Small reservoir batch extraction method based on high-precision remote sensing image

Abstract

The invention provides a small reservoir batch extraction method based on a high-precision remote sensing image, which comprises the step of collecting high-resolution land classification data for many years, and further comprises the following steps of: fusing the high-resolution land classification data; removing polygons without reservoir distribution areas from the water polygons by using multi-source data; intercepting a picture from the high-definition remote sensing image by using the water body polygon and marking the picture; and constructing and training a machine learning image classification model, and classifying the pictures. According to the small reservoir batch extraction method based on the high-precision remote sensing image provided by the invention, high-resolution land utilization data for many years are fused, water body data such as lakes and rivers are excluded by utilizing water surface occurrence frequency and environmental data such as land utilization, vegetation and population distribution, and a machine learning image recognition algorithm is utilized to extract the small reservoir batch extraction result. A small reservoir is extracted from a large amount of water body surfaces.

Description

Technical Field

[0001] The invention relates to the technical field of remote sensing target recognition, in particular to a small reservoir batch extraction method based on high-precision remote sensing images. Background Art

[0002] Small reservoirs have a significant impact on surface runoff. First, many model practitioners have proposed that the storage of small reservoirs is one of the factors that affect the accuracy of the model; second, small reservoirs provide a large amount of water supply and irrigation capacity. Third, small reservoirs are more prone to dam failure, which may lead to disasters such as flash floods and mudslides, causing losses. There are a large number of water conservancy projects represented by reservoirs in the world. Among them, the list and distribution of large reservoirs are relatively clear, but the location and number of small reservoirs are less clear. On the other hand, as the operating life of reservoirs increases, many reservoirs will be scrapped in an orderly manner, and the number of reservoirs actually in use is dynamically changing. Therefore, it is of great value to quickly and timely obtain information on small reservoirs in use for flash flood disaster prediction, small and medium-sized river flood forecasting and other work.

[0003] Currently, the most recognized reservoir datasets include HydroLAKES, OSM, GDAT, The Global DamTracker, GOODD, GRanD, GeoDAR, AQUASTAT, and the World Register of Dams (WRD). However, these data products mainly record large reservoirs and some medium-sized reservoirs, and lack the organization and record of small reservoirs.

[0004] The invention patent application with application number CN110826394A discloses a reservoir identification method and device based on a convolutional neural network algorithm, which includes the following steps: step 1, pre-establishing a reservoir classification neural network model combined with a convolutional neural network, and using the prepared reservoir and non-reservoir remote sensing images to train the reservoir classification neural network model; step 2, for the newly input original remote sensing image, using the water body sensitive band, extracting the water body area in the image, and then using morphological operations to lock the block water body area, and cutting the area in the original image; step 3, uniformly reducing the cropped image to the same square size; step 4, uniformly putting the image into the reservoir classification neural network model for discrimination, and identifying the reservoir area. The disadvantage of this method is that only single-period remote sensing images are used for reservoir extraction, and single-period remote sensing images have the situation that the reservoir is dry or the water surface is too small to extract the water body. Especially in the case of long-term drought, a large number of small reservoirs have no water surface or the water surface is too small. It is a very common problem, so many reservoirs will be missed, especially small reservoirs.

[0005] The invention patent application with application number CN109754025A discloses a method for identifying parameters of small reservoirs without data by combining hydrological simulation and continuous remote sensing images, which includes the following steps: Step 1, determine the reservoir area and upstream basin range of small reservoirs without data; Step 2, determine the reservoir discharge parameters and discharge curve; Step 3, verify the reservoir discharge parameters and discharge curve; Step 4, identify the reservoir operation mode. The disadvantage of this method is that the location of the small reservoir must be determined first, and then the reservoir area of ​​the small reservoir must be identified and the reservoir parameters must be extracted. It is impossible to realize the identification and extraction of large-scale small reservoir objects when the location and distribution of the small reservoirs are unknown. Summary of the invention

[0006] In order to solve the above technical problems, the present invention proposes a method for batch extraction of small reservoirs based on high-precision remote sensing images, which integrates many years of high-resolution land use data, uses water surface occurrence frequency and environmental data such as land use, vegetation, and population distribution to exclude water body data such as lakes and rivers, and uses machine learning image recognition algorithms to extract small reservoirs from a large number of water bodies.

[0007] The purpose of the present invention is to provide a method for batch extraction of small reservoirs based on high-precision remote sensing images, which includes collecting high-resolution land classification data for many years and further includes the following steps: Step 1: Fusion of the high-resolution land classification data; Step 2: Use multi-source data to remove polygons without reservoir distribution areas from water body polygons; Step 3: Use water body polygons to capture and annotate images from high-definition remote sensing images; Step 4: Build and train a machine learning image classification model to classify images.

[0008] Preferably, the high-resolution land classification data is a raster layer, the raster attribute values ​​are numbers, and different numbers represent different land use types.

[0009] In any of the above schemes, preferably, step 1 includes the following sub-steps: Step 11: Propose the water body type, form a new annual raster layer, and assign the attribute value of the raster to 1; Step 12: Add the rasters of all years to get a new raster layer:

[0010]

[0011] in, n is the year number, v i For the i The grid value of the year, Vis the grid value after the sum operation, V In 0- n between; Step 13: According to V Values, generate two new raster layers, use the RasterToPolygon tool in ArcGIS software to convert the two raster layers into two vector layers, polygon1 and polygon2; Step 14: Extract two new vector layers based on the areas of the polygons in the two vector layers.

[0012] In any of the above schemes, preferably, step 13 includes when V ≥1, it is the first raster layer; when V When ≥2, it is the second raster layer.

[0013] In any of the above schemes, preferably, step 14 includes extracting polygons whose areas are greater than a first area threshold from polygon1 to generate layer1; and extracting polygons whose areas are greater than a second area threshold from polygon2 to generate layer2.

[0014] In any of the above solutions, preferably, step 14 further comprises using a spatial selection tool of ArcGIS software to select polygons from polygon2 that intersect polygon1 and remove them from polygon2.

[0015] In any of the above schemes, preferably, step 2 comprises the following sub-steps: Step 21: Based on the polygon2, the watershed layer attributes are collected. The area of ​​each watershed is smaller than the first area threshold. Three watershed layers are extracted according to the watershed layer attribute requirements. The three layers are respectively dissolve processed, and the discrete watersheds are removed. Step 22: Use the three watershed layers to overlay polygon2, select the intersecting polygons, and remove them from polygon2.

[0016] In any of the above solutions, preferably, the watershed layer attributes include average slope, population size and snow cover attributes.

[0017] In any of the above solutions, preferably, the watershed layer attribute requirements are: 1) The average slope is less than 2 degrees; 2) The population is 0; 3) Snow cover is greater than 80%.

[0018] In any of the above schemes, it is preferred that the criterion for determining the discrete watershed is that the number of adjacent watershed units is less than 5.

[0019] In any of the above schemes, preferably, step 3 includes the following sub-steps: Step 31: Overlay the online high-definition image under the water body polygon layer and export the high-definition images of the polygons one by one; Step 32: Use a random algorithm to select no less than 1% of the pictures, and use a manual interpretation method to manually classify the selected pictures into two groups: reservoirs and non-reservoirs, to form a picture sample library.

[0020] In any of the above schemes, preferably, step 4 includes the following sub-steps: Step 41: using the picture samples in the picture sample library, obtaining the region of interest of each picture sample through image color analysis; Step 42: Standardize and crop the region of interest to obtain a standardized sample for training and learning; Step 43: Using the sample, based on the CNN convolutional neural network, learn and train the features in the image to generate a machine learning image classification model; Step 44: Import the image into the machine learning image classification model and output the classification result.

[0021] The present invention proposes a method for batch extraction of small reservoirs based on high-precision remote sensing images. Based on multi-period resolution land use data and high-precision remote sensing images, the method comprehensively utilizes the advantages of manual identification and machine learning to achieve rapid identification of small reservoirs over a large area, and avoids the problem of missing some small reservoirs due to drought or water use causing the water surface of the reservoir to be too small.

[0022] The RasterToPolygon tool is a tool that converts raster data to vector data. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 The present invention is a flowchart of a preferred embodiment of a method for batch extraction of small reservoirs based on high-precision remote sensing images.

[0024] Figure 2 The present invention is a flowchart of another preferred embodiment of the method for batch extraction of small reservoirs based on high-precision remote sensing images.

[0025] Figure 3 It is a schematic diagram of an embodiment of a summed raster file according to a method for batch extraction of small reservoirs based on high-precision remote sensing images of the present invention.

[0026] Figure 4It is a schematic diagram of an embodiment of the water body polygons of the method for batch extraction of small reservoirs based on high-precision remote sensing images according to the present invention.

[0027] Figure 5 This is not a schematic diagram of a reservoir class according to an embodiment of the manual annotation classification results of the small reservoir batch extraction method based on high-precision remote sensing images of the present invention.

[0028] Figure 6 The present invention is a schematic diagram of reservoir classes according to an embodiment of the manual annotation classification results of the small reservoir batch extraction method based on high-precision remote sensing images.

[0029] Figure 7 A schematic diagram of reservoir classes according to an embodiment of the classification results of a method for batch extraction of small reservoirs based on high-precision remote sensing images according to the present invention.

[0030] Figure 8 A schematic diagram of river classes according to an embodiment of the classification results of a method for batch extraction of small reservoirs based on high-precision remote sensing images according to the present invention.

[0031] Fig. 9 A schematic diagram of pond or lake classification according to an embodiment of the classification results of the small reservoir batch extraction method based on high-precision remote sensing images according to the present invention. DETAILED DESCRIPTION

[0032] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0033] Example 1 like Figure 1 As shown, a method for batch extraction of small reservoirs based on high-precision remote sensing images executes step 100 to collect high-resolution land classification data for many years. The high-resolution land classification data is a raster layer, and the raster attribute value is a number. Different numbers represent different land use types.

[0034] Executing step 110 to fuse the high-resolution land classification data includes the following sub-steps: Execute step 111 to extract the water body type, form a new yearly grid layer, and assign the attribute value of the grid to 1.

[0035] Execute step 112 to sum the rasters of all years to obtain a new raster layer:

[0036]

[0037] in, n is the year number, v i For the iThe grid value of the year, V is the grid value after the sum operation, V In 0- n between.

[0038] Execute step 113, according to V Values, generate two new raster layers, use the RasterToPolygon tool in ArcGIS software to convert the two raster layers into two vector layers, polygon1 and polygon2. V ≥1, it is the first raster layer; when V When ≥2, it is the second raster layer.

[0039] Execute step 114 to extract two new vector layers based on the areas of the polygons in the two vector layers. Extract polygons with an area greater than a first area threshold from polygon1 to generate layer1; extract polygons with an area greater than a second area threshold from polygon2 to generate layer2; use the spatial selection tool of ArcGIS software to select polygons from polygon2 that intersect with polygon1 and remove them from polygon2.

[0040] Executing step 120, using multi-source data to remove polygons of areas without reservoir distribution from water body polygons, includes the following sub-steps: Execute step 121, based on the polygon2, collect the watershed layer attributes, the area of ​​each watershed is less than the first area threshold, extract three watershed layers respectively according to the watershed layer attribute requirements, and dissolve the three layers respectively, and remove the discrete watersheds; the watershed layer attributes include average slope, population and snow cover attributes; the watershed layer attribute requirements are: 1) The average slope is less than 2 degrees; 2) The population is 0; 3) Snow cover is greater than 80%.

[0041] The criterion for determining the discrete watershed is that there are less than 5 adjacent watershed units.

[0042] Execute step 122, use the three watershed layers to overlay polygon2, select the intersecting polygons, and remove them from polygon2.

[0043] Execute step 130, using water body polygons to capture and annotate images from high-definition remote sensing images, including the following sub-steps: Execute step 131, overlay the online high-definition image under the water body polygon layer, and export the high-definition images of the polygons one by one; Execute step 132, use a random algorithm to select no less than 1% of the pictures, use a manual interpretation method to manually classify the selected pictures into two groups: reservoirs and non-reservoirs, to form a picture sample library.

[0044] Execute step 140 to build and train a machine learning image classification model to classify images, including the following sub-steps: Execute step 141, using the picture samples in the picture sample library, and obtaining the region of interest of each picture sample through image color analysis; Execute step 142 to standardize and crop the region of interest to obtain a standardized sample for training and learning; Execute step 143, using the sample, based on the CNN convolutional neural network, learn and train the features in the image to generate a machine learning image classification model; Execute step 144 to import the image into the machine learning image classification model and output the classification result.

[0045] Example 2 The present invention proposes a method for extracting small reservoirs in a large area based on multi-year high-resolution land use data and high-precision remote sensing images. First, the polygons of possible small reservoirs are screened out by extracting and fusing water body polygons from multi-year high-resolution land use data. Then, the remote sensing images of each polygon extending a certain range are intercepted from the high-resolution remote sensing images. Finally, the image recognition neural network is trained to identify small reservoirs.

[0046] (1) Integrate high-resolution land use data from many years, and use environmental data such as water surface frequency and land use, vegetation, and population distribution to exclude water body data such as lakes and rivers; (2) Use machine learning image recognition algorithms to extract small reservoirs from a large number of water bodies.

[0047] The calculation process of the present invention is as follows Figure 2 shown.

[0048] Step 1: Fusion of multi-year high-resolution land use classification data.

[0049] Collect high-resolution land classification data for many years. This type of data is a raster layer. The raster attribute value is a number. Different numbers represent different land use types. First, the water body type is proposed to form a new year-by-year raster layer, and the attribute value of the raster is uniformly assigned to 1. Then, the raster of all years is summed up to obtain a new raster layer:

[0050]

[0051] In the formula, n is the year number,v i For the i The grid value of the year, here, v i The value of is unified to 1, which is the grid value after the addition operation. V In 0- n between.

[0052] according to V value, generate two new raster layers, if V ≥1, for the first raster layer, if V ≥2, which is the second raster layer. Then use the RasterToPolygon tool in ArcGIS software to convert the two raster layers into two vector layers, polygon1 and polygon2.

[0053] According to the area of ​​the polygons in the two vector layers, two new vector layers are extracted: polygons with an area greater than 1km² are extracted from polygon1 to generate layer1. Here, if the area is greater than 1km 2 , generally medium-sized and large reservoirs, do not need to be extracted, can be obtained from existing data sets, and this area threshold can be adjusted in practice; extract polygons with an area greater than 1ha² from polygon2 to generate layer2. Here, it is considered that water bodies with an area less than 1ha² generally cannot reach the scale of small reservoirs. In practice, this area threshold can be adjusted. Using the spatial selection tool of ArcGIS software, select polygons from polygon2 that intersect with polygon1 and remove them from polygon2. This operation can remove large and medium-sized reservoirs from polygon2. This operation uses the Select By Location command in ArcGIS software, and the overlay option selects intersect the source layer feature.

[0054] Step 2: Use multi-source data to remove polygons without reservoirs from water body polygons.

[0055] Based on polygon2 obtained in step 1, collect watershed layers (including average slope, population, and snow cover attributes). The area of ​​each watershed should be less than 100km². Extract watershed layers according to the average slope less than 2 degrees, population of 0, and snow cover greater than 80%, and dissolve these three layers respectively, and then remove discrete watersheds. The judgment standard is: if there are less than 5 adjacent watershed units, remove them.

[0056] Use the three layers obtained to overlay polygon2, select the intersecting polygons, and remove them from polygon2. This operation uses the Select By Location command in ArcGIS software, and select intersect the source layer feature in the overlay option. Step 3: Use water body polygons to capture and label images from high-definition remote sensing images.

[0057] Using the VBA function of ArcGIS software, we overlaid online high-definition images under the water body polygon layer and exported high-definition pictures of polygons one by one. In order to include surrounding information, the rectangular outer box of each polygon was expanded by 1.5 times for export. Due to the huge number, multiple programs were started in parallel to export in actual operation, thus obtaining a large number of pictures.

[0058] Using a random algorithm, select no less than 1% (if the number is less than 10,000, select 10,000) of the pictures, such as the 1st, 101st, 201st, etc. Then use the manual interpretation method to manually classify the selected pictures into two groups: reservoirs and non-reservoirs.

[0059] Step 4: Build and train a machine learning image classification algorithm to classify all images.

[0060] Using the image samples obtained in step 3, we can obtain the region of interest (i.e., the boundary of the water body) of each image sample through image color analysis, standardize and crop the region, and obtain a standardized sample for training and learning. Using this sample, we can learn and train the features in the image based on the CNN convolutional neural network. After the model is trained, it can effectively identify the type of water body in the remote sensing impact image.

[0061] Example 3 This embodiment discloses an example of obtaining a small reservoir according to the extraction method of the present application.

[0062] In the first step, we collected 10-meter global land use grid data for seven consecutive years from 2017 to 2023. The data set is sliced ​​at 6° longitude and 8° latitude, and includes 712 slices per year. Using ArcGIS software, we first extracted the water body grid, then set the grid value to 1, and then added the grids for seven years to obtain 712 added grid files, as shown in the figure. Figure 3 shown.

[0063] Extract all rasters with raster values ​​greater than or equal to 1 and raster values ​​greater than or equal to 2 from the raster file and convert them into vector layers. Extract polygons with an area greater than 1km² and an area greater than 1ha² from the vector layers to obtain layer 1 and layer 2 respectively. Then remove polygons from layer 2 that intersect with polygons in layer 1. Finally, 7.87 million polygons are obtained. The results are as follows: Figure 4 As shown, the range is 110°-111° east longitude and 40°-41° north latitude.

[0064] The second step is to collect the global BasinATLAS_v10 basin data, which contains 1.18 million basins. The attributes of each basin include the average slope of the basin, the population in the basin, and the ice and snow coverage of the basin. Three layers are extracted according to the conditions of average slope greater than 2 degrees, population of 0, and ice and snow coverage greater than 80%, and the discrete basins in the three layers are removed respectively (there are less than 5 adjacent basins), and then the three layers are superimposed with the water body polygon layers obtained in the previous step, and the water body polygons that intersect with the basin faces in the three layers are removed from the water body polygons, and 1.51 million polygons are obtained.

[0065] Step 3: Use water polygons to capture and annotate images from high-definition remote sensing images. Remote sensing images use ESRI's free World Imagery data, which includes three resolutions: 30cm, 60cm, and 15m. The company processes the data into image data with 14 scales ranging from 1.9cm to 150m. A batch processing program is used to generate 1.51 million images, and 1% of them are selected for manual annotation and classification, such as Figure 5 The images shown are confirmed not to be reservoirs, such as Figure 6 Shown is an image confirmed to be a reservoir.

[0066] Step 4: Use the image samples obtained in step 3 to build and train a machine learning image classification algorithm to classify all images.

[0067] Based on the image samples in step 3, the water bodies in the images are preliminarily identified through image color analysis to obtain the region of interest of the image samples, and the region is standardized and cropped to obtain standardized training learning samples.

[0068] The standardized sample is used as a learning sample, and the CNN convolutional neural network model is used for feature-based recognition. According to the needs of daily work and the characteristics of the sample images, the training samples are classified into Pool (pond or lake), River (river) and Reservoir (reservoir). The results are as follows: Figure 7 The reservoir is shown; Figure 8 The river is shown; Fig. 9 A pond or lake is shown.

[0069] In order to better understand the present invention, the above is described in detail in conjunction with the specific embodiments of the present invention, but it is not intended to limit the present invention. Any simple modifications made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention. Each embodiment in this specification focuses on the differences from other embodiments, and the same or similar parts between the embodiments can be referenced to each other. For the system embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

Claims

1. A method for batch extraction of small reservoirs based on high-precision remote sensing images, including collecting high-resolution land classification data for many years, and also including the following steps: Step 1: Fusion of the high-resolution land classification data, including the following sub-steps: Step 11: Propose the water body type, form a new annual raster layer, and assign the attribute value of the raster to 1; Step 12: Add the rasters of all years to get a new raster layer: ， in, n is the year number, v i For the i The grid value of the year, V is the grid value after the sum operation, V In 0- n between; Step 13: According to V Values, generate two new raster layers, use the RasterToPolygon tool in ArcGIS software to convert the two raster layers into two vector layers, polygon1 and polygon2; Step 14: Extract two new vector layers according to the areas of the polygons in the two vector layers; Step 2: Use multi-source data to remove polygons without reservoir distribution areas from water body polygons; Step 3: Use water body polygons to capture and annotate images from high-definition remote sensing images; Step 4: Build and train a machine learning image classification model to classify images.

2. The method for batch extraction of small reservoirs based on high-precision remote sensing images according to claim 1 is characterized in that: The step 13 includes when V ≥1, it is the first raster layer; when V When ≥2, it is the second raster layer.

3. The method for batch extraction of small reservoirs based on high-precision remote sensing images as claimed in claim 2 is characterized in that: The step 14 includes extracting polygons whose areas are larger than a first area threshold from polygon1 to generate layer1; and extracting polygons whose areas are larger than a second area threshold from polygon2 to generate layer2.

4. The method for batch extraction of small reservoirs based on high-precision remote sensing images as claimed in claim 3 is characterized in that: The step 14 also includes using the spatial selection tool of ArcGIS software to select polygons from polygon2 that intersect with polygon1 and remove them from polygon2.

5. The method for batch extraction of small reservoirs based on high-precision remote sensing images as claimed in claim 4, characterized in that: The step 2 includes the following sub-steps: Step 21: Based on the polygon2, the watershed layer attributes are collected. The area of ​​each watershed is smaller than the first area threshold. Three watershed layers are extracted according to the watershed layer attribute requirements. The three layers are respectively dissolve processed, and the discrete watersheds are removed. Step 22: Use the three watershed layers to overlay polygon2, select the intersecting polygons, and remove them from polygon2.

6. The method for batch extraction of small reservoirs based on high-precision remote sensing images as claimed in claim 5, characterized in that: The watershed layer attributes include average slope, population, and snow cover attributes.

7. The method for batch extraction of small reservoirs based on high-precision remote sensing images according to claim 6 is characterized in that: The watershed layer attribute requirements are: 1) The average slope is less than 2 degrees; 2) The population is 0; 3) Snow cover is greater than 80%.

8. The method for batch extraction of small reservoirs based on high-precision remote sensing images according to claim 7 is characterized in that: The criterion for determining the discrete watershed is that there are less than 5 adjacent watershed units.

9. The method for batch extraction of small reservoirs based on high-precision remote sensing images according to claim 8, characterized in that: The step 3 includes the following sub-steps: Step 31: Overlay the online high-definition image under the water body polygon layer and export the high-definition images of the polygons one by one; Step 32: Use a random algorithm to select no less than 1% of the pictures, and use a manual interpretation method to manually classify the selected pictures into two groups: reservoirs and non-reservoirs, to form a picture sample library.

10. The method for batch extraction of small reservoirs based on high-precision remote sensing images according to claim 9, characterized in that: The step 4 includes the following sub-steps: Step 41: using the picture samples in the picture sample library, obtaining the region of interest of each picture sample through image color analysis; Step 42: Standardize and crop the region of interest to obtain a standardized sample for training and learning; Step 43: Using the sample, based on the CNN convolutional neural network, learn and train the features in the image to generate a machine learning image classification model; Step 44: Import the image into the machine learning image classification model and output the classification result.

Images