Apparatus and method for precisely detecting blue carbon vegetation area on basis of remote sensing

Abstract

The present invention relates to a method for precisely detecting a blue carbon vegetation area on the basis of remote sensing, the method comprising the steps of: receiving a spectral image and recording same in memory; determining a blue carbon vegetation area on the basis of spectral information of blue carbon vegetation from the spectral image; determining a non-blue carbon vegetation area on the basis of spectral information of non-blue carbon vegetation from the spectral image; generating information about an overlapping area in which the blue carbon vegetation area and the non-blue carbon vegetation area overlap; determining a final blue carbon vegetation area in which the overlapping area has been deleted from the blue carbon vegetation area by a processor; and providing the final blue carbon vegetation area.

Description

Remote Sensing-Based Blue Carbon Vegetation Precision Detection Device and Method

[0001] The present invention relates to a method for precise detection of blue carbon vegetation sites based on remote sensing, and more specifically, to a method for precisely estimating and detecting the habitat of blue carbon vegetation inhabiting tidal boundary areas.

[0002] Generally, blue carbon refers to carbon sinks in marine ecosystems, such as marine organisms, and includes mangroves, salt marshes, mudflats, and seagrasses. Blue carbon is attracting attention for its potential in carbon neutrality and responding to the climate crisis.

[0003] A known related technology is Korean registered patent No. 10-2366267 (registered on February 17, 2022, vegetation status analysis device and method of operation thereof).

[0004] The above registered patent describes an apparatus and method for acquiring multispectral images of a habitat area for halophytes using an unmanned aerial vehicle, and for analyzing the vegetation status of halophytes more accurately using multispectral images captured at various altitudes.

[0005] However, since the first multispectral image captured at a first altitude is preprocessed to determine two or more regions of interest over a wide area, and a second multispectral image captured at a second altitude lower than the first altitude is acquired and processed based on the degree of overlap of the regions of interest to analyze the vegetation status, it had limitations in that the processing speed was delayed due to the increased number of image acquisitions required for analysis, and costs, particularly regarding the operation of unmanned aerial vehicles, increased relatively.

[0006] Furthermore, while recent vegetation analysis requires technological development aimed at increasing precision, there are limitations to improving precision even when multispectral images are acquired again using the overlap between areas of interest, because the images captured at the corresponding altitude are simply processed.

[0007] The problem that the present invention aims to solve, taking into account the market demands mentioned above, is to provide a remote sensing-based method for precise detection of blue carbon vegetation sites that enables more precise analysis of blue carbon vegetation sites.

[0008] More specifically, the present invention aims to provide a method for detecting blue carbon vegetation more precisely by obtaining blue carbon vegetation detection results and non-blue carbon vegetation detection results separately, identifying the overlapping area of ​​the two detection results, and excluding the overlapping area.

[0009] A method for precise detection of a blue carbon vegetation site according to an embodiment of the present invention, for solving the above technical problem, may include the steps of: the processor receiving a spectral image and recording it in the memory; the processor determining a blue carbon vegetation site based on spectral information of blue carbon vegetation from the spectral image; the processor determining a non-blue carbon vegetation site based on spectral information of non-blue carbon vegetation from the spectral image; the processor generating information regarding an overlapping area in which the blue carbon vegetation site and the non-blue carbon vegetation site overlap; the processor determining a final blue carbon vegetation site by deleting the overlapping area from the blue carbon vegetation site; and providing the final blue carbon vegetation site.

[0010] In one embodiment of the present invention, the step of determining the blue carbon vegetation area may include: a step of calculating NDMI (Normalized Difference Mangrove Index) information for the spectral image; and a step of determining the blue carbon vegetation area based on the result of calculating the NDMI information.

[0011] In one embodiment of the present invention, the step of determining the non-blue carbon vegetation area may include: a step of calculating Normalized Difference Vegetation Index (NDVI) information for the spectral image; and a step of determining the blue carbon vegetation area based on the result of calculating the NDVI information.

[0012] In one embodiment of the present invention, the processor further comprises the steps of determining information regarding the degree of moisture of a tidal boundary region from the spectral image; and the processor determining a tidal line from the information regarding the degree of moisture of the tidal boundary region, wherein the blue carbon vegetation area may be subjected to filtering in which the land side is excluded based on the tidal line, and the non-blue carbon vegetation area may be subjected to filtering in which the moist region is excluded.

[0013] In one embodiment of the present invention, the memory further includes a first deep learning algorithm and a second deep learning algorithm, and the processor further includes the step of inputting the filtered blue carbon vegetation area into the first deep learning algorithm to generate an inferred blue carbon vegetation area; and the processor further includes the step of inputting the filtered non-blue carbon vegetation area into the second deep learning algorithm to generate an inferred non-blue carbon vegetation area, and the final blue carbon vegetation area may be determined by the processor deleting the overlapping area in which the inferred non-blue carbon vegetation area overlaps with the inferred blue carbon vegetation area.

[0014] A precision detection device for blue carbon vegetation sites according to an embodiment of the present invention, for solving the above technical problem, comprises: a memory storing a plurality of instructions; and a processor for executing the instructions. The processor receives a spectral image and records it in the memory, determines a blue carbon vegetation site based on spectral information of blue carbon vegetation from the spectral image, determines a non-blue carbon vegetation site based on spectral information of non-blue carbon vegetation from the spectral image, generates information regarding an overlapping area where the blue carbon vegetation site and the non-blue carbon vegetation site overlap, and the processor determines a final blue carbon vegetation site by deleting the overlapping area from the blue carbon vegetation site, and can provide the final blue carbon vegetation site.

[0015] The present invention utilizes a remote sensing method to overcome the high cost of field surveys or multi-elevation imaging methods, and minimizes misclassification caused by the overlapping growth environments of blue carbon vegetation sites and non-blue carbon vegetation sites, thereby enabling the precise detection of blue carbon vegetation sites.

[0016] FIG. 1 is a flowchart of a remote sensing-based method for precise detection of blue carbon vegetation sites according to a preferred embodiment of the present invention.

[0017] Figure 2 is a detailed flowchart of step S10.

[0018] Figure 3 is a detailed flowchart of step S20.

[0019] Figure 4 is a detailed flowchart of step S30.

[0020] Figure 5 shows a remote sensing-based blue carbon vegetation precision detection device.

[0021] In order to fully understand the structure and effects of the present invention, preferred embodiments of the present invention are described with reference to the attached drawings. However, the present invention is not limited to the embodiments disclosed below, but can be implemented in various forms and various modifications can be made. The description of the embodiments is provided merely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention. In the attached drawings, the components are depicted with their sizes enlarged compared to their actual size for convenience of explanation, and the proportions of each component may be exaggerated or reduced.

[0022] Terms such as 'first' and 'second' may be used to describe various components, but said components should not be limited by said terms. These terms may be used solely for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, 'first component' may be named 'second component,' and similarly, 'second component' may be named 'first component.' Furthermore, singular expressions include plural expressions unless the context clearly indicates otherwise. Unless otherwise defined, terms used in the embodiments of the present invention may be interpreted in the sense commonly known to those skilled in the art.

[0023] Hereinafter, a remote sensing-based method for precise detection of blue carbon vegetation sites according to an embodiment of the present invention will be described in detail with reference to the drawings.

[0024] The present invention is for precisely detecting blue carbon vegetation sites by electrically processing spectral information of images and images, and the entity performing the method of the present invention is a computing device comprising a memory storing a plurality of instructions and at least one artificial intelligence model and a processor, and the method can be performed by the processor executing the instructions stored in the memory.

[0025] Even if there is no explicit description of the entity performing each step of the method of the present invention in the description below, it should be understood that the entity performing the step is a computing device or a processor of a computing device.

[0026] Referring to FIG. 5, a remote sensing-based blue carbon vegetation precision detection device (100) according to one embodiment of the present invention may include a processor (110), a memory (120), and a communication unit (130).

[0027] The processor (110) is configured to control the memory (120) and the communication unit (130) to enable the method of the present invention to be performed.

[0028] The communication unit (130) is configured to communicate with another terminal to receive spectroscopic images.

[0029] The processor (110) can provide the finally determined blue carbon vegetation information to an external terminal through the communication unit (130). The processor (110) may also control the display (140) to output the blue carbon vegetation information visually and / or audibly.

[0030] FIG. 1 is a flowchart of a remote sensing-based method for precise detection of blue carbon vegetation sites according to a preferred embodiment of the present invention.

[0031] Referring to FIG. 1, the present invention comprises the steps of: detecting spectral information of blue carbon vegetation, spectral information of non-target vegetation excluding blue carbon, and spectral information of wetness of tidal boundary areas in an image acquired through an image acquisition device capable of remote operation, such as a satellite, a drone, or a fixed camera (S10); detecting a blue carbon vegetation area using the spectral information of the blue carbon vegetation area and the spectral information of the tidal boundary area (S20); detecting a non-blue carbon vegetation area using the spectral information of the non-blue carbon vegetation area and the spectral information of an area not a tidal boundary area (S30); and refining the blue carbon vegetation area using the results of steps S20 and S30 (S40).

[0032] Hereinafter, the configuration and operation of the remote sensing-based blue carbon vegetation site precision detection method of the present invention, configured as described above, will be explained in more detail.

[0033] First, Figure 2 is a detailed flowchart of the above S10 step.

[0034] Referring to FIGS. 1 and FIGS. 2, in step S10, data is collected to identify the habitat of the blue carbon vegetation to be detected, and various spectral information proposed in the present invention is obtained.

[0035] In step S11, spectral images are collected, and spectral images are secured using image acquisition devices equipped with spectral sensors, such as satellites, drones, and fixed cameras.

[0036] Spectral images are defined as containing all information corresponding to the ultraviolet, infrared, and visible light regions.

[0037] The resolution of a spectroscopic image can be determined by the performance of the image acquisition device and can be classified into R, G, B, NIR (Near-Infrared), SWIR (Short Wave InfraRed), etc.

[0038] Spectral images can be acquired by determining a specific period during the growth phase, taking into account the growth characteristics of the blue carbon being detected.

[0039] In addition, spectral images can be collected by considering the meteorological conditions of the target area. Meteorological conditions may include, for example, tropical regions, subtropical regions, the dry season, or the rainy season.

[0040] As such, the present invention can acquire suitable spectral images by considering geographical location and time.

[0041] In the present invention, a total of three types of spectral information are obtained based on the same spectral image.

[0042] In step S12, spectral information of the blue carbon vegetation to be detected is obtained. In this invention, terms such as 'vegetation', 'vegetated area', and 'vegetated region' may be used, but they can be substantially defined as referring to a planar space where blue carbon grows.

[0043] Blue carbon vegetation can be determined using the shape of the object, the wavelength of reflection, etc. For example, blue carbon vegetation can be determined based on NDMI (Normalized Difference Mangrove Index) information derived from spectroscopic images. Blue carbon vegetation sites do not necessarily have to be based on NDMI information and may be determined based on other methods for identifying blue carbon vegetation sites.

[0044] In step S13, spectral information of vegetation that is not a detection target is obtained. In other words, the present invention obtains spectral information of other terrestrial vegetation that coexists with blue carbon vegetation. For example, non-blue carbon vegetation can be determined based on NDVI (Normalized Difference Vegetation Index) information derived from spectral images. Non-blue carbon vegetation areas do not necessarily need to be based on NDVI information and may be determined based on other methods for identifying non-blue carbon vegetation areas.

[0045]

[0046] The present invention obtains NDMI for blue carbon vegetation and NDVI for non-blue carbon vegetation, and as described below, can detect blue carbon vegetation more precisely using NDMI and NDVI.

[0047] In addition, in step S14, spectral information is obtained that takes into account the degree of wetness of the tidal boundary area.

[0048] In the present invention, vegetation areas are detected specifically for blue carbon, and since blue carbon is distributed mainly along coastlines due to its characteristics, it must be possible to reflect the moisture characteristics of the soil, and the degree of moisture in the tidal boundary area is confirmed using the Short Wave InfraRed (SWIR) component of the spectral image to determine the tidal line.

[0049] That is, the present invention can determine the tidal line in areas where soil moisture is high or the difference between high and low tides is large and apply primary filtering to the growing area.

[0050] Steps S12, S13, and S14 can be processed sequentially or simultaneously, and in sequential processing, they can be processed regardless of their order.

[0051] Figure 3 is a detailed flowchart of step S20.

[0052] Referring to Figures 1 and 3 respectively, in step S20, a primary detection of the blue carbon vegetation area is performed.

[0053] First, in step S21, the spectral information obtained in step S12 and the spectral information obtained in step S14 are synthesized or combined to form a composite image or a combined image.

[0054] In other words, spectral information regarding blue carbon vegetation is filtered first based on the tide line.

[0055] Blue carbon vegetation is determined on the land side based on the tide line.

[0056] Next, in step S22, a first estimation and detection of blue carbon vegetation is performed using a first deep learning algorithm that takes the processing result of step S21 as input.

[0057] The training data of the first deep learning algorithm above may be spatial information constructed by utilizing open-source data and field survey data.

[0058] In other words, it learns various spectral images of blue carbon and interprets synthetic images of input blue carbon vegetation sites based on the learning results to estimate and detect blue carbon vegetation sites.

[0059] At this time, the first deep learning algorithm can use a known image learning algorithm.

[0060] Figure 4 is a detailed flowchart of step S30.

[0061] Referring to Figures 1 and 4 respectively, step S30 performs a primary detection of the non-blue carbon vegetation area.

[0062] In step S31, the spectral information obtained in step S13 is filtered using spatial information for regions that do not correspond to the spectral information obtained in step S14.

[0063] In other words, it generates images of non-blue carbon vegetation in non-humid regions.

[0064] Next, in step S32, a second deep learning algorithm is used as input to perform primary estimation and detection of non-blue carbon vegetation sites.

[0065] The second deep learning algorithm of the above S32 step is different from the first deep learning algorithm of the above S22 step, and uses a deep learning algorithm trained using open source data and field survey data regarding non-blue carbon vegetation.

[0066] Next, in step S40, the primary detected blue carbon vegetation resulting from step S20 and the primary detected non-blue carbon vegetation resulting from step S30 are superimposed.

[0067] In other words, the two images are superimposed, and if there is an overlapping area between blue carbon vegetation and non-blue carbon vegetation, the overlapping area within the blue carbon vegetation is deleted.

[0068] This processing takes into account that the previously explained NDMI may contain spectral information about blue carbon vegetation sites due to errors, and that the NDVI may also contain spectral information about blue carbon vegetation sites due to errors.

[0069] Therefore, in the present invention, by considering the error of each index, the overlapping area of ​​NDMI and NDVI is identified, and the overlapping area is removed from NDMI to detect blue carbon vegetation more accurately.

[0070] Although embodiments according to the present invention have been described above, they are merely illustrative and those skilled in the art will understand that various modifications and equivalent embodiments are possible therefrom. Accordingly, the true technical scope of protection of the present invention should be determined by the following claims.

Claims

1. A method for precise detection of blue carbon vegetation sites performed on a computing device comprising a memory and a processor in which multiple instructions are stored, The step of the processor receiving a spectroscopic image and recording it in the memory; A step in which the processor determines a blue carbon vegetation area based on spectral information of blue carbon vegetation from the spectral image; A step in which the processor determines a non-blue carbon vegetation area based on spectral information of non-blue carbon vegetation from the spectral image; The step of the processor generating information regarding an overlapping area in which the blue carbon vegetation area and the non-blue carbon vegetation area overlap; A step in which the processor determines a final blue carbon vegetation area by deleting the overlapping area from the blue carbon vegetation area; and A method for precise detection of blue carbon vegetation, comprising the step of providing the final blue carbon vegetation.

2. In Paragraph 1, The step of determining the blue carbon vegetation area comprises: calculating NDMI (Normalized Difference Mangrove Index) information for the spectral image; A method for precise detection of blue carbon vegetation, comprising the step of determining the blue carbon vegetation based on the above NDMI information calculation results.

3. In Paragraph 1, The step of determining the above-mentioned non-blue carbon vegetation site is, A step of calculating NDVI (Normalized Difference Vegetation Index) information for the spectroscopic image; and A step of determining the blue carbon vegetation area based on the calculation result of the above NDVI information Precise detection method for blue carbon vegetation.

4. In Paragraph 1, The step of the processor determining information regarding the degree of wetness of the tidal boundary region from the spectroscopic image; and The above processor further includes the step of determining a tide line from information regarding the degree of wetness of the tide boundary area, and The above blue carbon vegetation is subjected to filtering that excludes the land side based on the above tide line, and The above-mentioned non-blue carbon vegetation is a blue carbon vegetation precision detection method in which filtering is applied to exclude wet areas.

5. In Paragraph 4, The above memory further includes a first deep learning algorithm and a second deep learning algorithm, and The step of the processor inputting the filtered blue carbon vegetation into the first deep learning algorithm to generate an inferred blue carbon vegetation; and The above processor further includes the step of inputting the filtered non-blue carbon vegetation into the second deep learning algorithm to generate an inferred non-blue carbon vegetation, and A method for precise detection of blue carbon vegetation, wherein the final blue carbon vegetation is determined by the processor deleting the overlapping area in which the inferred non-blue carbon vegetation overlaps with the inferred blue carbon vegetation.

6. Memory where multiple instructions are stored; and It includes a processor that executes the above instructions, The above processor is, Receive a spectroscopic image and record it in the memory, Based on the spectral information of blue carbon vegetation from the above spectral image, a blue carbon vegetation area is determined, and Determining non-blue carbon vegetation sites based on spectral information of non-blue carbon vegetation from the above spectral image, and The above processor generates information regarding an overlapping area where the blue carbon vegetation area and the non-blue carbon vegetation area overlap, and The processor determines the final blue carbon vegetation area by deleting the overlapping area from the blue carbon vegetation area, and A blue carbon vegetation precision detection device that provides the above-mentioned final blue carbon vegetation.

Images