Remote Sensing Identification Method of Blue Ginseng Sheds Based on Sentinel-2 Optical Satellite Images

Abstract

The present invention proposes a remote sensing identification method for blue ginseng sheds based on Sentinel-2 optical satellite images, and the steps are as follows: First, preprocess the optical satellite images during the life cycle of the blue ginseng shed to obtain surface reflectance images, and collect the spectra of the blue ginseng shed and other ground objects using a spectral acquisition tool; Second, analyze the spectral characteristics and differences between the blue ginseng shed and other ground objects to determine the effective spectral bands for identifying the blue ginseng shed; Then, based on a function model, expand the spectral characteristic differences between the blue ginseng shed and other ground objects, that is, create a first remote sensing index for the blue ginseng shed and a second remote sensing index for the blue ginseng shed; Finally, based on the hierarchical classification idea, use the first remote sensing index for the blue ginseng shed, the second remote sensing index for the blue ginseng shed, the red light band, and the near-infrared band to construct an automatic identification model for the blue ginseng shed, and achieve precise automatic remote sensing identification of the blue ginseng shed. The present invention can effectively identify the blue ginseng shed and suppress the identification interference caused by blue-roofed buildings.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing target recognition, and particularly to a remote sensing recognition method for blue ginseng sheds based on Sentinel-2 optical satellite images. Background Art

[0002] With the improvement of people's living standards, the market demand for ginseng with high medicinal value and health care effects has increased significantly, and the ginseng industry has played an important role in the continuous growth of the local economy. The supply and demand changes affect the market price. Timely and accurately grasping the ginseng planting area has important guiding significance for the sustainable development of the ginseng planting industry. The method of obtaining the ginseng planting area through manual ground surveys is costly and inefficient. Ginseng is artificially planted in greenhouses covered with shading nets. Affected by the growth habits of ginseng, covering with blue shading nets is a mainstream way of ginseng greenhouse planting. This way of planting ginseng in blue ginseng sheds provides an opportunity for obtaining the ginseng planting area by using remote sensing technology. Its main principle is that blue ginseng sheds will form special blue imaging spectral characteristics on optical images. However, a large number of blue-roofed buildings will cause serious interference to the remote sensing recognition of blue ginseng sheds. Summary of the Invention

[0003] Aiming at the technical problem that a large number of blue-roofed buildings cause serious interference to the remote sensing recognition of blue ginseng sheds and reduce the recognition accuracy, the present invention proposes a remote sensing recognition method for blue ginseng sheds based on Sentinel-2 optical satellite images, which can effectively identify blue ginseng sheds and suppress the recognition interference caused by blue-roofed buildings.

[0004] The technical solution of the present invention is realized as follows:

[0005] A remote sensing recognition method for blue ginseng sheds based on Sentinel-2 optical satellite images, the steps are as follows:

[0006] S1. Collect any phase of Sentinel-2 optical satellite images during the life cycle of blue ginseng sheds, and preprocess the Sentinel-2 optical satellite images to obtain Sentinel-2 surface reflectance images;

[0007] S2. Use a GNSS positioning tool to collect the geographical coordinates of blue ginseng sheds and other ground objects, and then use a spectral collection tool to collect the spectral samples of blue ginseng sheds and other ground objects at each geographical coordinate position in the Sentinel-2 surface reflectance image obtained in step S1;

[0008] S3. Analyze the spectral samples obtained in step S2, and determine the effective spectral bands in which the blue ginseng shed is different from other ground objects, namely the blue band, the red band, the second red edge band, the near-infrared band, the first short-wave infrared band, and the second short-wave infrared band;

[0009] S4. Based on the effective spectral bands obtained in step S3, construct the first remote sensing index of the blue ginseng shed and the second remote sensing index of the blue ginseng shed;

[0010] S5. Based on the Sentinel-2 surface reflectance image obtained in step S1 and the first remote sensing index of the blue ginseng shed obtained in S4, calculate the first remote sensing index image of the blue ginseng shed;

[0011] S6. Based on the Sentinel-2 surface reflectance image obtained in step S1 and the second remote sensing index of the blue ginseng shed obtained in S4, calculate the second remote sensing index image of the blue ginseng shed;

[0012] S7. Identify the blue ginseng shed in the area to be identified in the Sentinel-2 surface reflectance image obtained in step S1. If the pixel value of the pixel position i in the red band in the area to be identified is less than the first threshold γ, and the pixel value in the near-infrared band is greater than the second threshold δ, then execute step S8; otherwise, assign the attribute of this pixel position to other ground objects;

[0013] S8. If the pixel value of the pixel position i in the first remote sensing index image of the blue ginseng shed in the area to be identified is greater than the third threshold α, then execute step S9; otherwise, assign the attribute of this pixel position to other ground objects;

[0014] S9. If the pixel value of the pixel position i in the second remote sensing index image of the blue ginseng shed in the area to be identified is greater than the fourth threshold β, then the attribute of this pixel position is the blue ginseng shed; otherwise, assign the attribute of this pixel position to the blue roof building;

[0015] S10. Loop through steps S7 to S9 until all pixel positions in the area to be identified are traversed, that is, complete the remote sensing automatic identification of the blue ginseng shed.

[0016] The method for determining the first threshold γ is as follows: statistically analyze the blue ginseng shed spectral samples obtained in step S2, and obtain that the pixel value of the blue ginseng shed in the red band is less than γ; the method for determining the second threshold δ is as follows: statistically analyze the blue ginseng shed spectral samples obtained in step S2, and obtain that the pixel value of the blue ginseng shed in the near-infrared band is greater than δ; the method for determining the third threshold α is as follows: based on the blue ginseng shed spectral samples obtained in step S2 and the first remote sensing index of the blue ginseng shed obtained in step S4, calculate the value range of the blue ginseng shed in the first remote sensing index of the blue ginseng shed, that is, the first remote sensing index of the blue ginseng shed is greater than α; the method for determining the fourth threshold β is as follows: based on the blue ginseng shed spectral samples obtained in step S2 and the second remote sensing index of the blue ginseng shed obtained in step S4, calculate the value range of the blue ginseng shed in the second remote sensing index of the blue ginseng shed, that is, the second remote sensing index of the blue ginseng shed is greater than β.

[0017] The expression of the first remote sensing index of the blue ginseng shed is as follows:

[0018]

[0019] In the formula, BGS 1 represents the value of the first remote sensing index of the blue ginseng shed, and ζ blue represents the surface reflectance in the blue band, and ζ red represents the surface reflectance in the red band, and ζ redEdge2 represents the surface reflectance in the second red-edge band.

[0020] The expression of the second remote sensing index of the blue ginseng shed is as follows:

[0021]

[0022] In the formula, BGS 2 represents the value of the second remote sensing index of the blue ginseng shed, and ζ nir represents the surface reflectance in the near-infrared band, and ζ swir1 represents the surface reflectance in the first short-wave infrared band, and ζ swir2 represents the surface reflectance in the second short-wave infrared band.

[0023] Compared with the prior art, the beneficial effects produced by the present invention are as follows:

[0024] (1) The present invention proposes the first remote sensing index and the second remote sensing index of the blue ginseng shed based on Sentinel-2 optical satellite images, enriching the remote sensing theory.

[0025] (2) The automatic recognition method of the blue ginseng shed proposed by the present invention can accurately and quickly identify the blue ginseng shed, providing accurate basic data for relevant departments and industries. Description of the Drawings

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0027] Figure 1 is the flow chart of the present invention;

[0028] Figure 2 is the spectrum of the main land cover types on Sentinel-2 imagery;

[0029] Figure 3 is the blue ginseng shed identification result of the Liuhe County area in Jilin Province of the present invention. Detailed implementation manners

[0030] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0031] As Figure 1 shown, the embodiments of the present invention provide a method for remotely sensing and identifying blue ginseng sheds based on Sentinel-2 optical satellite imagery. A blue ginseng shed refers to a ginseng planting greenhouse covered with a blue shading net. The specific steps are as follows:

[0032] S1. Taking a partial area of Liuhe County in Jilin Province as the test area, collecting Sentinel-2 optical satellite imagery of September 2, 2022, covering the test area during the life cycle of the blue ginseng shed based on the Google Earth Engine cloud platform, and performing preprocessing such as atmospheric correction and geometric correction on the Sentinel-2 optical satellite imagery to obtain a Sentinel-2 surface reflectance image.

[0033] S2. Using a GNSS positioning tool to collect the geographical coordinates of the blue ginseng shed and other ground objects, and then using a spectral collection tool (spectral collection module in ENVI software) to collect spectral samples of the blue ginseng shed and other ground objects at each geographical coordinate position in the Sentinel-2 surface reflectance image obtained in step S1. The results are as Figure 2 shown.

[0034] S3. Analyze the spectral samples obtained in step S2 to determine the effective spectral bands in which the blue ginseng shed is different from other ground objects, namely the blue band, the red band, the second red-edge band, the near-infrared band, the first short-wave infrared band, and the second short-wave infrared band.

[0035] S4. Based on the effective spectral bands obtained in step S3, further analyze the spectral characteristics of the main ground objects (such as Figure 2 ) and find that the spectral reflectance of the blue ginseng shed and the blue-roofed building has an obvious characteristic of decreasing from the blue band to the red band; the spectral reflectance of vegetation has a slight decreasing characteristic from the blue band to the red band, which has a certain similarity with the spectral characteristics of the blue ginseng shed; the spectral reflectance of bare land and other buildings shows an obvious increasing characteristic from the blue band to the red band, which is significantly different from the spectral characteristics of the blue ginseng shed. Introducing the characteristic that the spectral reflectance of vegetation in the second red-edge band is significantly higher than its spectral reflectance in the blue band can further expand the spectral difference between vegetation and the blue ginseng shed. Therefore, the present invention proposes a normalized difference function model of the blue band reflectance and the mean value of the red band reflectance and the second red-edge band reflectance to distinguish the blue ginseng shed from vegetation, bare land, other buildings, and water bodies. In addition, there are some other ground objects whose spectra have a drifting similarity characteristic with the spectrum of the blue ginseng shed, that is, the change trend of the spectral reflectance in each band is the same, but the reflectance value has an overall upward shift, which will cause the "normalized difference function model of the blue band reflectance and the mean value of the red band reflectance and the second red-edge band reflectance" to be unable to distinguish this type of ground object from the blue ginseng shed. Therefore, the present invention further introduces a correction parameter of the model, that is, the difference between the red band reflectance and 0.2. To sum up, construct the first remote sensing index of the blue ginseng shed (First Blue Ginseng Shed, BGS1), and the expression of the first remote sensing index of the blue ginseng shed is:

[0036]

[0037] In the formula, BGS 1 represents the value of the first remote sensing index of the blue ginseng shed, and ζ blue represents the surface reflectance of the blue band, and ζ red represents the surface reflectance of the red band, and ζ redEdge2 represents the surface reflectance of the second red-edge band.

[0038] To further distinguish the blue ginseng shed from the blue-roofed building, analyze Figure 2It is found that the spectral reflectance of the blue ginseng shed shows an obvious decreasing characteristic in the near-infrared band to the short-wave infrared band, while the spectral reflectance of the blue roof building shows an obvious increasing characteristic in the near-infrared band to the short-wave infrared band. Therefore, the present invention proposes a normalized difference function model of the reflectance in the near-infrared band and the mean values of the reflectances in the first short-wave infrared band and the second short-wave infrared band to realize the separation of the spectral characteristics of the blue ginseng shed and the blue roof building, that is, the second remote sensing index of the blue ginseng shed (Second Blue Ginseng Shed, BGS2). The expression of the second remote sensing index of the blue ginseng shed is as follows:

[0039]

[0040] In the formula, BGS 2 represents the value of the second remote sensing index of the blue ginseng shed, and ζ nir represents the surface reflectance in the near-infrared band, and ζ swir1 represents the surface reflectance in the first short-wave infrared band, and ζ swir2 represents the surface reflectance in the second short-wave infrared band.

[0041] S5. Calculate the first remote sensing index image of the blue ginseng shed based on the Sentinel-2 surface reflectance image obtained in step S1 and the expression (1) of the first remote sensing index of the blue ginseng shed obtained in S4. First, construct a blank image I with the same pixel size and the number of pixel rows and columns as those of the Sentinel-2 surface reflectance image obtained in step S1; then, substitute the blue band pixel value, the red band pixel value, and the second red edge band pixel value at the pixel position i in the Sentinel-2 surface reflectance image into the expression (1) of the first remote sensing index of the blue ginseng shed to obtain a value of the first remote sensing index of the blue ginseng shed, and assign this value to the pixel position i in the blank image I. Traverse all pixel positions in turn, and then the first remote sensing index image of the blue ginseng shed can be obtained. The first remote sensing index image of the blue ginseng shed has the same pixel size and the number of pixel rows and columns as the Sentinel-2 surface reflectance image obtained in step S1.

[0042] S6. Calculate the second remote sensing index image of blue ginseng shed based on the Sentinel-2 surface reflectance image obtained in step S1 and the second remote sensing index expression (2) of blue ginseng shed obtained in S4. First, construct a blank image II with the same pixel size and number of pixel rows and columns as those of the Sentinel-2 surface reflectance image obtained in step S1. Then, substitute the pixel values of the blue band, red band, and second red edge band at pixel position i in the Sentinel-2 surface reflectance image into the second remote sensing index expression (2) of blue ginseng shed to obtain a value of the second remote sensing index of blue ginseng shed, and assign this value to pixel position i in the blank image II. Traverse all pixel positions in sequence, and then the second remote sensing index image of blue ginseng shed can be obtained. The second remote sensing index image of blue ginseng shed has the same pixel size and number of pixel rows and columns as the Sentinel-2 surface reflectance image obtained in step S1.

[0043] S7. Conduct remote sensing identification of blue ginseng shed for the area to be identified in the Sentinel-2 surface reflectance image obtained in step S1. If the pixel value of the red band at pixel position i in the area to be identified is less than the first threshold γ, and the pixel value of the near-infrared band is greater than the second threshold δ, then execute step S8; otherwise, assign the attribute of this pixel position to other ground objects. The determination method of the first threshold γ is as follows: Import the reflectance values of the blue ginseng shed spectral samples obtained in step S2 in the red band into the EXCEL software, and count the distribution range of these values to obtain the maximum value γ'. To overcome the limitations of the samples, expand the sample maximum value γ' by 10%, that is, obtain the first threshold γ. The determination method of the second threshold δ is as follows: Import the reflectance values of the blue ginseng shed spectral samples obtained in step S2 in the near-infrared band into the EXCEL software, and count the distribution range of these values to obtain the minimum value δ'. To overcome the limitations of the samples, shrink the sample minimum value δ' by 10%, that is, obtain the first threshold δ. In this embodiment, γ = 0.21 and δ = 0.24.

[0044] S8. If the pixel value on the first remote sensing index image of blue ginseng shed at pixel position i in the area to be identified is greater than the third threshold α, then execute step S9; otherwise, assign the attribute of this pixel position to other ground objects. The determination method of the third threshold α is as follows: Based on the geographical coordinates of the blue ginseng shed obtained in step S2, use the spectral acquisition tool (the spectral acquisition module in ENVI software) to collect samples of blue ginseng shed at each geographical coordinate position in the first remote sensing index image of blue ginseng shed obtained in step S5, import these sample values into the EXCEL software, and count the distribution range of these values to obtain the minimum value α'. To overcome the limitations of the samples, shrink the sample minimum value α' by 10%, that is, obtain the third threshold α. In this embodiment, α = 0.

[0045] S9. If the pixel value of the second remote sensing index image of the blue ginseng shed at the pixel position i in the area to be recognized is greater than the fourth threshold β, then the attribute of this pixel position is the blue ginseng shed; otherwise, the attribute of this pixel position is assigned as the blue roof building. The method for determining the fourth threshold β is as follows: Based on the geographical coordinates of the blue ginseng shed obtained in step S2, use the spectral acquisition tool (the spectral acquisition module in ENVI software) to collect samples of the blue ginseng shed at each geographical coordinate position in the second remote sensing index image of the blue ginseng shed obtained in step S6, import these sample values into the EXCEL software, and statistically analyze the distribution interval of these values to obtain the minimum value β'. To overcome the limitations of the samples, reduce the sample minimum value β' by 10%, that is, obtain the third threshold β. In this embodiment, β = 0.

[0046] S10. Loop and execute steps S7 to S9 until all pixel positions in the area to be recognized are traversed, that is, complete the remote sensing automatic recognition of the blue ginseng shed.

[0047] In this embodiment, the remote sensing recognition result of the blue ginseng shed is as Figure 3 shown. Through Figure 3 it can be seen that the texture information of the blue ginseng shed is complete. After comparing with the actual situation on the spot, the accuracy reaches 94%, indicating the reliability and accuracy of the recognition of the blue ginseng shed by the present invention.

[0048] The above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A remote sensing identification method for blue ginseng sheds based on Sentinel-2 optical satellite images, characterized in that, the steps are as follows: S1. Collect any phase of Sentinel-2 optical satellite images during the life cycle of the blue ginseng shed, and preprocess the Sentinel-2 optical satellite images to obtain Sentinel-2 surface reflectance images; S2. Use a GNSS positioning tool to collect the geographical coordinates of the blue ginseng shed and other ground objects, and then use a spectral collection tool to collect the spectral samples of the blue ginseng shed and other ground objects at each geographical coordinate position in the Sentinel-2 surface reflectance image obtained in step S1; S3. Analyze the spectral samples obtained in step S2 to determine the effective spectral bands different between the blue ginseng shed and other ground objects, namely the blue band, the red band, the second red edge band, the near-infrared band, the first short-wave infrared band, and the second short-wave infrared band; S4. Based on the effective spectral bands obtained in step S3, construct the first remote sensing index of the blue ginseng shed and the second remote sensing index of the blue ginseng shed; The expression of the first remote sensing index of the blue ginseng shed is: where BGS 1 represents the first remote sensing index value of the blue reference canopy, ζ blue represents the surface reflectance of the blue band, ζ red represents the surface reflectance of the red band, ζ redEdge2 represents the surface reflectance of the second red edge band; The expression of the second remote sensing index of the blue ginseng shed is: In the formula, BGS 2 represents the second remote sensing index value of the blue reference shed, ζ nir represents the surface reflectance in the near-infrared band, ζ swir1 represents the surface reflectance in the first short-wave infrared band, ζ swir2 represents the surface reflectance in the second short-wave infrared band; S5. Calculate the first remote sensing index image of the blue ginseng shed based on the Sentinel-2 surface reflectance image obtained in step S1 and the first remote sensing index of the blue ginseng shed obtained in step S4; S6. Calculate the second remote sensing index image of the blue ginseng shed based on the Sentinel-2 surface reflectance image obtained in step S1 and the second remote sensing index of the blue ginseng shed obtained in step S4; S7. Conduct blue ginseng shed identification on the area to be identified in the Sentinel-2 surface reflectance image obtained in step S1. If the pixel value in the red band at pixel position i in the area to be identified is less than the first threshold γ, and the pixel value in the near-infrared band is greater than the second threshold δ, then execute step S8; otherwise, assign the attribute of this pixel position to other ground objects; S8. If the pixel value on the first remote sensing index image of the blue ginseng shed at pixel position i in the area to be identified is greater than the third threshold α, then execute step S9; otherwise, assign the attribute of this pixel position to other ground objects; S9. If the pixel value on the second remote sensing index image of the blue ginseng shed at pixel position i in the area to be identified is greater than the fourth threshold β, then the attribute of this pixel position is the blue ginseng shed; otherwise, assign the attribute of this pixel position to the blue roof building; S10. Loop through steps S7 to S9 until all pixel positions in the area to be identified are traversed, that is, complete the automatic remote sensing identification of the blue ginseng shed.

2. The remote sensing identification method for blue ginseng sheds based on Sentinel-2 optical satellite images according to claim 1, characterized in that, The method for determining the first threshold γ is as follows: statistically analyze the blue ginseng shed spectral samples obtained in step S2 to obtain that the pixel value of the blue ginseng shed in the red band is less than γ; the method for determining the second threshold δ is as follows: statistically analyze the blue ginseng shed spectral samples obtained in step S2 to obtain that the pixel value of the blue ginseng shed in the near-infrared band is greater than δ; the method for determining the third threshold α is as follows: based on the blue ginseng shed spectral samples obtained in step S2 and the first remote sensing index of the blue ginseng shed obtained in step S4, calculate the value range of the blue ginseng shed in the first remote sensing index of the blue ginseng shed, that is, the first remote sensing index of the blue ginseng shed is greater than α; the method for determining the fourth threshold β is as follows: based on the blue ginseng shed spectral samples obtained in step S2 and the second remote sensing index of the blue ginseng shed obtained in step S4, calculate the value range of the blue ginseng shed in the second remote sensing index of the blue ginseng shed, that is, the second remote sensing index of the blue ginseng shed is greater than β.

Images