Automatic identification method and automatic identification system for floating-leaf vegetation and emergent vegetation

Abstract

The invention provides an automatic identification method and an automatic identification system for floating-leaf vegetation and emergent vegetation, and relates to the technical field of ecological environment monitoring. The identification method provided by the invention comprises the following steps: acquiring a target radar image of a target area, delimiting a gray threshold based on a gray value of the target radar image, and re-classifying the target radar image based on the gray threshold to obtain an emergent vegetation area and a to-be-distinguished area; obtaining multispectral remote sensing data of the target area, calculating a normalized vegetation index of each pixel in the target area image based on the multispectral remote sensing data, and dividing a water body area and a vegetation mixing area based on the normalized vegetation indexes; and acquiring an overlapping area of the to-be-distinguished area and the vegetation mixing area as a floating-leaf vegetation area. According to the method, the radar image technology and the multispectral remote sensing technology are combined, the floating-leaf vegetation and the emergent vegetation can be effectively identified, technical support is provided for large-scale aquatic vegetation investigation and water ecology detection, and the method has high accuracy and efficient identification capability.

Description

Technical Field

[0001] The present invention relates to the technical field of ecological environment monitoring, and in particular to an automatic identification method and system for floating leaf vegetation and emergent vegetation. Background Art

[0002] Aquatic vegetation plays an indispensable role in maintaining water quality, improving regional water environment, protecting the health of aquatic ecosystems, and protecting species diversity. In the study and management of ecosystems, floating leaf vegetation and emergent vegetation play different roles in aquatic ecosystems, and have their own unique contributions to water purification, ecological balance maintenance, and biodiversity protection. Therefore, accurate identification of floating leaf vegetation and emergent vegetation is of great significance for a deep understanding of the structure and function of aquatic ecosystems, formulating scientific and reasonable protection and management measures, and ensuring the sustainable use of water resources and maintaining ecological balance.

[0003] Traditional aquatic vegetation monitoring is mainly based on manual field surveys. Manual surveys are highly accurate and can identify specific species, but they are time-consuming and labor-intensive, and the monitoring range is small. Satellite remote sensing technology provides a fast, large-scale aquatic vegetation monitoring method with the advantages of real-time, spatiotemporal continuity, and can observe the area of ​​aquatic vegetation in a specific area and its spatiotemporal change trends for a long time. Remote sensing technology can effectively monitor aquatic vegetation and provide important data support for the protection and research of aquatic ecosystems such as lakes and wetlands.

[0004] Multispectral remote sensing technology identifies aquatic vegetation groups through the differences in spectral characteristics of different aquatic vegetation on satellite images. Therefore, a series of vegetation indices for extracting different groups have been constructed, such as the Normalized Difference Vegetation Index (NDVI), the Floating Leaf Vegetation Sensitive Spectral Index (FVSI), and the Floating Algae Index (FAI).

[0005] Based on the relevant vegetation index, a decision tree classification method or threshold method can be used to extract aquatic vegetation. However, since the spectral reflectance of emergent vegetation and floating leaf vegetation is relatively small in the near-infrared band, there is a problem of "different objects with the same spectrum", and it is impossible to effectively distinguish emergent vegetation and floating leaf vegetation based on spectral characteristics using vegetation index. Therefore, it is urgent to provide a solution to improve the above problem. Summary of the invention

[0006] The purpose of the present invention is to provide an automatic identification method and system for floating leaf vegetation and emergent vegetation.

[0007] In a first aspect, the present invention provides an automatic identification method for floating leaf vegetation and emergent vegetation, comprising the following steps: obtaining a target radar image of a target area, defining a grayscale threshold based on the grayscale value of the target radar image, and reclassifying the target radar image based on the grayscale threshold to obtain an emergent vegetation area and an area to be distinguished; obtaining multispectral remote sensing data of the target area, calculating a normalized vegetation index for each pixel in the target area image based on the multispectral remote sensing data, and dividing a water body area and a vegetation mixed area based on the normalized vegetation index; obtaining an overlapping area between the area to be distinguished and the vegetation mixed area as a floating leaf vegetation area.

[0008] The identification method provided by the present invention, by combining radar imaging technology and multispectral remote sensing technology, can effectively identify floating leaf vegetation and emergent vegetation, provide technical support for large-scale aquatic vegetation surveys and water ecological detection, and has high accuracy and efficient identification capabilities.

[0009] Optionally, when acquiring a target radar image of a target area, when defining a grayscale threshold based on the grayscale value of the target radar image, the grayscale value of each pixel in the target radar image is acquired, and a grayscale histogram is constructed based on the grayscale value, and the valley point of two cross peaks in the grayscale histogram is selected as the grayscale threshold.

[0010] Optionally, the grayscale value of each pixel in the emergent vegetation area is greater than or equal to the grayscale threshold, and the grayscale value of each pixel in the area to be distinguished is less than the grayscale threshold.

[0011] Optionally, after acquiring the target radar image, the target radar image is preprocessed, and a grayscale threshold is divided based on the preprocessed target radar image.

[0012] Optionally, the preprocessing includes radiation correction, speckle noise suppression, image registration, geometric correction and linear scale conversion.

[0013] Optionally, when calculating the normalized vegetation index of each pixel in the target area based on the multispectral remote sensing data, it is as shown in the following formula:

[0014]

[0015] Among them: NDVI i is the normalized vegetation index of the ith pixel; NIR i is the near infrared band reflectivity of the i-th pixel; RED i is the infrared band reflectivity of the i-th pixel.

[0016] In a second aspect, the present invention further provides an automatic identification system for floating leaf vegetation and emergent vegetation, comprising:

[0017] Data acquisition module, which acquires target radar images and multispectral remote sensing data of the target area;

[0018] A grayscale processing module, which defines a grayscale threshold based on the grayscale value of the target radar image;

[0019] An emergent vegetation recognition module reclassifies the target radar image based on a grayscale threshold to obtain an emergent vegetation area and an area to be distinguished;

[0020] A normalized vegetation index calculation module calculates the normalized vegetation index of each pixel in the target area image based on the multispectral remote sensing data;

[0021] The floating leaf vegetation recognition module divides the water body area and the vegetation mixed area based on the normalized vegetation index, and obtains the overlapping area between the area to be distinguished and the vegetation mixed area as the floating leaf vegetation area. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A flowchart of a method for automatically identifying floating leaf vegetation and emergent vegetation provided by the present invention;

[0023] Figure 2 A process block diagram of a method for automatically identifying floating leaf vegetation and emergent vegetation provided by the present invention;

[0024] Figure 3 A diagram showing the identification process when the automatic identification method provided by the present invention is applied to a part of the Poyang Lake area to identify floating leaf vegetation and emergent vegetation;

[0025] Figure 4 A schematic diagram of a grayscale histogram and grayscale threshold when the automatic identification method provided by the present invention is applied to a part of the Poyang Lake area to identify floating leaf vegetation and emergent vegetation;

[0026] Figure 5 The present invention provides a structural schematic diagram of a floating leaf vegetation and emergent vegetation automatic identification system. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Unless otherwise defined, the technical terms or scientific terms used herein should be the common meanings understood by people with general skills in the field to which the present invention belongs.

[0028] See also Figure 1 and Figure 2The present invention provides a method for automatically identifying floating leaf vegetation and emergent vegetation, comprising the following steps:

[0029] S1. Obtain a target radar image of a target area, and define a grayscale threshold based on the grayscale value of the target radar image;

[0030] S2, reclassifying the target radar image based on the grayscale threshold to obtain the emergent vegetation area and the area to be distinguished;

[0031] S3, obtaining multispectral remote sensing data of the target area, and calculating the normalized vegetation index of each pixel in the image of the target area based on the multispectral remote sensing data;

[0032] S4. Divide the water area and the mixed vegetation area based on the normalized difference vegetation index;

[0033] S5. Obtain an overlapping area between the area to be distinguished and the vegetation mixed area as a floating leaf vegetation area.

[0034] In fact, the target radar image obtained in step S1 records the backscattering information of the target object in the target area to the radar beam. Since these scattering information are affected by the roughness of the target object, they can show different features in the radar image. Therefore, targets with different roughness can be identified from the target radar image.

[0035] In fact, since both emergent vegetation and floating leaf vegetation grow in water bodies, the upper parts of the stems and leaves of emergent vegetation are above the water surface, while floating leaf vegetation floats on the surface of the water body. The roughness of the floating leaf vegetation is similar to that of the water body surface. In addition, the influence of environmental factors will also cause the roughness of the water body surface to change, such as wind, rain, ripples, etc.

[0036] In some embodiments, when the grayscale threshold is defined based on the grayscale value of the target radar image in step S1, the grayscale value of each pixel in the target radar image is obtained, and a grayscale histogram is constructed based on the grayscale value, and the valley point of two cross peaks in the grayscale histogram is selected as the grayscale threshold.

[0037] In fact, the grayscale histogram constructed when defining the grayscale threshold shows the frequency of occurrence of different grayscale values ​​in the target radar image. Therefore, the concentration trend and dispersion of the grayscale values ​​in the target radar image can be quickly understood from the grayscale histogram, which is helpful for judging the category of the target object in the target radar image.

[0038] Specifically, when defining the grayscale threshold, the different grayscale peak positions in the grayscale histogram reflect the main grayscale value distribution, which corresponds to the regions of different targets in the target radar image, and the two grayscale peaks can be accurately divided by the valley point between the two cross peaks. In fact, in step S2, the grayscale value of each pixel in the emergent vegetation area is greater than or equal to the grayscale threshold, and the grayscale value of each pixel in the area to be distinguished is less than the grayscale threshold.

[0039] In some embodiments, after the target radar image is acquired in step S1, the target radar image is preprocessed, and grayscale thresholds are divided based on the preprocessed target radar image. Specifically, the preprocessing includes radiation correction, speckle noise suppression, image registration, geometric correction, and linear scale conversion.

[0040] In fact, by preprocessing the target radar image, the data quality of the radar image can be significantly improved, which is helpful for comparing and analyzing image data obtained by the same sensor in multiple phases or different sensors, and can highlight the differences in radar intensity, which helps to distinguish ground objects with large brightness differences on radar images.

[0041] In some embodiments, when calculating the normalized vegetation index of each pixel in the target area based on the multispectral remote sensing data in step S3, it is as shown in the following formula:

[0042]

[0043] Among them: NDVI i is the normalized vegetation index of the ith pixel; NIR i is the near infrared band reflectivity of the i-th pixel; RED i is the infrared band reflectivity of the i-th pixel.

[0044] In fact, by calculating the normalized vegetation index of the target area, the growth status and coverage of vegetation in the target area can be effectively reflected, and the vegetation coverage area and water area in the target area can be effectively distinguished through the numerical value of the normalized vegetation index.

[0045] In fact, when the water body area and the vegetation mixed area are divided based on the normalized vegetation index in step S4, since the normalized vegetation index of the water body and the vegetation area is quite different, the classification threshold can be preset or the classification threshold can be adaptively adjusted through the model. For example, the classification threshold can be set with reference to the existing research results and experience range in this field, or a supervised machine learning model can be used to use sample data of known target categories as a training set to adaptively adjust the classification threshold.

[0046] In some embodiments, see Figure 3When using the identification method provided by the present invention to identify floating leaf vegetation and emergent vegetation in a part of Poyang Lake, the following steps are included:

[0047] S1. Select a clear Landsat-8 image (20141008) with few clouds and the corresponding target radar image (20141008). After performing radiation correction, speckle noise suppression, image registration, geometric correction and linear scale conversion on the target radar image, and converting the backscatter coefficient of the target radar image into a logarithmic scale, obtain the gray value of each pixel in the image, and construct the grayscale histogram of the target radar image based on the grayscale value, and select the valley point of the two cross peaks in the grayscale as the grayscale threshold (-21), such as Figure 4 As shown;

[0048] S2. Using the reclassification tool in ArcGIS software, the target radar image is reclassified based on the grayscale threshold, so as to obtain the emergent vegetation area with grayscale values ​​greater than or equal to the grayscale threshold, and the area to be distinguished with grayscale values ​​less than the grayscale threshold;

[0049] S3, obtain multispectral remote sensing data of Landsat-8 image (20141008), use the raster calculator tool in ArcGIS software to build a normalized vegetation index model, and use the normalized vegetation index model to calculate the normalized vegetation index of each pixel in the image;

[0050] S4, setting the classification threshold to 0, and classifying the area with the normalized vegetation index greater than or equal to 0 as a mixed vegetation area, and classifying the area with the normalized vegetation index less than 0 as a water area through the classification threshold;

[0051] S5. Using the intersection tool in ArcGIS software, the area to be resolved and the mixed vegetation area are intersected to obtain the overlapping area as the floating leaf vegetation area.

[0052] The accuracy of the confusion matrix was verified for the above classification results. 600 sample points were randomly generated in the Poyang Lake area using ArcGIS software. These 600 sample points were visually identified and compared with the above classification results to construct a verification sample confusion matrix as shown in Table 1. It can be seen that the overall accuracy of the recognition method provided by the present invention during recognition is 93.33%, and the Kappa coefficient is 0.8798, which indicates that the recognition method provided by the present invention can classify floating leaf vegetation and emergent vegetation more accurately.

[0053] Table 1 Evaluation table of extraction accuracy of floating leaf vegetation and emergent vegetation

[0054]

[0055] See also Figure 5The present invention also provides an automatic identification system for floating leaf vegetation and emergent vegetation, comprising:

[0056] The data acquisition module 100 acquires target radar images and multi-spectral remote sensing data of the target area;

[0057] A grayscale processing module 200 is used to define a grayscale threshold based on the grayscale value of the target radar image;

[0058] The emergent vegetation recognition module 300 reclassifies the target radar image based on the grayscale threshold to obtain the emergent vegetation area and the area to be distinguished;

[0059] A normalized vegetation index calculation module 400 calculates the normalized vegetation index of each pixel in the target area image based on multispectral remote sensing data;

[0060] The floating leaf vegetation identification module 500 divides the water body area and the vegetation mixed area based on the normalized vegetation index, and obtains the overlapping area between the area to be distinguished and the vegetation mixed area as the floating leaf vegetation area.

[0061] Although the embodiments of the present invention are described in detail above, it is obvious to those skilled in the art that various modifications and variations can be made to these embodiments. However, it should be understood that such modifications and variations are within the scope and spirit of the present invention as described in the claims. Moreover, the present invention described herein may have other embodiments and may be implemented or realized in a variety of ways.

Claims

1. A method for automatically identifying floating leaf vegetation and emergent vegetation, characterized in that: The following steps are involved: A target radar image of a target area is obtained, a grayscale threshold is defined based on the grayscale value of the target radar image, and the target radar image is reclassified based on the grayscale threshold to obtain an emergent vegetation area and an area to be distinguished; multispectral remote sensing data of the target area is obtained, a normalized vegetation index of each pixel in the target area image is calculated based on the multispectral remote sensing data, and a water area and a mixed vegetation area are divided based on the normalized vegetation index; an overlapping area between the area to be distinguished and the mixed vegetation area is obtained as a floating leaf vegetation area.

2. The automatic identification method according to claim 1, characterized in that: When defining the grayscale threshold based on the grayscale value of the target radar image, the grayscale value of each pixel in the target radar image is obtained, and a grayscale histogram is constructed based on the grayscale value, and the valley point of two cross peaks in the grayscale histogram is selected as the grayscale threshold.

3. The automatic identification method according to claim 2, characterized in that: The grayscale value of each pixel in the emergent vegetation area is greater than or equal to the grayscale threshold, and the grayscale value of each pixel in the area to be distinguished is less than the grayscale threshold.

4. The automatic identification method according to claim 2, characterized in that: After acquiring the target radar image, the target radar image is preprocessed, and grayscale thresholds are divided based on the preprocessed target radar image.

5. The automatic identification method according to claim 4, characterized in that: The preprocessing includes radiation correction, speckle noise suppression, image registration, geometric correction and linear scale conversion.

6. The automatic identification method according to claim 1, characterized in that: When calculating the normalized vegetation index of each pixel in the target area based on the multispectral remote sensing data, it is as shown in the following formula: Among them: NDVI i is the normalized vegetation index of the ith pixel; NIR i is the near infrared band reflectivity of the i-th pixel; RED i is the infrared band reflectivity of the i-th pixel.

7. An automatic identification system for floating leaf vegetation and emergent vegetation, characterized in that: include: Data acquisition module, which acquires target radar images and multispectral remote sensing data of the target area; A grayscale processing module, which defines a grayscale threshold based on the grayscale value of the target radar image; An emergent vegetation recognition module reclassifies the target radar image based on a grayscale threshold to obtain an emergent vegetation area and an area to be distinguished; A normalized vegetation index calculation module calculates the normalized vegetation index of each pixel in the target area image based on the multispectral remote sensing data; The floating leaf vegetation recognition module divides the water body area and the vegetation mixed area based on the normalized vegetation index, and obtains the overlapping area between the area to be distinguished and the vegetation mixed area as the floating leaf vegetation area.

Images