A rapid extraction method of plastic greenhouses based on Sentinel-2 remote sensing images

Through the APGI index of Sentinel-2 remote sensing images, the problem of insufficient robustness caused by the diversity of greenhouse types and seasonal changes in the existing technology is solved, and the rapid and accurate extraction of large-scale plastic greenhouses is achieved, showing the powerful Lu Strong and applicability.

CN114187519BActive Publication Date: 2024-12-06NANJING UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202111535348.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-15
Publication Date
2024-12-06
Estimated Expiration
2041-12-15

AI Technical Summary

Technical Problem

When facing the diversity of greenhouse types, greenhouse background diversity and seasonal changes, the existing plastic greenhouse index is insufficiently robust, making it difficult to achieve rapid and accurate extraction of large areas and large scales.

Method used

Adaptive greenhouse index (APGI) based on Sentinel-2 remote sensing images is adopted, and the image resolution is resampled through radiation calibration and atmospheric correction, and the APGI index value is calculated, and binary segmentation is performed in combination with preset thresholds to distinguish greenhouse and Not a greenhouse.

Benefits of technology

It has achieved rapid and accurate extraction of greenhouses in different seasons in typical greenhouse planting areas around the world, showing strong robustness and stability, which is better than the accuracy and applicability of existing indexes in large-area applications.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114187519B_ABST
    Figure CN114187519B_ABST
Patent Text Reader

Abstract

The present invention relates to a method for quickly extracting plastic greenhouses based on Sentinel‑2 remote sensing images, and belongs to the field of remote sensing geoscience application technology. The main steps include: 1) atmospheric correction of Sentinel‑2 remote sensing images to obtain the true reflectivity of the surface; 2) using the coastal aerosol band, red band, near infrared (NIR) band, and shortwave infrared 2 (SWIR2) band of the image to calculate the Advanced Plastic Greenhouse Index (APGI); 3) using a preset threshold Threshold to perform binary segmentation on the APGI index image to obtain a plastic greenhouse thematic map. The method provided by the present invention can quickly and effectively extract different types of greenhouses in typical greenhouse planting areas around the world, and has strong robustness to seasonal changes.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a method for quickly extracting plastic greenhouses based on Sentinel-2 remote sensing images, and belongs to the technical field of remote sensing geoscience applications. Background Art

[0002] In the past few decades, the application of plastic greenhouses has promoted the rapid development of traditional agriculture. Plastic greenhouses use plastic film to provide a suitable growth environment for crops, which can greatly increase the yield and quality of crops. As of 2016, the area of ​​plastic greenhouses in the world exceeded 30,000 square kilometers, mainly distributed in East Asia, Europe, North Africa and the Middle East. However, with the large-scale application and rapid expansion of plastic greenhouses, the ecological and environmental problems they bring are becoming increasingly prominent. The excessive and intensive use of plastic greenhouses will inevitably change the local land use and landscape pattern. Specifically, the high reflectivity and airtightness of greenhouses will change the energy balance and water cycle, and even lead to a long-term decrease in surface temperature. Waste plastics and fertilizers and pesticides will also cause problems such as soil pollution, biodiversity decline and food safety. The rapid and large-scale mapping of plastic greenhouses is of great significance to the decision-making and planning of modern agriculture and to reveal the ecological and environmental effects of greenhouses. Therefore, it is of great significance to study the rapid and large-scale extraction methods of plastic greenhouses. Remote sensing images provide the possibility for rapid and large-scale greenhouse mapping.

[0003] The current plastic greenhouse extraction methods are mainly divided into classifier-based and index-based methods. Among them, although the classifier-based method has a high extraction accuracy, it requires a large number of samples to train the classifier, so it is not suitable for large-scale applications. The index-based method has the advantages of fast calculation speed and easy use, and is suitable for large-scale greenhouse extraction. The current greenhouse indexes mainly include PGI [see Yang, D., Chen, J., Zhou, Y., Chen, Xiang, Chen, Xuehong, Cao, X., 2017. Mapping plastic greenhouse with medium spatial resolution satellite data: Development of a new spectral index. ISPRS J. Photogramm. Remote Sens. 128, 47–60.], PMLI [see Lu, L., Di, L., Ye, Y., 2014. A decision-tree classifier for extracting transparent plastic-mulched Landcover from Landsat-5TM images. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 7, 4548–4558.], V I [See Zhao, G.-X., Li, J., Li, T., Yue, Y.-D., T. W., 2004. Utilizing Landsat TM imagery to map greenhouses in Qingzhou, Shandong Province, China. Pedosphere 14, 363–369.] The current greenhouse index faces three main problems, namely, the diversity of greenhouse types, the diversity of greenhouse backgrounds, and the seasonal changes of greenhouses. The above problems restrict the robustness of the current greenhouse index in greenhouse mapping in large areas. Summary of the invention

[0004] The technical problem to be solved by the present invention is to overcome the problems faced by the existing greenhouse index, such as the diversity of greenhouse types, the diversity of greenhouse backgrounds and the seasonal changes of greenhouses, and to propose a greenhouse index with wide adaptability and strong robustness to achieve the purpose of fast and accurate extraction of large-scale plastic greenhouses.

[0005] In order to solve the above technical problems, the present invention provides a

[0006] Method for Rapid Extraction of Plastic Greenhouses Based on Sentinel-2 Remote Sensing Images, the method comprising the following steps:

[0007] Step 1, perform radiometric calibration and atmospheric correction on the Sentinel-2 remote sensing image to obtain the true surface reflectance;

[0008] Step 2, resample the bands with an image resolution of less than 10 meters to 10 meters;

[0009] Step 3, calculate the APGI index value for each pixel to obtain the APGI index image, the APGI index calculation formula is as follows:

[0010]

[0011] where ρ coastal , ρ red , ρ NIR and ρ SWIR2 are the surface reflectances of the 1st, 4th, 8th, and 12th bands of the Sentinel-2 remote sensing image respectively;

[0012] Step 4, perform binary segmentation on the APGI index image based on a preset threshold Threshold, the pixels with APGI >= Threshold are greenhouses; the pixels with APGI < Threshold are non-greenhouses.

[0013] Further, the preset threshold Threshold for the Chinese region is 0.25; the preset threshold Threshold for the Mediterranean region is 0.4.

[0014] The beneficial effects of the present invention are:

[0015] Based on the APGI index of the present invention, the Sentinel-2 remote sensing image can be effectively used to extract plastic greenhouses in a large area, quickly and accurately. Relevant experiments show that APGI can effectively extract greenhouses in different seasons in typical greenhouse planting areas around the world (Almería, Spain; Anamur, Turkey; Weifang, China; Nantong, China). Description of the Drawings

[0016] The following drawings are intended to illustrate and explain the present invention schematically and do not limit the scope of the present invention. Among them:

[0017] Figure 1 is the flow chart of the method for extracting plastic greenhouses based on Sentinel-2 remote sensing images.

[0018] Figure 2 is the spectral curve graph of greenhouses in each research case.

[0019] Figure 3 It is the spectral curve of the greenhouse and other background objects.

[0020] Figure 4 It is a statistical graph of the reflectivity of each type of land in each band or band combination.

[0021] Figure 5 It is a graph showing the calculation results of each greenhouse index (APGI, PGI, PMLI, VI) in each research case.

[0022] Figure 6 APGI and PGI in Almeria in summer ( Figure 6 (a)~(c))、Winter Almeria( Figure 6 Detailed results of (d) to (f).

[0023] Figure 7 Here are the detailed results of APGI and PGI in Anamur.

[0024] Figure 8 These are the detailed results of APGI and PGI in Nantong.

[0025] Fig. 9 These are the detailed results of APGI and PGI in Weifang.

[0026] Fig.10 are the accuracy evaluation results of APGI and PGI corresponding to different thresholds in each study case: (a) Almeria in summer, (b) Almeria in winter, (c) Anamur, (d) Nantong, and (e) Weifang.

[0027] Fig.11 are the greenhouse extraction results of APGI and PGI in each study case: (a1)~(a3) Almeria in summer, (b1)~(b3) Almeria in winter, (c1)~(c3) Anamur, (d1)~(d3) Nantong, (a1)~(a3) Weifang.

[0028] Fig.12 Almeria Fig.12 (a),(b)) and Weifang ( Fig.12 (c), (d)) APGI and PGI statistics of greenhouse and non-greenhouse in each season. DETAILED DESCRIPTION

[0029] The present invention is described in detail below with reference to the accompanying drawings to make the technical route and operation steps of the present invention clearer.

[0030] Those skilled in the art should know that the following implementation modes are not the only limitations on the technical solutions of the present invention, and any equivalent changes or modifications made within the spirit of the technical solutions of the present invention should be deemed to be within the protection scope of the present invention.

[0031] Figure 1 This is a flow chart of a method for extracting plastic greenhouses based on Sentinel-2 remote sensing images. The method of the present invention mainly includes the following steps, which will be introduced in detail below.

[0032] Step 1: Radiometric calibration and atmospheric correction are performed on the Sentinel-2 remote sensing images to obtain the true surface reflectance. The present invention directly uses the surface reflectance products that have been radiometrically calibrated and atmospherically corrected provided by the European Space Agency.

[0033] Step 2: Resample the bands with a resolution of less than 10 meters in the image to 10 meters. The present invention uses the nearest neighbor method of ENVI software to resample the bands with a resolution of less than 10 meters to 10 meters.

[0034] Step 3: Calculate the APGI index value for each pixel to obtain the APGI index image. The APGI index is defined as follows:

[0035]

[0036] where ρ coastal , ρ red , ρ NIR and ρ SWIR2 are the surface reflectances of the 1st, 4th, 8th, and 12th bands of the Sentinel-2 remote sensing image, respectively.

[0037] Step 4: Binary segmentation is performed on the APGI index image based on a preset threshold Threshold. Pixels with APGI >= Threshold are greenhouses; pixels with APGI < Threshold are non-greenhouses.

[0038] The preset threshold Threshold for the Chinese region is 0.25, and the preset threshold Threshold for the Mediterranean region is 0.4.

[0039] Regarding the design process of the APGI index in the present invention, the present invention elaborates as follows:

[0040] 1. Select four typical greenhouse planting areas globally as the study areas, namely Almería, Spain; Anamur, Turkey; Weifang, China; and Nantong, China. Among them, since the spectral characteristics of the greenhouses in Almería are significantly different in summer and winter, summer Almería and winter Almería are taken as two separate study cases for analysis. Therefore, this study includes a total of five study cases.

[0041] 2. For each study case, analyze the spectral characteristics of its greenhouses, such as Figure 2As shown. Obviously, the Almeria greenhouse has significantly different spectral characteristics in summer, while the spectral characteristics of the greenhouses in the other four research cases are relatively similar. Since the Almeria greenhouses are generally sprayed with white paint in summer to prevent the temperature inside the greenhouse from being too high, the transparency of the greenhouse is reduced, most of the sunlight is reflected, and only a small amount of sunlight can penetrate the plastic film of the greenhouse, so the greenhouse in this case has a higher reflectivity. In view of this, the greenhouses are divided into high-transparency greenhouses and low-transparency greenhouses according to their transparency.

[0042] 3. For each research case, eight land types were selected for spectral analysis, including two types of greenhouses (high transparency greenhouses and low transparency greenhouses), four main land surface types (artificial surface, cultivated land, vegetation and water bodies), and two special land types (blue-roofed buildings and mixed pixels of artificial surface and vegetation), such as Figure 3 The reflectivity of each type of land in each band or band combination is statistically analyzed, as shown in Figure 4 As shown. Figure 3 and Figure 4 It can be seen that: 1) The difference between the near infrared and red light bands (NIR-red) can effectively distinguish high-transparency greenhouses from artificial surfaces and water bodies ( Figure 4 (m)); 2) The difference between the near infrared and short wave infrared 2 bands (NIR-SWIR2) can effectively distinguish greenhouses from artificial surfaces and blue roof buildings ( Figure 4 (n)); 3) The red band can effectively distinguish greenhouses from vegetation and cultivated land; 4) The near-infrared band can effectively distinguish greenhouses from mixed pixels of artificial surfaces and vegetation. Based on the above findings, a new greenhouse index (APGI) is proposed, which is defined as follows:

[0043]

[0044] Among them, ρ coastal ,ρ red ,ρ NIR and ρ SWIR2 They are the surface reflectance of bands 1, 4, 8, and 12 of the Sentinel-2 remote sensing image.

[0045] In order to better illustrate the technical effect of the present invention, five research cases are calculated using the APGI index, PGI index, PMLI index, and VI index proposed in the present invention. The images obtained after calculation are as follows: Figure 5 As shown. Figure 5 It can be seen that the PMLI index and the VI index cannot effectively distinguish the greenhouse from the background in almost all cases; while APGI and PGI performed well in the three cases of Almeria in winter, Anamur, and Weifang. In Nantong and Almeria in summer, APGI performed well while PGI could not effectively distinguish the greenhouse from the background. In order to more finely reflect the effect of APGI, Figure 6 to Figure 9 The detailed images of APGI and PGI in each case are presented, and the results show that APGI exhibits stronger greenhouse extraction and background suppression capabilities than PGI in each case.

[0046] The selection of the threshold in step 4 comes from the statistical graph of the accuracy of greenhouse extraction in each research case under each threshold. Fig.10 As shown in the figure, the APGI threshold with the best accuracy in the Mediterranean region (Almeria and Anamur) is around 0.4; the APGI threshold with the best accuracy in China (Weifang and Nantong) is around 0.25. Fig.10 It can be seen that the greenhouse extraction accuracy of APGI in each research case is better than that of PGI. Fig.11 These are the greenhouse extraction results using APGI and PGI at the optimal threshold in each research case, which also shows that APGI has stronger greenhouse extraction ability than PGI. Fig.12 Almeria Fig.12 (a),(b)) and Weifang ( Fig.12 (c), (d)) The APGI and PGI statistics of greenhouses and non-greenhouses in each season show that APGI can effectively extract greenhouses in all seasons, except for the summer in Weifang. This is because the region is in the greenhouse replacement period from early July to late August. During this period, the crops in the greenhouses are harvested and there is only bare soil in the greenhouses, so APGI and PGI cannot effectively detect the greenhouses in the region in summer. In addition, APGI shows robustness to seasonal changes and stronger stability than PGI.

Claims

1. A method for rapid extraction of plastic greenhouses based on Sentinel-2 remote sensing images, the method comprising the following steps: Step 1: Perform radiometric calibration and atmospheric correction on the Sentinel-2 remote sensing image to obtain the true reflectivity of the surface; Step 2: resample the bands with image resolution less than 10 meters to 10 meters; Step 3: Calculate the APGI index value of each pixel to obtain an APGI index image. The APGI index calculation formula is as follows: in, ρ coastal ,ρ red ,ρ NIR and ρ SWIR2 They are the surface reflectance of the 1st, 4th, 8th and 12th bands of the Sentinel-2 remote sensing image; Step 4: Binarize and segment the APGI index image based on a preset threshold Threshold. The pixels with APGI >= Threshold are greenhouses; the pixels with APGI < Threshold are non-greenhouses.

2. The method for rapid extraction of plastic greenhouses based on Sentinel-2 remote sensing images according to claim 1, characterized in that: The preset threshold Threshold for the Chinese region takes a value of 0.

25.

3. The method for rapid extraction of plastic greenhouses based on Sentinel-2 remote sensing images according to claim 1, characterized in that: The preset threshold Threshold for the Mediterranean region takes a value of 0.4.

Citation Information

Patent Citations

  • Plastic greenhouse extraction method and coverage inversion method based on medium-resolution remote sensing image

    CN106934405A

  • A method and device for extracting greenhouse based on remote sensing image

    CN109087316A