A method for monitoring green tide algae area based on multi-spectrometer onboard drone

Abstract

The invention belongs to the technical field of green tide monitoring, and provides a method for monitoring green tide area based on a multi-spectrometer mounted on an unmanned aerial vehicle, which includes data collection and data processing. The data collection utilizes a multi-spectrometer mounted on an unmanned aerial vehicle to carry out aerial photography data collection in a green tide outbreak area. The data processing is to process the collected multi-spectral data, including radiation correction processing, multi-spectral image stitching, green tide algae identification and segmentation, and area calculation. The invention collects data through an airborne multi-spectrometer, and then processes the collected multi-spectral data, identifies and segments the green tide algae in the multi-spectral image, counts the number of pixels of the green tide algae, and then calculates the actual area represented by a single pixel according to the ground sampling distance, thereby obtaining the real area of ​​the green tide. The method for monitoring green tide area based on the multi-spectrometer mounted on an unmanned aerial vehicle provided by the invention can realize real-time monitoring of the green tide area, and provide data support for the prevention, control and management of the green tide.

Description

Technical Field

[0001] The invention relates to a method for monitoring the area of ​​green tide algae based on a multi-spectrometer carried by an unmanned aerial vehicle, and belongs to the technical field of green tide monitoring. Background Art

[0002] In recent years, due to atmospheric changes and human activities, green tides have erupted. Green tide is a harmful algal bloom phenomenon formed by the rapid growth and aggregation of large marine green algae, which generally occurs in coastal areas. Existing studies have shown that the algae that form green tides are mainly opportunistic seaweeds with strong nutrient absorption and environmental stress adaptation capabilities, such as Enteromorpha, Ulva, Cladophora, and Cladophora. The outbreak of green tides will have a negative impact on the environment and economy: first, after the outbreak of green tides, the biomass is extremely large, covering a large area of ​​the sea surface, blocking sunlight, absorbing nutrients in seawater, and affecting the growth of other animals and plants in the ocean; second, large-scale accumulation of green tides will rot and stink, affecting the tourism industry and people's health in coastal cities; third, the cleaning of green tides requires a lot of manpower and material resources. It is precisely because of the negative impact of green tide outbreaks that green tide algae must be controlled.

[0003] Green tides are widely distributed on the sea surface, sometimes reaching tens of thousands of square kilometers. Therefore, it is difficult to conduct comprehensive and real-time monitoring of large-scale green tides at sea by relying on traditional voyage surveys. Satellite remote sensing images are highly effective, economical and capable of large-scale synchronous observations. They are an important means of dynamic monitoring of green tide disasters and are widely used in green tide monitoring in the Yellow Sea. However, due to the relatively low resolution of satellite remote sensing, when the green tide outbreak is first observed using satellites, the area covered by floating algae has often reached tens of square kilometers, resulting in the prevention and control of green tide disasters still being in a passive response stage.

[0004] Therefore, there is currently a lack of scientific and systematic means to conduct comprehensive, real-time and rapid monitoring of the occurrence and development of green tides. Summary of the invention

[0005] To solve the above problems, the present invention provides a method for monitoring the area of ​​green tide algae based on a multi-spectrometer onboard an unmanned aerial vehicle, which can monitor the green tide area in real time and quickly, and provide data support for the prevention, control and management of the green tide.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0007] The following steps are involved:

[0008] Step 1: Green tide algae multispectral data collection: Use a drone equipped with a multispectral instrument to collect data in the green tide outbreak area;

[0009] Step 2, multispectral data preprocessing: performing radiation correction preprocessing on the collected multispectral data;

[0010] Step 3, image stitching: performing multispectral image stitching processing on the multispectral data after radiation correction;

[0011] Step 4, green tide algae identification: using the vegetation index method to identify green tide algae in multispectral images;

[0012] Step 5, background segmentation: using the threshold segmentation method, the multispectral image after vegetation index recognition is segmented into two parts: green tide algae and other background;

[0013] Step 6, calculation of green tide area: using the ground sampling distance of the drone, the actual area represented by a single pixel is obtained, the corresponding relationship formula between the drone flight altitude and the aerial photography area is derived, and the number of pixels of green tide algae is counted to obtain the actual area of ​​the green tide.

[0014] Step 7, calculation of green tide area monitoring error: a “preset standard area method” is proposed to set up multiple green tide algae samples of fixed areas in the aerial photography area in advance, and compare the actual area of ​​the green tide algae samples with the monitoring area of ​​the algorithm to verify the error of green tide area monitoring.

[0015] A further improvement of the technical solution of the present invention is that in the step 1, a DJI Phantom 4 multispectral version integrated system is used to plan the aerial photography area, a multispectral instrument is carried out on a drone, the shooting mode is set to take pictures at equal time intervals, the flight altitude is set to 50 meters, the aerial photography pitch angle is set to vertically downward, the heading and lateral overlap rate are set, and after setting the flight parameters, take off to collect multispectral data; wherein each set of multispectral data collected contains a visible light image and five single-band images, and the bands are: blue B: 450nm±16nm, green G: 560nm±16nm, red R: 650nm±16nm, red edge RE: 730nm±16nm, near infrared NIR: 840nm±26nm, and the effective pixels of a single sensor are 2.08 million.

[0016] A further improvement of the technical solution of the present invention is that the method of rapid radiation correction in step 2 is an "alternative sensor correction method", and the specific implementation method is as follows: each single-band data is imported in sequence according to the wavelength, and then the five single-band data are band-synthesized to synthesize a multispectral data containing five bands, and a rapid atmospheric correction algorithm is used to select an alternative sensor containing these five bands for radiation correction. The corrected green tide algae spectral curve conforms to the spectral characteristics of the vegetation spectrum, indicating that the correction result is correct.

[0017] A further improvement of the technical solution of the present invention is that in step 3, the radiation corrected images are stitched using Pix4Dampper software: first, all images are arranged according to the aerial photography positions based on the coordinate positions, and then feature points in adjacent images are extracted and stitched using a feature point matching method, so that multiple single multispectral images are stitched into a large image covering the entire aerial photography area.

[0018] The further improvement of the technical solution of the present invention is that: in the step 4, the vegetation index is a parameter formed by combining the spectral reflectance of different bands. The reflectance of different objects in each band is different. Green tide algae has a strong absorption capacity in the red light band and a strong reflection capacity in the near-infrared band. The "red edge effect" is based on this principle to identify the green tide in the multispectral image; the vegetation index is the normalized green band difference vegetation index GNDVI, and its formula is as follows:

[0019] GNDVI=(R nir -R green ) / (R nir +R green ) (1),

[0020] Where R represents the reflectivity of each band. Based on the MATLAB environment, a green tide algae recognition algorithm based on the GNDVI vegetation index is compiled to identify green tide algae in multispectral images.

[0021] A further improvement of the technical solution of the present invention is that in step 5, the multispectral image processed by the vegetation index is segmented using the maximum inter-class variance threshold segmentation method. The maximum inter-class variance method uses the idea of ​​clustering to divide the grayscale value of the multispectral image into two parts, so that the grayscale value difference between the two parts is maximized, and the grayscale value difference within each part is minimized. A suitable grayscale threshold is selected by calculating the variance for segmentation, and the multispectral image is segmented into two parts of the target green tide algae and other background objects by threshold segmentation.

[0022] A further improvement of the technical solution of the present invention is that in step 6, according to the ground sampling distance GSD of the multispectral instrument of the drone, GSD refers to the actual size represented by the unit pixel in the multispectral image. GSD is related to the flight altitude of the aircraft, the lens focal length of the imaging system, and the size of the pixel. The GSD formula of the DJI Phantom 4 multispectral version is as follows:

[0023] GSD = H / 18.9 (cm / pixel) (2),

[0024] Among them, H is the flight height of the drone. The corresponding formula of the drone flight height and the aerial photography area is deduced according to the ground sampling distance. The formula is as follows:

[0025] S=(H / 18.9) 2 *(P / 1000) (3),

[0026] Among them, S is the actual area of ​​the aerial photography area in square meters, H is the flight altitude of the drone, and P is the total pixels of the multispectral image. By counting the total number of pixels in the image and the number of pixels of green tide algae, the actual area of ​​the multispectral image and the actual area of ​​the green tide can be obtained respectively.

[0027] A further improvement of the technical solution of the present invention is that in step 7, an innovative "preset standard area method" is proposed to verify the error of area monitoring. By setting multiple green tide algae samples of different areas in the aerial photography area, the error of area monitoring can be obtained by comparing the actual area and the monitored area.

[0028] Due to the adoption of the above technical solution, the technical progress achieved by the present invention is:

[0029] The drone images used in the present invention have higher spatial resolution than satellite images, and the monitoring area can be flexibly adjusted. They are more portable and less affected by the atmosphere. Therefore, drone monitoring of green tides can greatly make up for the insufficient accuracy of remote sensing images in the early monitoring of green tides. The authenticity of satellite remote sensing monitoring can also be verified based on the characteristic that the spatial resolution of drones is close to the true value.

[0030] The present invention uses a multi-spectral instrument onboard an unmanned aerial vehicle to collect data in the green tide algae outbreak area. Through the recognition and segmentation processing in the multi-spectral image, combined with the corresponding relationship formula between the flight altitude of the unmanned aerial vehicle and the aerial photography area, the green tide area can be quickly monitored anytime, anywhere.

[0031] The "substitution sensor" radiation correction method used in the present invention is simple to operate and can achieve rapid radiation correction of multispectral images without any additional parameters.

[0032] The "preset standard area" error verification method proposed in the present invention is simple to operate, highly reliable, and can truly reflect the accuracy of green tide area monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a data processing flow chart of the present invention. DETAILED DESCRIPTION

[0034] In order to make the purpose, content and advantages of the present invention clearer, the specific implementation methods of the present invention are further described in detail below in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all 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.

[0035] Reference Figure 1 As shown, the present invention provides a method for monitoring the area of ​​green tide algae based on a multi-spectrometer on a drone. The process of using the multi-spectrometer on a drone to monitor the area of ​​green tide is as follows: first, set parameters: plan the aerial photography area, plan the route, and set the overlap rate; then execute the aerial photography task: use the multi-spectral version of the DJI Phantom 4 flight system to perform aerial photography in the green tide outbreak area; then export the multi-spectral image data: after completing the task, export the drone data to a computer; then perform multi-spectral image processing: use Pix4dmapper software to splice the aerial multi-spectral images into a large image; use vegetation index to identify green tide algae in multi-spectral images; use the threshold segmentation method to segment the multi-spectral image, segment the green tide algae, and count the number of pixels occupied by the green tide algae; finally, calculate the green tide area: the pixel corresponding to the actual area obtained by GSD, combined with the number of pixels occupied by the green tide algae, can obtain the actual area of ​​the green tide.

[0036] In this embodiment, the green tide area monitoring method based on the multi-spectrometer onboard an unmanned aerial vehicle includes the following steps:

[0037] Step 1: Multispectral data acquisition method. The equipment used in the present invention is the DJI Phantom 4 multispectral version integrated system. The aerial photography area is planned. The multispectral instrument is used on a drone. The shooting mode is set to take photos at equal time intervals. The flight altitude is set to 50 meters. The aerial photography pitch angle is set to vertically downward. The heading and lateral overlap rate are set. After setting the flight parameters, take off to collect multispectral data. Each set of multispectral data collected contains a visible light image and five single-band images. The bands are: blue (B): 450nm±16nm, green (G): 560nm±16nm, red (R): 650nm±16nm, red edge (RE): 730nm±16nm, near infrared (NIR): 840nm±26nm, and the effective pixels of a single sensor are 2.08 million.

[0038] Step 2: Radiation correction method, using the "alternative sensor parameters" method to achieve rapid radiation correction of a single set of images, importing each single-band data in turn according to the wavelength, and then synthesizing the five single-band data into a multispectral data containing five bands. Using the fast atmospheric correction algorithm, select alternative sensors containing these five bands (such as AISA-ES, Landsat, MODIS sensors) for radiation correction. The corrected green tide algae spectral curve is consistent with the spectral characteristics of the vegetation spectrum, indicating that the correction result is correct.

[0039] Step 3: Image stitching method, use Pix4Dampper software to stitch the radiation corrected images. First, arrange all images according to the aerial photography position based on the coordinate position, then extract the feature points in adjacent images, and use the feature point matching method to stitch multiple single multispectral images into a large image covering the entire aerial photography area.

[0040] Step 4: Image recognition method. The vegetation index is a parameter formed by combining the spectral reflectance of different bands. Different objects have different reflectances in different bands. Since the green tide has a strong absorption capacity in the red light band and a strong reflection capacity in the near-infrared band, the "red edge effect" can be used to identify the green tide in the multispectral image. The vegetation index used in the present invention is the normalized green band difference vegetation index (GNDVI), and its formula is as follows:

[0041] GNDVI=(R nir -R green ) / (R nir +R green ) (1),

[0042] Where R represents the reflectance of each band. Based on the MATLAB environment, a green tide algae recognition algorithm based on the GNDVI vegetation index was written to identify green tide algae in multispectral images.

[0043] Step 5: Image segmentation method, use the maximum inter-class variance threshold segmentation method to segment the multispectral image after vegetation index processing. The maximum inter-class variance method uses the clustering idea to divide the grayscale value of the multispectral image into two parts, so that the grayscale value difference between the two parts is the largest, and the grayscale value difference within each part is the smallest. A suitable grayscale threshold is selected for segmentation through variance calculation, and the multispectral image is segmented into two parts of target green tide algae and other background objects through threshold segmentation.

[0044] Step 6: Area calculation method, based on the ground sampling distance (GSD) of the drone's multispectral instrument. GSD refers to the actual size represented by a unit pixel in a multispectral image. GSD is related to the flight altitude of the aircraft, the focal length of the imaging system, and the size of the pixel. The GSD formula for the DJI Phantom 4 Multispectral Edition is as follows:

[0045] GSD = H / 18.9 (cm / pixel) (2),

[0046] Among them, H is the flight height of the drone. The corresponding formula of the drone flight height and the aerial photography area is deduced according to the ground sampling distance. The formula is as follows:

[0047] S=(H / 18.9) 2 *(P / 1000) (3),

[0048] Among them, S is the real area of ​​the aerial photography area in square meters, H is the flight altitude of the drone, and P is the total number of pixels in the multispectral image. By counting the total number of pixels in the image and the number of pixels of green tide algae, the real area of ​​the multispectral image and the real area of ​​the green tide algae can be obtained respectively.

[0049] Step 7: Error verification method. The “preset standard area method” is proposed. By setting up multiple green tide algae samples of different areas in the aerial photography area and comparing the monitored area with the actual area, the error of green tide area monitoring can be obtained.

[0050] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for monitoring green tide algae area based on a multi-spectrometer on an unmanned aerial vehicle, characterized in that: The following steps are involved: Step 1: Green tide multispectral data collection: Use a drone equipped with a multispectral instrument to collect data in the green tide algae outbreak area; In the step 1, a multi-spectral integrated system of the DJI Phantom 4 is used to plan the aerial photography area, a multi-spectral instrument is mounted on a drone, the shooting mode is set to take pictures at equal time intervals, the flight altitude is set to 50 meters, the aerial photography pitch angle is set to vertically downward, the heading and lateral overlap rate are set, and after setting the flight parameters, take off to collect multi-spectral data; wherein each set of multi-spectral data collected contains a visible light image and five single-band images, and the bands are: blue B: 450nm±16nm, green G: 560nm±16nm, red R: 650nm±16nm, red edge RE: 730nm±16nm, near infrared NIR: 840nm±26nm, and the effective pixels of a single sensor are 2.08 million; Step 2, multispectral data preprocessing: performing radiation correction preprocessing on the collected multispectral data; Step 3, image stitching: performing multispectral image stitching processing on the multispectral data after radiation correction; Step 4, green tide algae identification: using the vegetation index method to identify green tide algae in multispectral images; In step 4, the vegetation index is a parameter formed by combining the spectral reflectance of different bands. Different objects have different reflectances in different bands. Green tide algae have a strong absorption capacity in the red light band and a strong reflection capacity in the near-infrared band. The "red edge effect" is based on this principle to identify green tide algae in multispectral images. The vegetation index is the normalized green band difference vegetation index GNDVI, and its formula is as follows: GNDVI=(R nir -R green ) / (R nir +R green ) (1), Where R represents the reflectivity of each band; Based on the MATLAB environment, the green tide algae recognition algorithm of the GNDVI vegetation index is compiled to identify the green tide algae in multispectral images; Step 5, background segmentation: using the threshold segmentation method, the multispectral image after vegetation index recognition is segmented into two parts: green tide algae and other background; In the step 5, the multispectral image processed by the vegetation index is segmented using the maximum inter-class variance threshold segmentation method. The maximum inter-class variance method uses the idea of ​​clustering to divide the grayscale value of the multispectral image into two parts, so that the grayscale value difference between the two parts is maximized, and the grayscale value difference within each part is minimized. A grayscale threshold is selected by calculating the variance for segmentation, and the multispectral image is segmented into two parts of target green tide algae and other background objects by threshold segmentation; Step 6, calculation of green tide area: using the ground sampling distance of the drone, the actual area represented by a single pixel is obtained, the corresponding relationship formula between the drone flight altitude and the aerial photography area is derived, and the number of pixels of green tide algae is counted to obtain the actual area of ​​the green tide; In step 6, according to the ground sampling distance GSD of the multispectral instrument of the UAV, GSD refers to the actual size represented by the unit pixel in the multispectral image. GSD is related to the flight altitude of the aircraft, the lens focal length of the imaging system, and the size of the pixel. The formula is as follows: GSD = H / 18.9 (cm / pixel) (2), Among them, H is the flight height of the drone. The corresponding formula of the drone flight height and the aerial photography area is deduced according to the ground sampling distance. The formula is as follows: S=(H / 18.9) 2 *(P / 10000) (3), Among them, S is the real area of ​​the aerial photography area, in square meters, H is the flight altitude of the drone, and P is the total pixels of the multispectral image. By counting the total number of pixels in the image and the number of pixels of the green tide algae, the real area of ​​the multispectral image and the real area of ​​the green tide can be obtained respectively; Step 7, calculation of green tide area monitoring error: set up multiple green tide algae samples of fixed area in the aerial photography area in advance. By comparing the actual area of ​​the green tide algae samples with the monitoring area of ​​the algorithm, the error of green tide area monitoring can be verified.

2. The method for monitoring the area of ​​green tide algae based on a multi-spectrometer on an unmanned aerial vehicle according to claim 1, characterized in that: The method of rapid radiation correction in step 2 is the "alternative sensor correction method", and the specific implementation method is: importing each single-band data in sequence according to the wavelength, and then synthesizing the five single-band data into a multi-spectral data containing five bands, and using the rapid atmospheric correction algorithm, selecting an alternative sensor containing these five bands for radiation correction. The corrected green tide algae spectral curve conforms to the spectral characteristics of the vegetation spectrum, indicating that the correction result is correct.

3. The method for monitoring green tide algae area based on a multi-spectrometer on an unmanned aerial vehicle according to claim 1, characterized in that: In step 3, the radiation-corrected images are stitched using Pix4Dampper software: first, all images are arranged according to the aerial photography positions based on the coordinate positions, and then feature points in adjacent images are extracted and stitched using a feature point matching method, so that multiple single multispectral images are stitched into a large image covering the entire aerial photography area.

4. The method for monitoring green tide algae area based on a multi-spectrometer on an unmanned aerial vehicle according to claim 1, characterized in that: In step 7, by setting a plurality of green tide algae samples of different areas in the aerial photography area, the error of area monitoring can be obtained by comparing the actual area of ​​the green tide algae samples with the monitored area.

Images