A method and apparatus for mapping seagrass distribution

Abstract

The present application relates to satellite remote sensing image technical field, disclose a kind of seaweed distribution map drawing method and device, including obtaining the time series remote sensing image of to be identified area, based on water body index of each satellite remote sensing image therein, select sea-land separation boundary, construct time series marine area remote sensing image;Using multiscale segmentation method obtains multiple homogeneous patches in each marine area remote sensing image, and is classified;The region of all homogeneous patches belonging to marine water green plant type therein is obtained, as greenery area remote sensing image;Based on the red light band reflection value and the first vegetation red edge band reflection value of each greenery area remote sensing image, calculate submergence seagrass spectral index, to mark suspected seagrass pixel;The amplitude dispersion index of each suspected seagrass pixel point is calculated, and target seagrass pixel point is selected;Extract target seagrass pixel point of all greenery area remote sensing image in time series greenery area remote sensing image, and draw seaweed distribution map.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of satellite remote sensing image, in particular to a sea grass distribution map drawing method and device. BACKGROUND

[0002] Sea grass is a kind of higher aquatic plants growing in the shallow water area of the sea, which has high ecological value and is a key component of the marine blue carbon ecosystem. However, in recent years, the global sea grass is declining rapidly, and it is of great significance to protect the sea grass and its ecological function.

[0003] Sea grass survey is the prerequisite for implementing sea grass protection. The current conventional sea grass survey method is field investigation and remote sensing technology survey. Based on field observation, this method has defects such as long operation period, high implementation cost, and difficulty in detecting unknown sea areas. Satellite remote sensing technology not only has the ability of convenient and rapid large-scale statistical inference, but also has lower cost; using remote sensing technology to survey sea grass can quickly, conveniently, efficiently and cheaply realize large-scale sea grass distribution map drawing, which has significant advantages compared with the traditional field observation survey method.

[0004] However, in the process of using remote sensing technology to survey sea grass, both sea grass and seaweed floating on the surface of the sea water contain chlorophyll, and both of them show the spectral characteristics of plants in the remote sensing image, so it is difficult to accurately distinguish sea grass and seaweed, resulting in low accuracy of sea grass survey and low precision of sea grass distribution map drawing. SUMMARY

[0005] Therefore, the technical problem to be solved by the present application is to overcome the problem that the prior art cannot effectively distinguish sea grass and floating algae when using remote sensing technology to draw sea grass distribution map, resulting in inaccurate identification of sea grass distribution area and low precision of sea grass distribution map drawing.

[0006] To solve the above technical problems, the present application provides a sea grass distribution map drawing method, comprising:

[0007] Obtaining satellite remote sensing images of each sampling time of the to-be-identified area in a preset time period to form a time sequence remote sensing image of the to-be-identified area;

[0008] Based on the water body index of each satellite remote sensing image in the time sequence remote sensing image, a sea-land separation boundary is selected, and a marine area in each satellite remote sensing image is extracted to form a time sequence marine area remote sensing image;

[0009] For each marine area remote sensing image in the time sequence marine area remote sensing image, a multi-scale segmentation method is used for image segmentation to obtain a plurality of homogeneous image patches in each marine area remote sensing image;

[0010] Based on the feature parameters of the homogeneous patches, the category of each homogeneous patch in each marine area remote sensing image is obtained, and the category includes the marine water green plant type, the marine water body type and the benthic type;

[0011] A region composed of all homogeneous patches belonging to the marine water green plant type in each marine area remote sensing image is obtained as a green plant region remote sensing image corresponding to each marine area remote sensing image, and the time sequence green plant region remote sensing images are composed;

[0012] For each green plant region remote sensing image in the time sequence green plant region remote sensing images, based on the red light band reflection value and the first vegetation red edge band reflection value of each green plant region remote sensing image, the submergent seagrass spectral index of each green plant region remote sensing image is calculated, compared with the preset spectral index interval, and the suspected seagrass pixel points are marked; the amplitude dispersion index of each suspected seagrass pixel point is calculated, and all suspected seagrass pixel points with an amplitude dispersion index not greater than a preset amplitude threshold are obtained as target seagrass pixel points;

[0013] The target seagrass pixel points of all green plant region remote sensing images in the time sequence green plant region remote sensing images are extracted, and a seagrass distribution map is drawn.

[0014] Preferably, after obtaining the time sequence remote sensing images of the to-be-identified region, the method further comprises pre-processing each satellite remote sensing image in the time sequence remote sensing images, including:

[0015] Radiometric calibration, atmospheric correction, geometric correction, splicing and cropping are performed on each satellite remote sensing image to obtain a corresponding optimized remote sensing image;

[0016] One of the satellite remote sensing images after atmospheric correction is selected as a reference image, and the scale-invariant feature transform features of the reference image are extracted;

[0017] Based on the scale-invariant feature transform features of each optimized remote sensing image and the scale-invariant feature transform features of the reference image, the registration of each optimized remote sensing image is performed with the reference image as the reference, and the pre-processing of each satellite remote sensing image is completed.

[0018] Preferably, the water body index of each satellite remote sensing image in the time sequence remote sensing images is used to select a sea-land separation boundary, extract a marine area in each satellite remote sensing image, and compose time sequence marine area remote sensing images, including:

[0019] Based on the green band and near-infrared band reflection values of each satellite remote sensing image in the time sequence remote sensing images, the water body index of each satellite remote sensing image in the time sequence remote sensing images is calculated;

[0020] obtaining a water boundary corresponding to each satellite remote sensing image based on a water body index of each satellite remote sensing image and a preset water body index threshold value; obtaining a water boundary with a largest marine area range as a sea-land separation boundary;

[0021] cutting each satellite remote sensing image in the time-series remote sensing images of the to-be-identified region using the sea-land separation boundary, extracting a marine area of each satellite remote sensing image as a corresponding marine area remote sensing image, and composing time-series marine area remote sensing images.

[0022] Preferably, the water body index of each satellite remote sensing image in the time-series remote sensing images is calculated based on the green band and near-infrared band reflectance values of each satellite remote sensing image, and is represented as:

[0023] ;

[0024] wherein, the water body index is represented by, the green band is represented by, and the near-infrared band reflectance value is represented by.

[0025] Preferably, the segmentation scale, shape factor, compactness, spectral weight, and spatial weight of the preset multi-scale segmentation method are used to divide each marine area remote sensing image to obtain a plurality of homogeneous image patches in each marine area remote sensing image.

[0026] Preferably, the class of each homogeneous image patch in each marine area remote sensing image is obtained based on the feature parameters of the homogeneous image patch, and the class includes:

[0027] A training set is constructed by obtaining a plurality of typical homogeneous image patches and their feature parameters, and a target gradient boosting decision tree is obtained by training the gradient boosting decision tree;

[0028] The feature parameters of each homogeneous image patch include spectral features, texture features, contour features, and shape features.

[0029] Based on the feature parameters of each homogeneous image patch, the target gradient boosting decision tree is used for classification, and the homogeneous image patch is divided into one of the marine water green plant type, the marine water type, or the benthic type.

[0030] Preferably, the submersed seagrass spectral index of each green plant area remote sensing image is calculated based on the red band reflectance value and the first vegetation red edge band reflectance value of each green plant area remote sensing image, and is represented as:

[0031] ;

[0032] wherein, the submersed seagrass spectral index is represented by, represents the first vegetation red edge band reflectance value, represents the red band reflectance value, represents a preset adjustment coefficient.

[0033] Preferably, the amplitude dispersion index is represented as:

[0034] ;

[0035] wherein, represents the amplitude dispersion index, is the amplitude standard deviation of the pixel points, is the amplitude mean value of the pixel points.

[0036] Preferably, the satellite remote sensing image is a Sentinel-2 satellite remote sensing image.

[0037] The embodiment provides a seaweed distribution map drawing device, comprising:

[0038] an image acquisition module configured to acquire satellite remote sensing images of a to-be-identified region at each sampling time in a preset time period to form time-series remote sensing images of the to-be-identified region;

[0039] an ocean region acquisition module configured to select a land-sea separation boundary based on a water body index of each satellite remote sensing image in the time-series remote sensing images, extract an ocean region in each satellite remote sensing image, and form time-series ocean region remote sensing images;

[0040] a homogeneous patch extraction and classification module configured to, for each ocean region remote sensing image in the time-series ocean region remote sensing images, perform image segmentation by using a multi-scale segmentation method to obtain a plurality of homogeneous patches in each ocean region remote sensing image, and obtain a category of each homogeneous patch in each ocean region remote sensing image based on a feature parameter of the homogeneous patch, wherein the category includes a marine green plant type, an ocean water type and a benthic type;

[0041] a green plant region acquisition module configured to obtain a region composed of all homogeneous patches belonging to the marine green plant type in each ocean region remote sensing image as a green plant region remote sensing image corresponding to each ocean region remote sensing image, and form time-series green plant region remote sensing images;

[0042] The target seagrass pixel extraction module is configured to, for each green plant area remote sensing image in the time-series green plant area remote sensing images, calculate a submergent seagrass spectral index of each green plant area remote sensing image based on a red light band reflection value and a first vegetation red edge band reflection value of each green plant area remote sensing image, compare the submergent seagrass spectral index with a preset spectral index interval, and mark a suspected seagrass pixel; calculate an amplitude dispersion index of each suspected seagrass pixel, and obtain suspected seagrass pixels with an amplitude dispersion index not greater than a preset amplitude threshold as target seagrass pixels.

[0043] The distribution map drawing module is configured to extract the target seagrass pixels of all green plant area remote sensing images in the time-series green plant area remote sensing images, and draw a seagrass distribution map.

[0044] The above technical solution of the present application has the following beneficial effects compared with the prior art:

[0045] The seagrass distribution map drawing method provided by the present application obtains satellite remote sensing images of a to-be-identified region, uses a water body index to distinguish sea and land regions, and obtains a seagrass possible growth region; uses a multi-scale segmentation method to obtain a plurality of homogeneous patches, classifies the homogeneous patches, and obtains a green plant type in seawater; calculates a submergent seagrass spectral index and an amplitude dispersion index of a green plant area remote sensing image composed of the homogeneous patches of the green plant type in seawater, obtains target seagrass pixels from the submergent seagrass spectral index and the amplitude dispersion index, and draws a seagrass distribution map; the present application uses the plant spectral index, the submergent seagrass spectral index, and the amplitude dispersion index in the remote sensing image to distinguish the image pixels of seagrass and floating algae based on the different ecological properties of seagrass and floating algae, i.e., the growth position of seagrass is fixed over time, while the growth position of floating algae changes, accurately obtains the seagrass region in the to-be-identified region, improves the seagrass remote sensing investigation precision, and improves the accuracy of the drawn seagrass distribution map. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to make the content of the present application more easily understood, the present application will be further described in detail below according to specific embodiments of the present application and in conjunction with the drawings, in which

[0047] Figure 1 is a step flow chart of the seagrass distribution map drawing method provided by the present application;

[0048] Figure 2 is a detailed step discrimination flow chart of the seagrass map drawing provided by the present application;

[0049] Figure 3 is a seagrass distribution map drawing flow chart based on Sentinel-2 satellite remote sensing images provided by the present application. DETAILED DESCRIPTION

[0050] The present application will be further described below in conjunction with the drawings and specific embodiments so that those skilled in the art can better understand and implement the present application, but the embodiments are not intended to limit the present application.

[0051] Although both seaweed and algae show the spectral characteristics of plants in remote sensing images, in practice, seaweed is a perennial plant with developed root systems fixed to the bottom of the sea, so the location of most seaweed growth does not change except for the areas of natural death and new growth of seaweed; and algae have no root systems fixed, and the growth outbreak is often closely related to the deterioration of the sea water quality, and the location is always changing, which is affected by waves and sea winds. Therefore, the present application will be based on the different ecological properties of seaweed and algae, i.e. the significant difference between the fixed / changed growth locations of seaweed and algae over time, and the amplitude dispersion index ADI of the spectral index of submerged seaweed in remote sensing images will be used to distinguish the image pixels of seaweed and algae, so as to improve the accuracy of seaweed remote sensing investigation.

[0052] Referring to FIG. 1, a step flow chart of a seaweed distribution map drawing method provided by the present application is shown, which specifically includes: Figure 1

[0053] S101: acquiring satellite remote sensing images of each sampling time of a to-be-identified region in a preset time period to form time sequence remote sensing images of the to-be-identified region;

[0054] S102: selecting a sea-land separation boundary based on a water body index of each satellite remote sensing image in the time sequence remote sensing images, extracting a marine area in each satellite remote sensing image to form time sequence marine area remote sensing images;

[0055] The water body index is represented as: is a green band, is a near-infrared band reflection value; S103: for each marine area remote sensing image in the time sequence marine area remote sensing images, using a multi-scale segmentation method to perform image segmentation to obtain a plurality of homogeneous image patches in each marine area remote sensing image;

[0056] S104: based on the feature parameters of the homogeneous image patches, obtaining the category of each homogeneous image patch in each marine area remote sensing image, the category including marine water green plant type, marine water type and benthic type;

[0057]

[0058] ​​​​S105: Obtain a region composed of all homogeneous patches belonging to the green plant type in the sea water in each marine area remote sensing image as a green plant region remote sensing image corresponding to each marine area remote sensing image, and compose a time sequence green plant region remote sensing image;

[0059] S106: For each green plant region remote sensing image in the time sequence green plant region remote sensing image, based on the red light band reflection value and the first vegetation red edge band reflection value of each green plant region remote sensing image, calculate the submergent seagrass spectral index of each green plant region remote sensing image, compare with the preset spectral index interval, mark the suspected seagrass pixel points; calculate the amplitude dispersion index of each suspected seagrass pixel point, and obtain all suspected seagrass pixel points with an amplitude dispersion index not greater than a preset amplitude threshold as target seagrass pixel points;

[0060] The submergent seagrass spectral index , is expressed as: ; The first vegetation red edge band reflection value is represented by Rr, The red light band reflection value is represented by R, The preset adjustment coefficient is represented by a;

[0061] The amplitude dispersion index , is expressed as: ; The amplitude standard deviation of the pixel point is represented by σ, The amplitude mean of the pixel point is represented by μ;

[0062] S107: Extract the target seagrass pixel points of all green plant region remote sensing images in the time sequence green plant region remote sensing image, and draw a seagrass distribution map.

[0063] In the embodiment of the present application, after the time sequence remote sensing image of the to-be-identified region is obtained in step S101, the time sequence remote sensing image is further preprocessed, including:

[0064] Radiometric calibration, atmospheric correction, geometric correction, splicing and cropping are performed on each satellite remote sensing image to obtain a corresponding optimized remote sensing image;

[0065] Select one of the satellite remote sensing images after atmospheric correction as a reference image, and extract the scale-invariant feature transform feature of the reference image;

[0066] Based on the scale-invariant feature transform feature of each optimized remote sensing image and the scale-invariant feature transform feature of the reference image, the reference image is taken as a reference to register each optimized remote sensing image, and the preprocessing of each satellite remote sensing image is completed.

[0067] Specifically, in step S102, the construction of the time sequence marine area remote sensing image includes:

[0068] S102-1: Calculate a water body index of each satellite remote sensing image in the time-series remote sensing images based on the green band and near-infrared band reflection values of the satellite remote sensing image in each scene of the time-series remote sensing images;

[0069] S102-2: Obtain a water boundary corresponding to each satellite scene remote sensing image based on the water body index of each satellite scene remote sensing image and a preset water body index threshold; obtain a water boundary with the largest range of marine area as a sea-land separation boundary;

[0070] S102-3: Use the sea-land separation boundary to crop each satellite remote sensing image in the time-series remote sensing images of the to-be-identified region, extract the marine area of each satellite remote sensing image as a corresponding marine area remote sensing image, and compose a time-series marine area remote sensing image.

[0071] In the embodiment of the present application, when the multi-scale segmentation method is used for image segmentation to obtain homogeneous image patches, the segmentation scale, shape factor, compactness, spectral weight and spatial weight of the preset multi-scale segmentation method are used to divide each marine area remote sensing image to obtain a plurality of homogeneous image patches in each marine area remote sensing image.

[0072] Specifically, in step S104, the homogeneous image patches are classified, including:

[0073] S104-1: Obtain a plurality of typical homogeneous image patches and their characteristic parameters to construct a training set, train the gradient boosting decision tree, and obtain a target gradient boosting decision tree;

[0074] S104-2: Extract the characteristic parameters of each homogeneous image patch, including spectral features, texture features, contour features and shape features;

[0075] S104-3: Based on the characteristic parameters of each homogeneous image patch, use the target gradient boosting decision tree for classification, and divide the homogeneous image patches into one of the marine water green plant type, marine water type or benthic type.

[0076] The seaweed distribution mapping method provided by the present application, after obtaining the satellite remote sensing image of the to-be-identified region, distinguishes the sea and land regions by using the water body index, and obtains the seaweed possible growth region; after obtaining a plurality of homogeneous image patches by using a multi-scale segmentation method, the homogeneous image patches are classified to obtain the green plant type in the seawater; the seaweed distribution map is drawn by obtaining the target seaweed pixel points from the seaweed spectrum index and the amplitude deviation index of the green plant region remote sensing image composed of the homogeneous image patches of the green plant type in the seawater; the present application distinguishes the image pixels of seaweed and floating algae by using the calculation of the plant spectrum index, the seaweed spectrum index and the amplitude deviation index in the remote sensing image according to the different ecological properties of seaweed and floating algae, i.e. the growth position of seaweed is fixed over time, while the growth position of floating algae changes, so that the seaweed region in the to-be-identified region is accurately obtained, the seaweed remote sensing investigation precision is improved, and the accuracy of the drawn seaweed distribution map is improved.

[0077] Referring to Figure 2 Fig. 1 is a detailed step discrimination flowchart of the seaweed mapping provided by the embodiment of the present application; the present application obtains the time series remote sensing image in the to-be-identified region, and pre-processes the time series remote sensing image; the water boundary of the time series image is extracted and cropped to determine the seawater coverage region as the seaweed possible growth region, i.e. the marine region; the time series remote sensing image of the marine region is classified by using the object-oriented method, the time series remote sensing image is divided into three categories of green plants in seawater, seawater and other benthic organisms by the steps of segmenting homogeneous image patches, extracting feature parameters and running classification algorithms; the submerged seaweed extraction index is calculated for the green plants in seawater to obtain the time series seaweed spectrum image; the amplitude deviation index of each pixel is calculated for the time series seaweed spectrum image; the amplitude deviation index can reflect whether the seaweed spectrum in each marine position remains unchanged; according to the calculation result of the amplitude deviation index, the positions with unchanged seaweed spectrum in the image are marked as seaweed, and the positions with changed seaweed spectrum are marked as floating algae and deleted, so that the floating algae in the seaweed identification process is removed.

[0078] Based on the above embodiment, in the embodiment of the present application, the satellite remote sensing image collected by the Sentinel-2 satellite is used to draw the seaweed distribution map, referring to Figure 3 Fig. 2 is a seaweed distribution map drawing flowchart based on the Sentinel-2 satellite remote sensing image, and the specific steps include:

[0079] S201: obtaining the time series Sentinel-2 satellite remote sensing image of the research region within 2 years with a time interval of 10 days;

[0080] S202: The Sentinel-2 satellite remote sensing image is radiometrically calibrated using the SANP software, the dark spectrum fitting method is used to perform atmospheric correction using the ALLCOT software, and the ARCGIS software is used to perform geometric correction, splicing and cutting of all images;

[0081] S203: Select one image of the middle period from the time series images after atmospheric correction as the reference image, extract the SIFT features of the image using the ARCGIS software; use the ARCGIS software to register other images based on the reference image;

[0082] S204: Calculate the water body index of each image in the registered time series image using the ENVI software , the calculation formula is: ; is the green band, is the reflectance value of the near-infrared band.

[0083] S205: Set threshold value of the calculation result, determine the water boundary extraction threshold range; determine the maximum range water boundary at the highest tide from the extracted multi-period water boundary as the sea-land separation boundary of seagrass growth;

[0084] S206: Use the maximum range water boundary to cut the time series image, determine the seawater coverage range as the possible seagrass growth area (i.e. marine area), and retain it, remove the remaining part to eliminate the spectral interference of land vegetation and ground objects;

[0085] S207: Perform image segmentation on each image of the possible seagrass growth area, extract homogeneous patches, and use a multi-scale segmentation method to achieve it; divide the segmented image homogeneous patches into three types: marine green plants in water, marine water body and other bottom;

[0086] During classification, feature parameters are extracted for each type of patch, including spectral features, texture features, contour features and shape features, etc., and typical homogeneous patches are selected as samples to use the gradient boosting decision tree GBDT method for classification, obtaining three image types of marine green plants in water, marine water body and other bottom;

[0087] S208: For the marine green plants in water type, extract the distribution range, perform image cutting, and obtain the green vegetation area time series image;

[0088] S209: For the green vegetation area time series image, calculate the submerged seagrass spectral index SSII to obtain the time series seagrass spectral image;

[0089] The calculation formula of the submerged seagrass spectral index SSII is: ; a reflectance value representing a first vegetation red edge band, a reflectance value representing a red band, a preset adjustment coefficient, which is a minimum value other than 0;

[0090] The SSII value of the image of the seaweed coverage area should be greater than 1, and the greater the SSII value, the higher the seaweed coverage density of the area.

[0091] S210: setting an SSII threshold for the time-series seaweed spectral image, marking the pixel points in the threshold range as suspected seaweed; calculating the amplitude dispersion index of each pixel in the time-series seaweed spectral image , which can reflect whether the seaweed spectrum at each position remains unchanged;

[0092] The calculation formula of the amplitude dispersion index is as follows: ; is the amplitude standard deviation, and the amplitude mean.

[0093] Setting a threshold for the amplitude dispersion index , marking the pixel points greater than the threshold as floating algae;

[0094] S211: removing all pixel points marked as algae in the image, and then determining the remaining pixel points as seaweed, thereby realizing the removal of algae in the seaweed identification process, so as to draw a seaweed distribution map according to the area composed of the remaining pixel points.

[0095] Based on the attribute feature that the geographical spatial positions of the seaweed and floating algae growth areas in the real ground environment are fixed or changed, the present application effectively distinguishes seaweed from algae by using remote sensing images, reduces the error influence caused by floating algae in seaweed remote sensing monitoring, improves the seaweed remote sensing investigation accuracy, and further improves the drawing accuracy of the seaweed distribution map.

[0096] Based on the above embodiment, the seaweed distribution map drawing device provided by the embodiment of the present application can specifically include:

[0097] An image acquisition module 100 is configured to acquire satellite remote sensing images of a to-be-identified area at each sampling time in a preset time period to form a time-series remote sensing image of the to-be-identified area.

[0098] An ocean area acquisition module 200 is configured to select a land-sea separation boundary based on a water body index of each satellite remote sensing image in the time-series remote sensing image, extract an ocean area in each satellite remote sensing image, and form a time-series ocean area remote sensing image.

[0099] The homogeneous image patch extraction and classification module 300 is configured to perform image segmentation on each marine area remote sensing image in the time-series marine area remote sensing images by using a multi-scale segmentation method to obtain a plurality of homogeneous image patches in each marine area remote sensing image; and obtain the category of each homogeneous image patch in each marine area remote sensing image based on the feature parameters of the homogeneous image patch, wherein the category includes a marine water green plant type, a marine water body type and a benthic type.

[0100] The green plant area acquisition module 400 is configured to obtain a region composed of all homogeneous image patches belonging to the marine water green plant type in each marine area remote sensing image as a green plant area remote sensing image corresponding to each marine area remote sensing image, and to compose a time-series green plant area remote sensing image.

[0101] The target seagrass pixel point extraction module 500 is configured to calculate a submergent seagrass spectral index of each green plant area remote sensing image based on the red light band reflectance and the first vegetation red edge band reflectance of each green plant area remote sensing image, compare the submergent seagrass spectral index with a preset spectral index interval, and mark a suspected seagrass pixel point; and calculate an amplitude dispersion index of each suspected seagrass pixel point, and obtain suspected seagrass pixel points with an amplitude dispersion index not greater than a preset amplitude threshold as target seagrass pixel points.

[0102] The distribution map drawing module 600 is configured to extract the target seagrass pixel points of all green plant area remote sensing images in the time-series green plant area remote sensing images, and draw a seagrass distribution map.

[0103] The seagrass distribution map drawing device of the embodiment is used to implement the seagrass distribution map drawing method described above, and thus the specific embodiments of the seagrass distribution map drawing device can refer to the embodiments of the seagrass distribution map drawing method described above, for example, the image acquisition module 100 and the marine area acquisition module 200 are respectively used to implement steps S101 and S102 in the seagrass distribution map drawing method described above; the homogeneous image patch extraction and classification module 300 is used to implement steps S103 and S104 in the seagrass distribution map drawing method described above; and the green plant area acquisition module 400, the target seagrass pixel point extraction module 500 and the distribution map drawing module 600 are respectively used to implement steps S105, S106 and S107 in the seagrass distribution map drawing method described above, and thus the specific embodiments can refer to the descriptions of the respective embodiments, and details are not repeated here.

[0104] The seaweed distribution mapping method provided by the present application, after obtaining the satellite remote sensing image of the region to be identified, distinguishes the sea and land regions by using the water body index, and obtains the seaweed possible growth region; after obtaining a plurality of homogeneous image patches by using the multi-scale segmentation method, the homogeneous image patches are classified to obtain the green plant type in the marine water; the seaweed distribution map is drawn by obtaining the target seaweed pixel points from the underwater seaweed spectral index and the amplitude deviation index of the green plant region remote sensing image composed of the homogeneous image patches of the green plant type in the marine water; the present application distinguishes the image pixels of seaweed and floating algae by using the calculation of the plant spectral index, the underwater seaweed spectral index and the amplitude deviation index in the remote sensing image according to the different ecological properties of seaweed and floating algae, i.e., the growth position of seaweed is fixed over time, while the growth position of floating algae is variable, so that the seaweed region in the region to be identified can be accurately obtained, the seaweed remote sensing investigation precision is improved, and the accuracy of the drawn seaweed distribution map is improved.

[0105] Those skilled in the art will understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer usable program code.

[0106] The present application is described with reference to flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks. Figure 1 The devices that implement the functions specified in one or more flows and / or blocks.

[0107] These computer program instructions can also be stored in a computer-readable memory that can guide the computer or other programmable data processing devices to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction devices that implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks. Figure 1 The devices that implement the functions specified in one or more flows and / or blocks.

[0108] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable data processing devices to generate a computer implemented process, so that the instructions executed on the computer or other programmable data processing devices provide steps for implementing the function specified in the flowchart Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0109] Obviously, the above embodiments are only examples for clearly illustrating the present application, and are not intended to limit the embodiments. Based on the above description, other different forms of changes or variations can be made by those skilled in the art. Here, all the embodiments are not required to be exhausted, and the obvious changes or variations derived therefrom are still within the protection scope of the present application.

Claims

1. A method for drawing a seagrass distribution map, characterized in that, include: The satellite remote sensing images of the area to be identified are acquired at each sampling time within a preset time period to form a time-series remote sensing image of the area to be identified. Based on the water index of each satellite remote sensing image in the time series remote sensing image, the land-sea separation boundary is selected, and the marine area in each satellite remote sensing image is extracted to form a time series marine area remote sensing image. For each scene of ocean region remote sensing image in the time series ocean region remote sensing image, the multi-scale segmentation method is used to segment the image and obtain multiple homogeneous image spots in each scene of ocean region remote sensing image. Based on the feature parameters of homogeneous patches, the category of each homogeneous patch in each marine remote sensing image is obtained, and the category includes marine green plant type, marine water body type and benthic type; The region consisting of all homogeneous patches belonging to the type of green plants in the ocean is obtained in each marine remote sensing image and used as the green plant region remote sensing image corresponding to each marine remote sensing image, forming a time-series green plant region remote sensing image. For each scene of a green vegetation area in a time-series remote sensing image, the submerged seagrass spectral index is calculated based on the red band reflectance value and the first vegetation red edge band reflectance value of each scene. This index is compared with a preset spectral index range to mark suspected seagrass pixels. The amplitude deviation index of each suspected seagrass pixel is calculated, and a threshold for the amplitude deviation index is set. Pixels with an amplitude deviation index greater than the threshold are marked as floating algae. All pixels marked as algae in the image are removed. All suspected seagrass pixels with an amplitude deviation index not greater than the preset amplitude threshold are obtained as target seagrass pixels. Extract the target seagrass pixels from all remote sensing images of green vegetation areas in the time-series green vegetation area remote sensing images, and draw a seagrass distribution map; The spectral index of submerged seagrass in each remote sensing image of a green vegetation area is represented as follows: ; Indicates the spectral index of submerged seagrass. This represents the reflectance value of the first vegetation red-edge band. This indicates the reflectance value in the red light band. This represents the preset adjustment coefficient; the amplitude deviation index of each suspected seaweed pixel is calculated and expressed as: ;in, Indicates the amplitude deviation index. The standard deviation of the amplitude of a pixel. This represents the average amplitude of the pixel.

2. The method for drawing seagrass distribution maps according to claim 1, characterized in that, After acquiring the time-series remote sensing imagery of the area to be identified, the process also includes preprocessing each satellite remote sensing image in the time-series imagery, including: For each satellite remote sensing image, radiometric calibration, atmospheric correction, geometric correction, mosaicking, and cropping are performed to obtain the corresponding optimized remote sensing image. Select one scene from all atmospherically corrected satellite remote sensing images as a reference image, and extract the scale-invariant feature transformation features of the reference image. Based on the scale-invariant feature transformation characteristics of each optimized remote sensing image and the scale-invariant feature transformation characteristics of the reference image, the optimized remote sensing image is registered with the reference image as a reference, thus completing the preprocessing of each satellite remote sensing image.

3. The method for drawing seagrass distribution maps according to claim 1, characterized in that, The method involves selecting the land-sea separation boundary based on the water index of each satellite remote sensing image in the time-series remote sensing imagery, extracting the ocean region from each satellite remote sensing image, and assembling a time-series ocean region remote sensing image, including: Based on the green band and near-infrared band reflectance values ​​of each satellite remote sensing image in the time-series remote sensing image, the water index of each satellite remote sensing image in the time-series remote sensing image is calculated. Based on the water index of each satellite remote sensing image and the preset water index threshold, the water boundary corresponding to each satellite remote sensing image is obtained; the water boundary with the largest marine area is obtained as the sea-land separation boundary. By utilizing the land-sea separation boundary, each satellite remote sensing image in the time-series remote sensing image of the area to be identified is cropped, and the marine region of each satellite remote sensing image is extracted as the corresponding marine region remote sensing image, thus forming a time-series marine region remote sensing image.

4. The method for drawing seagrass distribution maps according to claim 3, characterized in that, The water index for each satellite remote sensing image in the time-series remote sensing image is calculated based on the green band and near-infrared band reflectance values ​​of each image. This index is expressed as follows: ； in, Indicates water quality index, Green band This is the reflectance value in the near-infrared band.

5. The method for drawing seagrass distribution maps according to claim 1, characterized in that, The multi-scale segmentation method is pre-defined with segmentation scale, shape factor, compactness, spectral weight, and spatial weight. Each marine remote sensing image is then segmented to obtain multiple homogeneous patches within each marine remote sensing image.

6. The method for drawing seagrass distribution maps according to claim 1, characterized in that, The method of obtaining the category of each homogeneous patch in each marine remote sensing image based on the feature parameters of homogeneous patches includes: A training set is constructed by acquiring multiple typical homogeneous images and their feature parameters, and the gradient boosting decision tree is trained to obtain the target gradient boosting decision tree. Extract the feature parameters of each homogeneous image spot, including spectral features, texture features, contour features, and shape features; Based on the feature parameters of each homogeneous image patch, a target gradient boosting decision tree is used for classification, dividing the homogeneous image patches into one of the following: marine green plant type, marine water body type, or benthic type.

7. The method for drawing seagrass distribution maps according to claim 1, characterized in that, The satellite remote sensing image is a Sentinel-2 satellite remote sensing image.

8. A seagrass distribution map drawing apparatus based on the seagrass distribution map drawing method according to any one of claims 1 to 7, characterized in that, include: The image acquisition module is used to acquire satellite remote sensing images of the area to be identified at each sampling time within a preset time period, forming a time-series remote sensing image of the area to be identified. The ocean region acquisition module is used to select the land-sea separation boundary based on the water index of each satellite remote sensing image in the time-series remote sensing image, extract the ocean region in each satellite remote sensing image, and form a time-series ocean region remote sensing image. The homogeneous image patch extraction and classification module is used to segment each scene of marine remote sensing imagery in time-series marine remote sensing images using a multi-scale segmentation method to obtain multiple homogeneous images in each scene of marine remote sensing imagery. Based on the feature parameters of homogeneous patches, the category of each homogeneous patch in each marine remote sensing image is obtained, and the category includes marine green plant type, marine water body type and benthic type; The green plant area acquisition module is used to acquire the area composed of all homogeneous images of green plants in the ocean in each marine remote sensing image, which is used as the green plant area remote sensing image corresponding to each marine remote sensing image, forming a time-series green plant area remote sensing image. The target seagrass pixel extraction module is used to calculate the submerged seagrass spectral index of each green vegetation area remote sensing image in the time-series green vegetation area remote sensing image based on the red band reflectance value and the first vegetation red edge band reflectance value of each green vegetation area remote sensing image. It compares the index with a preset spectral index range and marks suspected seagrass pixels. It calculates the amplitude deviation index of each suspected seagrass pixel, sets a threshold for the amplitude deviation index, marks pixels with an amplitude deviation index greater than the threshold as floating algae, removes all pixels marked as algae in the image, and obtains all suspected seagrass pixels with an amplitude deviation index not greater than the preset amplitude threshold as target seagrass pixels. The distribution map drawing module is used to extract the target seagrass pixels from all remote sensing images of green vegetation areas in the time-series green vegetation area remote sensing images and draw a seagrass distribution map.

Images