l, col, row), where:

[0086]

[0087] Tile pixel size: The pixel size of each tile is the same, with a width and height of 256 pixels. The total number of pixels contained in each tile is represented by P, and P = 256*256.

[0088] The image data contains N×N pixels, where N=2k, corresponding to level k, and the insufficient pixels can be filled with null values. Assume that every 2×2 pixels in the k layer are synthesized into one pixel to generate the k-1 level image, and so on, until the 0 level image is generated, which is 1 pixel. Among them, the sizes of the k-1 pixel matrices are N / 2×N / 2, N / 4×N / 4 and 1×1 respectively. In addition to the original data, the number of pixels M of the resampled pyramid is:

[0089]

[0090] like Figure 3 As shown, according to the corresponding position of the pixels of the tile and the longitude and latitude projection, the pixels in the image are written into the tile. When cutting, the image resolution corresponding to the TIFF data file is cut to the adaptation level. At the same time, the upper level of the adaptation level is sampled, and the images of the two levels are saved in the intermediate file named after the grid, thus completing the cutting of the image to the tile.

[0091] Step 4: Merge tiles.

[0092] After slicing, the tile images that have been split into small pieces are merged. According to predefined rules and algorithms, the overlapping areas between adjacent tiles are matched and spliced ​​to generate a seamless large image. Through an efficient tile merging algorithm and according to predefined grid division, tiles under the same grid are stored in the same compact file.

[0093] The construction process of the remote sensing image tile pyramid is actually an iterative segmentation and resampling process. When slicing, only the resolution level corresponding to the TIFF data file is sampled, and low-level tiles are sampled when merging.

[0094] Sampling process, such as Figure 4 As shown in the figure, firstly, the 3-level original remote sensing image is obtained, which is divided into 32 tiles in total. Then, based on the 3-level tiles, four tiles are resampled to generate 2-level tiles. Finally, based on the 2-level tiles, four tiles are resampled to generate 1-level tiles. Note that in the resampling process, any tile of level 2 is resampled by four fixed tiles of level 3. For example, the 0-column 0-row tile of level 2 (2, 0, 0) is resampled by four tiles of level 3: 0-column 0-row, 1-column 0-row, 0-column 1-row, and 1-column 1-row. The whole resampling process is as follows: Figure 4 shown.

[0095] Tile resampling: The input of the tile resampling operation is four adjacent tiles in layer l (l>=2). Only when the input conditions are met (that is, all four input tiles exist), the corresponding resampling operation can be performed during the tile pyramid construction process. The output of tile resampling is a tile at level l-1. The encoding of the input tile and the encoding of the output tile of the tile resampling operation have a certain correspondence. Figure 5 As shown, assuming that the code of the output tile is (l-1, col, row), and the codes of the four input tiles are T00, T10, T01, and T11, the codes of the four input tiles can be inferred, and the calculation formula is as follows:

[0096] T00 = (l, 2*col, 2*row)

[0097] T10 = (l, 2*col+1, 2*row)

[0098] T01 = (l, 2*col, 2*row+1)

[0099] T11 = (l, 2*col+1, 2*row+1)

[0100] Conversely, the code (Li-1, Ci-1, Ri-1) of a tile generated when it is used as one of the four input tiles of the tile resampling operation can be inferred according to the code (Li, Ci, Ri) of a tile in layer l (l>=2), where:

[0101] L i-1 =L i -1,

[0102] The input tile of the tile resampling operation is called the parent tile of the output tile; conversely, the output tile is called the child tile of the input tile, so one child tile corresponds to four parent tiles.

[0103] like Figure 6 As shown, the data in the same grid are merged into one grid and sampled, thus completing the merging of tile data.

[0104] Step 5: Data storage and archiving.

[0105] After the data is merged, the data information is put into storage for management. By defining the data storage structure and rules, the data is stored in the database or storage system in a specific organization and format, and a unique identifier and metadata information are assigned to it.

[0106] Through the data storage and archiving subsystem, large amounts of data can be easily managed and retrieved, ensuring data security and traceability, achieving long-term storage and effective use of data, and more conveniently managing the update and replacement of slice data.

[0107] The compact tile cutting method applicable to multi-source heterogeneous global remote sensing images of the present invention is further described in detail below in conjunction with the accompanying drawings.

[0108] like Figure 7-23 As shown, the compact tile cutting method applicable to multi-source heterogeneous global remote sensing images of the present invention is applied, and the implementation process of multi-source heterogeneous global remote sensing image slicing is described as follows:

[0109] (1) Tile task division: Figure 8 and 9 As shown in the figure, the slicing task strategy first divides the world into grids according to the projection. According to the level of the data, it can be divided into a certain number of levels, and each grid corresponds to a final compact cache file; after obtaining the original TIFF data files of multi-source heterogeneous global remote sensing images, each TIFF data file corresponds to a task, and the TIFF data file is cut into one or more grids according to the range.

[0110] (2) Data preprocessing: Obtain original multi-source heterogeneous global remote sensing image data. Preprocess the original data, such as Fig. 20As shown, it includes: correction, denoising, radiation calibration, geometric correction and other operations to ensure data quality and accuracy.

[0111] (3) Data slicing: The preprocessed multi-source heterogeneous global remote sensing images are cut into small pieces, usually rectangular areas of fixed size, called slices. Fig.10 , 11 and Fig.21 As shown, the slices are divided according to the grid or specific requirements to adapt to different application scenarios and needs. Parallel computing technology is used to process tile data in parallel to accelerate data processing and analysis. Tile blocks are compressed to reduce storage space and improve data transmission efficiency.

[0112] (4) Data merging and remapping: For large-scale, multi-source, heterogeneous global remote sensing image data, data merging and remapping operations may be required. Figure 12-19 as well as Fig. 22 As shown, the sliced ​​data is merged according to certain rules to fill the missing areas to generate complete image data. The storage method of the tile block is a compact cache interface to improve data storage efficiency and access speed.

[0113] (5) Data storage and archiving: The multi-source heterogeneous global remote sensing image data after slicing and merging is stored in a compact cache file for archiving and management. Fig.23 As shown, the database can use a relational database, an object storage or a distributed file system to support efficient data storage and retrieval. An index structure is established for the tile blocks obtained by division. The tile blocks can contain metadata information, such as location, size, timestamp, etc., to quickly retrieve and access specific tiles.

[0114] The compact tile cutting method for multi-source heterogeneous global remote sensing images of the present invention cuts multi-source heterogeneous remote sensing image data into tile blocks and saves them as compact cache format files, thereby realizing efficient processing and storage management of remote sensing image data. This cutting technology can greatly increase the cutting speed, reduce the storage space occupied, speed up the data retrieval speed, and improve the efficiency of remote sensing image processing and use.

[0115] Obviously, the above embodiments are merely examples for the purpose of clear explanation, and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the invention.

Claims

1. A compact tile cutting method suitable for multi-source heterogeneous global remote sensing images, characterized in that: The following steps are involved: Step 1: Tile task division; The slicing task strategy first divides the world into grids according to the projection, and divides it into a certain number of levels according to the level of the data. Each grid corresponds to a final compact cache file. After obtaining the original TIFF data files of multi-source heterogeneous global remote sensing images, each TIFF data file corresponds to a task, and the TIFF data file is cut into one or more grids according to the range. Step 2: Data preprocessing; Standardize the input multi-source heterogeneous global remote sensing image data; Step 3: Tile cutting; After data preprocessing, the input multi-source heterogeneous global remote sensing image data is sliced ​​and cut into intermediate files according to the format, coordinate system and projection information of the image and the corresponding vector. The information corresponding to the image in the vector file is read and saved in the database to build a spatiotemporal information index. Step 4: Tile merging; According to predefined rules and algorithms, the overlapping areas between adjacent tiles are matched and spliced ​​to generate a seamless large image; through the tile merging algorithm, according to the predefined grid division, the tiles under the same grid are stored in the same compact file; Step 5: Data storage and archiving; Data information is stored in the database or storage system in a specific organization and format by defining data storage structure and rules, and unique identifiers and metadata information are assigned to the data.

2. The compact tile cutting method applicable to multi-source heterogeneous global remote sensing images according to claim 1 is characterized in that: The standardization processing in step 2 includes: image format processing, coordinate system correction, image band verification, and image resolution verification.

3. The compact tile cutting method applicable to multi-source heterogeneous global remote sensing images according to claim 1 is characterized in that: Step three is as follows: Cutting TIFF data files into tile images one by one involves the level and encoding of tiles. Through the correspondence between projection and pixels, pixels are sampled from the TIFF data file to form a tile image.

4. The compact tile cutting method applicable to multi-source heterogeneous global remote sensing images according to claim 3 is characterized in that: Tile images include: Tile layer numbering: The tile pyramid is set to 18-22 layers, each layer has a number, and the layer number is represented by l. The value range of l is an integer in the interval [1, 22]. Level 1 is the top layer and level 22 is the bottom layer. Tile column number: given level l, the total number of columns in this layer is represented by Cl, with Cl = 2 l , the tile column number starts from 0, and the range of the level l column number is [0, 2 l –1]; Tile row numbering: Given a level l, the total number of rows in that layer is represented by Rl, and Rl = 2 l–1 The row number of the tile starts from 0, and the range of the row number of level l is [0, 2 l–1 -1]; Tile encoding: The encoding of a tile is a triple consisting of level and row and column number (l, col, row), where l represents the level, col represents the column number, and row represents the row number; Longitude and latitude bounding box: For a certain geographic space range, a rectangular longitude and latitude bounding box is used to represent it. The bounding box is a four-tuple (Xmin, Ymin, Xmax, Ymax), where Xmin is the longitude of the point at the lower left corner of the bounding box, Ymin is the latitude of the point at the lower left corner of the bounding box, Xmax is the longitude of the point at the upper right corner of the bounding box, and Ymax is the latitude of the point at the upper right corner of the bounding box; Conversion of tile encoding to latitude and longitude bounding box: Given a tile encoding (l, col, row), calculate its corresponding latitude and longitude bounding box (Xmin, Ymin, Xmax, Ymax), where: Conversion of geographic coordinates to the code of the tile to which they belong: Given a certain latitude and longitude coordinate (x, y) and layer number l, get the code of the tile to which the geographic location belongs (l, col, row), where: Tile pixel size: The pixel size of each tile is the same, with a width and height of 256 pixels. The total number of pixels contained in each tile is represented by P, and P = 256*256; The image data contains N×N pixels, where N=2k, corresponding to level k, and the missing pixels are filled with null values; assuming that every 2×2 pixels in the k layer are synthesized into one pixel, a k-1 level image is generated, and so on, until a level 0 image is generated, which is 1 pixel; the sizes of the k-1 pixel matrices are N / 2×N / 2, N / 4×N / 4, and 1×1, respectively; in addition to the original data, the number of pixels M in the resampled pyramid is: According to the corresponding position of the tile pixels and the longitude and latitude projection, the pixels in the image are written into the tile. When cutting, the image resolution of the TIFF data file is correspondingly cut to the adaptation level; Sample the previous level of the adaptation level and save the two levels of images to an intermediate file named after the grid to complete the image-to-tile cutting.

5. The compact tile cutting method applicable to multi-source heterogeneous global remote sensing images according to claim 1, characterized in that: Step 4 is as follows: The sampling process is as follows: first, obtain the 3-level original remote sensing image, divide it into 3 blocks to obtain 32 tiles in total, then resample the 3-level tiles in groups of four to generate 2-level tiles, and finally resample the 2-level tiles in groups of four to generate 1-level tiles; The tile resampling process is as follows: Assuming that the code of the output tile is (l-1, col, row), and the codes of the four input tiles are T00, T10, T01, and T11, the codes of the four input tiles are calculated as follows: T00 = (l, 2*col, 2*row) T10 = (l, 2*col+1, 2*row) T01 = (l, 2*col, 2*row+1) T11 = (l, 2*col+1, 2*row+1) According to the code (Li, Ci, Ri) of a tile in layer l, l>=2, the code (Li-1, Ci-1, Ri-1) of the tile generated when it is used as one of the four input tiles of the tile resampling operation is calculated, where: Merge the data in the same grid into one grid and perform sampling to complete the merging of tile data.

Images