Vegetation coverage measuring method and system based on computer vision
By using computer vision technology, combined with deep learning and spectral compensation, the error problem in vegetation cover measurement under complex terrain has been solved, and high-precision and fast vegetation cover calculation and ecological monitoring have been achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SOUTH CHINA SEA FISHERIES RES INST CHINESE ACAD OF FISHERY SCI
- Filing Date
- 2026-02-04
- Publication Date
- 2026-04-24
AI Technical Summary
Existing technologies for measuring vegetation cover, especially in complex terrain or densely vegetated areas, suffer from problems such as large measurement errors, high labor costs, time and effort consumption, and difficulty in achieving large-scale dynamic monitoring.
Using a computer vision-based approach, low-altitude aerial photography is conducted using a drone equipped with a multispectral camera. Combined with deep learning and an adaptive threshold shadow detection model, spectral compensation and correction of shadow areas are performed. Furthermore, vegetation pixels are identified through HSV color space conversion and geometric correction, coverage is calculated, and early warning and repair strategies are generated.
It significantly improves measurement accuracy under complex lighting conditions, enabling rapid and objective calculation of vegetation projection cover, generating archiveable gridded images and quantitative data for ecological monitoring and statistical analysis.
Smart Images
Figure CN121921655A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of technology, and in particular to a method and system for measuring vegetation cover based on computer vision. Background Technology
[0002] Vegetation fractional cover (VFC), as a key indicator for measuring the growth status and spatial distribution of surface vegetation, has significant application value in fields such as ecological assessment, soil and water conservation, climate change research, and agricultural production. The development of its precise measurement technology has evolved from traditional ground measurement to modern remote sensing inversion, with the core objective being to efficiently and accurately obtain the proportion of the vertically projected area of vegetation to the total area of the statistical region at different scales.
[0003] Early vegetation cover measurements relied primarily on field surveys, with typical methods including quadrat surveys, visual estimation, and sampling instrument methods (such as fisheye photography and leaf area indexers). While these methods could obtain relatively accurate point data, they suffered from inherent drawbacks such as high labor costs, time-consuming and labor-intensive processes, and limited spatial representativeness, making them unsuitable for large-scale, continuous dynamic monitoring. The implementation of traditional methods was particularly limited in areas with complex terrain or those that were difficult to access.
[0004] With the rise and development of remote sensing technology, vegetation cover measurement has entered a spatialized and dynamic era. Remote sensing methods indirectly invert vegetation cover by acquiring information on electromagnetic waves reflected or emitted from the Earth's surface, achieving a leap from "point" to "area." Among these methods, optical remote sensing, especially multispectral and hyperspectral remote sensing, has become the most important technological approach. By constructing empirical or physical relationship models between vegetation indices (such as the Normalized Difference Vegetation Index (NDVI) and the Enhanced Vegetation Index (EVI)) and vegetation cover (such as pixel-based bisection models, regression models, and machine learning models), coverage estimation at regional and even global scales can be achieved. However, optical remote sensing is susceptible to the influence of atmospheric conditions, cloud cover, soil background, and vegetation saturation, resulting in certain errors in complex terrain areas and densely vegetated areas. Summary of the Invention
[0005] This invention overcomes the shortcomings of the prior art and provides a method and system for measuring vegetation coverage based on computer vision.
[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: The first aspect of this invention provides a method for measuring vegetation cover based on computer vision, comprising the following steps: Image data information in the target area is acquired, and color space conversion and recognition are performed on the image data information, followed by geometric correction processing. Vegetation detection is performed on the corrected image. Vegetation pixels are identified using a preset HSV threshold, and related shapes are filtered based on shape features. The vegetation coverage is calculated from the images after vegetation detection, and the images are then divided according to the coverage. Based on coverage data, early warnings are issued for target areas, and relevant vegetation restoration strategies are generated.
[0007] Furthermore, in the computer vision-based vegetation cover measurement method, image data information of the target area is collected, specifically as follows: A drone equipped with a multispectral camera takes low-altitude aerial photos of the area to be tested to acquire high spatial resolution image data, and builds an adaptive threshold shadow detection model based on deep learning. The model inputs are the strong reflection characteristics of shadows in the near-infrared band, the solar elevation angle, the shooting time, and the brightness values of RGB images. The shadow areas in the images are used as the prediction results output. The adaptive threshold shadow detection model is used to segment the shadow areas in the image. For the identified shadow area vegetation, the spectral values of the shadow area pixels are compensated and corrected by using the spectral characteristics of the same type of vegetation under sunlight in the same plot. The compensated and corrected image data information is used as the image data information in the target area, and the image data information is output.
[0008] Furthermore, in the computer vision-based vegetation cover measurement method, the image data information is subjected to color space conversion and recognition, and geometric correction processing is performed, specifically as follows: The image data information is converted from RGB to HSV color space, and the color of vegetation and soil is sampled. The HSV threshold range is determined based on the sampled vegetation and soil colors. Based on the HSV threshold range, the image data information is divided into vegetation pixels and non-vegetation pixels, and by selecting the four corner points of the sample plot, the quadrilateral sample plot taken at an angle is corrected into a standard square. The original image coordinates are mapped to standard coordinates using a homography matrix to complete the geometric correction process and form the corrected image data information.
[0009] Furthermore, in the computer vision-based vegetation cover measurement method, vegetation detection is performed on the calibrated image. A preset HSV threshold is used to identify vegetation pixels, and relevant shapes are filtered based on shape features. Specifically: Vegetation detection is performed on the corrected image. A preset HSV threshold is used to determine whether the relevant pixels are greater than the preset HSV threshold. If they are greater, they are considered as vegetation pixels; otherwise, they are considered as non-vegetation pixels. The system counts vegetation pixels and non-vegetation pixels, removes noise and fills holes through opening and closing operations, and filters non-vegetation objects based on shape features.
[0010] Furthermore, in the computer vision-based vegetation cover measurement method, the cover is calculated on the image after vegetation detection, and the image is divided according to the cover, specifically as follows: The corrected image is divided into 800×800 pixel squares, and then the 800×800 pixel squares are further divided into several small grids; By performing pixel statistics on the image data of the several small grids, the proportion of vegetation pixels in each small grid to the total pixels is calculated, and the coverage evaluation index data is preset. The image data of the several small grids are divided based on the coverage evaluation index data to generate coverage membership evaluation data, which is then output as coverage division data.
[0011] Furthermore, in computer vision-based vegetation cover measurement methods, early warnings are issued for target areas based on cover segmentation data, specifically as follows: Set the coverage membership evaluation index data information, and determine whether the coverage division data is greater than the coverage membership evaluation index data information; If the coverage membership evaluation index data is greater than the coverage membership evaluation index data, then an early warning will be issued for the corresponding target area. If the coverage membership evaluation index data is not greater than the coverage membership evaluation index data, then the corresponding target area is regarded as a normal vegetation growth area.
[0012] Furthermore, in the computer vision-based vegetation cover measurement method, relevant vegetation restoration strategies are generated, specifically: Acquire land use, soil, and DEM data in the target area, and generate a spatial distribution layer of regional soil conservation, water conservation, and carbon sequestration ecosystem service value based on the land use, soil, and DEM data in the target area. By spatially overlaying the vegetation cover decline patches with the value spatial distribution layer, a two-dimensional risk matrix of the coverage decline rate and the service value of the location is constructed. Estimate the potential loss of ecosystem service value caused by the continued existence of the current degradation trend for a preset period of time, and set a threshold for the loss of ecosystem service value. When the loss of ecosystem service value caused by the current degradation trend continuing for a preset time is greater than the threshold of the loss of ecosystem service value, the corresponding target area will be designated as the priority vegetation restoration area, and a relevant vegetation restoration strategy will be generated.
[0013] A second aspect of the present invention provides a vegetation coverage measurement system based on computer vision, including a memory and a processor. The memory includes a program for a vegetation coverage measurement method based on computer vision. When the program is executed by the processor, the vegetation coverage measurement method based on computer vision implements the steps of any of the above-described vegetation coverage measurement methods based on computer vision.
[0014] This invention addresses the shortcomings of the prior art and has the following beneficial effects: This invention acquires image data from a target area, performs color space conversion and recognition on the image data, and then performs geometric correction. The corrected image is then used for vegetation detection, identifying vegetation pixels using a preset HSV threshold and filtering related shapes based on shape features. Coverage is calculated on the vegetation-detected image, and the area is divided according to coverage. Finally, warnings are issued for the target area based on the coverage division data, and relevant vegetation restoration strategies are generated. This invention combines shadow detection with spectral compensation depth to specifically correct measurement errors caused by shadows in high-resolution images, significantly improving measurement accuracy under complex lighting conditions. Furthermore, it uses computer vision technology to quickly and objectively calculate vegetation projection cover within a fixed sample plot, and can generate archiveable gridded images and quantitative data for ecological monitoring and statistical analysis. Attached Figure Description
[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained from these drawings without creative effort.
[0016] Figure 1 A flowchart illustrating the overall process of a computer vision-based vegetation cover measurement method is shown. Figure 2 A system block diagram of a computer vision-based vegetation cover measurement system is shown. Detailed Implementation
[0017] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.
[0018] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.
[0019] The first aspect of this invention provides a method for measuring vegetation cover based on computer vision, comprising the following steps: The system acquires image data from the target area, performs color space conversion and recognition on the image data, and then performs geometric correction. Vegetation detection is performed on the corrected image. Vegetation pixels are identified using a preset HSV threshold, and related shapes are filtered based on shape features. The vegetation coverage is calculated from the images after vegetation detection, and the images are then divided according to the coverage. Based on coverage data, early warnings are issued for target areas, and relevant vegetation restoration strategies are generated.
[0020] It should be noted that this invention combines shadow detection with spectral compensation depth to specifically correct measurement errors caused by shadows in high-resolution images, significantly improving measurement accuracy under complex lighting conditions. Furthermore, by using computer vision technology, it can quickly and objectively calculate vegetation projection cover within a fixed sample plot, and generate archiveable gridded images and quantitative data for ecological monitoring and statistical analysis.
[0021] Furthermore, in the computer vision-based vegetation cover measurement method, image data information of the target area is collected, specifically as follows: A drone equipped with a multispectral camera takes low-altitude aerial photos of the area to be tested to acquire high spatial resolution image data, and builds an adaptive threshold shadow detection model based on deep learning. The model inputs are the strong reflection characteristics of shadows in the near-infrared band, the solar elevation angle, the shooting time, and the brightness values of RGB images. The shadow areas in the images are used as the prediction results output. The adaptive threshold shadow detection model is used to segment the shadow areas in the image. For the identified shadow area vegetation, the spectral values of the shadow area pixels are compensated and corrected by using the spectral characteristics of the same type of vegetation under sunlight in the same plot. The compensated and corrected image data is used as the image data information in the target area, and the image data information is output.
[0022] It should be noted that traditional visible vegetation indices (such as ExG) are easily affected by lighting, shadows, and soil background, leading to heterospectral variations in the same vegetation. This invention combines shadow detection with spectral depth compensation to specifically correct measurement errors caused by shadows in high-resolution images, significantly improving measurement accuracy under complex lighting conditions.
[0023] Furthermore, in the computer vision-based vegetation cover measurement method, color space conversion and recognition are performed on the image data, followed by geometric correction processing, specifically: The image data information is converted from RGB to HSV color space, the colors of vegetation and soil are sampled, and the HSV threshold range is determined based on the sampled colors of vegetation and soil. Based on the HSV threshold range, image data information is divided into vegetation pixels and non-vegetation pixels, and by selecting the four corner points of the quadrangular quadrangle, the tilted quadrangle is corrected into a standard square. The original image coordinates are mapped to standard coordinates using a homography matrix to complete the geometric correction process and form the corrected image data information.
[0024] It should be noted that this method can transform the complex problem of vegetation cover measurement into an image processing problem.
[0025] Furthermore, in the computer vision-based vegetation cover measurement method, vegetation detection is performed on the calibrated image. A preset HSV threshold is used to identify vegetation pixels, and relevant shapes are filtered based on shape features. Specifically: Vegetation detection is performed on the corrected image. A preset HSV threshold is used to determine whether the relevant pixels are greater than the preset HSV threshold. If they are greater, they are considered as vegetation pixels; otherwise, they are considered as non-vegetation pixels. The system counts vegetation pixels and non-vegetation pixels, removes noise and fills holes through opening and closing operations, and filters non-vegetation objects based on shape features.
[0026] Furthermore, in the computer vision-based vegetation cover measurement method, the cover is calculated on the image after vegetation detection, and the image is divided according to the cover, specifically as follows: The corrected image is divided into 800×800 pixel squares, and then the 800×800 pixel squares are further divided into several small grids; By performing pixel statistics on the image data of several small grids, the proportion of vegetation pixels in each small grid to the total pixels is calculated, and the coverage evaluation index data is preset. Based on the coverage evaluation index data, the image data of several small grids are divided to generate coverage membership evaluation data, which is then output as coverage partitioning data.
[0027] Furthermore, in computer vision-based vegetation cover measurement methods, early warnings are issued for target areas based on cover segmentation data, specifically as follows: Set up coverage membership evaluation index data information and determine whether the coverage division data is greater than the coverage membership evaluation index data information; If the coverage membership evaluation index data is greater than the coverage membership evaluation index data, an early warning will be issued for the corresponding target area. If the coverage membership evaluation index data is not greater than the coverage membership evaluation index data, then the corresponding target area is considered a normal vegetation growth area.
[0028] It should be noted that the grid division is as follows: the calibrated 800×800 pixel sample plot is divided into 16 small grids of 4×4 grids. Pixel statistics are as follows: the proportion of vegetation pixels in each small grid is counted to the total number of pixels. Grading display is as follows: the coverage is divided into three levels: low (<33%), medium (33-67%), and high (>67%), forming coverage membership evaluation data.
[0029] Furthermore, in the computer vision-based vegetation cover measurement method, relevant vegetation restoration strategies are generated, specifically: Acquire land use, soil, and DEM data in the target area, and generate a spatial distribution layer of regional soil conservation, water conservation, and carbon sequestration ecosystem service value based on the land use, soil, and DEM data in the target area. By spatially overlaying the vegetation cover decline patches with the value spatial distribution layer, a two-dimensional risk matrix of the coverage decline rate and the service value of the location is constructed. Estimate the potential loss of ecosystem service value caused by the continued existence of the current degradation trend for a preset period of time, and set a threshold for the loss of ecosystem service value. When the loss of ecosystem service value caused by the current degradation trend continuing for a preset time exceeds the threshold of ecosystem service value loss, the corresponding target area will be designated as the priority vegetation restoration area, and relevant vegetation restoration strategies will be generated.
[0030] It should be noted that this method can upgrade purely biophysical early warning to ecological function risk early warning, directly linking early warning information to management priorities and benefits. DEM data is a digital terrain model that represents the surface elevation using an ordered numerical array. It belongs to the most basic branch of digital terrain model (DTM) and only expresses the elevation of "bare land" (excluding buildings, trees, etc.). It is the underlying supporting data for geospatial analysis such as earthwork calculation, flood simulation, visibility analysis, and 3D visualization.
[0031] In addition, this method also includes: Use airborne or vehicle-mounted hyperspectral imagers and sun-induced chlorophyll fluorescence (SIF) detectors to simultaneously acquire hyperspectral cube data and SIF intensity data; The red edge position, red edge slope and photosynthetic pigment sensitivity index are calculated from the hyperspectral cube data. The intensity ratio of SIF intensity data at 687nm and 760nm is calculated, and a random forest or support vector regression model is established. Stress data is constructed based on the red edge position, red edge slope, photosynthetic pigment sensitivity index and intensity ratio. The stress data is used as input to output a vegetation stress coefficient between 0 and 1. A preliminary vegetation cover is calculated based on the classification of hyperspectral data. The cover of high-stress areas is then corrected downward using the stress coefficient, while healthy vegetation is maintained or fine-tuned.
[0032] It should be noted that by introducing chlorophyll fluorescence, an optical signal that directly reflects the physiological state of vegetation, and combining it with high-spectral fine features, physiological corrections are made to the vegetation coverage in an unhealthy state, so that the measurement results are closer to the effective green coverage.
[0033] In addition, this method also includes: When conducting remote sensing photography, satellite data with multi-angle observation capabilities or by planning drones to fly and photograph the same area from multiple directions to obtain a set of multi-angle images of the same ground feature; By introducing the Ross-Li model and using reflectance data observed from multiple angles as input, the isotropic parameters, geometric optical parameters, and volume scattering parameters of the Ross-Li model are derived. A quantitative relationship between the geometric optical parameters and the canopy structure and coverage of vegetation is then constructed. A lookup table for geometric optical parameters and vegetation cover is established based on the quantitative relationship between the geometric optical parameters and the canopy structure and coverage of the vegetation. The vegetation coverage under the current optical parameters is estimated by combining the lookup table of geometric optical parameters and coverage with the real-time geometric optical parameters, and the vegetation coverage under the current optical parameters is obtained. The real-time vegetation coverage is then corrected based on the vegetation coverage under the current optical parameters.
[0034] It should be noted that this method eliminates the influence of observation angle on the coverage at standard angles (such as nadir angle), resulting in better spatiotemporal comparability. Furthermore, by utilizing multi-angle observation and physical optics models, it fundamentally eliminates reflectivity variations caused by observation geometry, making vegetation coverage measurement results more objective and stable, and suitable for large-scale, multi-temporal comparative analysis.
[0035] A second aspect of the present invention provides a vegetation cover measurement system based on computer vision, including a memory and a processor. The memory includes a program for a vegetation cover measurement method based on computer vision. When the processor executes the vegetation cover measurement method based on computer vision, it performs the following steps: The system acquires image data from the target area, performs color space conversion and recognition on the image data, and then performs geometric correction. Vegetation detection is performed on the corrected image. Vegetation pixels are identified using a preset HSV threshold, and related shapes are filtered based on shape features. The vegetation coverage is calculated from the images after vegetation detection, and the images are then divided according to the coverage. Based on coverage data, early warnings are issued for target areas, and relevant vegetation restoration strategies are generated.
[0036] It should be noted that this invention combines shadow detection with spectral compensation depth to specifically correct measurement errors caused by shadows in high-resolution images, significantly improving measurement accuracy under complex lighting conditions. Furthermore, by using computer vision technology, it can quickly and objectively calculate vegetation projection cover within a fixed sample plot, and generate archiveable gridded images and quantitative data for ecological monitoring and statistical analysis.
[0037] Furthermore, in a computer vision-based vegetation cover measurement system, image data information is collected from the target area, specifically as follows: A drone equipped with a multispectral camera takes low-altitude aerial photos of the area to be tested to acquire high spatial resolution image data, and builds an adaptive threshold shadow detection model based on deep learning. The model inputs are the strong reflection characteristics of shadows in the near-infrared band, the solar elevation angle, the shooting time, and the brightness values of RGB images. The shadow areas in the images are used as the prediction results output. The adaptive threshold shadow detection model is used to segment the shadow areas in the image. For the identified shadow area vegetation, the spectral values of the shadow area pixels are compensated and corrected by using the spectral characteristics of the same type of vegetation under sunlight in the same plot. The compensated and corrected image data is used as the image data information in the target area, and the image data information is output.
[0038] It should be noted that traditional visible vegetation indices (such as ExG) are easily affected by lighting, shadows, and soil background, leading to heterospectral variations in the same vegetation. This invention combines shadow detection with spectral depth compensation to specifically correct measurement errors caused by shadows in high-resolution images, significantly improving measurement accuracy under complex lighting conditions.
[0039] Furthermore, in a computer vision-based vegetation cover measurement system, color space conversion and recognition are performed on image data, followed by geometric correction. Specifically: The image data information is converted from RGB to HSV color space, the colors of vegetation and soil are sampled, and the HSV threshold range is determined based on the sampled colors of vegetation and soil. Based on the HSV threshold range, image data information is divided into vegetation pixels and non-vegetation pixels, and by selecting the four corner points of the quadrangular quadrangle, the tilted quadrangle is corrected into a standard square. The original image coordinates are mapped to standard coordinates using a homography matrix to complete the geometric correction process and form the corrected image data information.
[0040] It should be noted that this method can transform the complex problem of vegetation cover measurement into an image processing problem.
[0041] Furthermore, in the computer vision-based vegetation cover measurement system, vegetation detection is performed on the calibrated image. A preset HSV threshold is used to identify vegetation pixels, and relevant shapes are filtered based on shape features. Specifically: Vegetation detection is performed on the corrected image. A preset HSV threshold is used to determine whether the relevant pixels are greater than the preset HSV threshold. If they are greater, they are considered as vegetation pixels; otherwise, they are considered as non-vegetation pixels. The system counts vegetation pixels and non-vegetation pixels, removes noise and fills holes through opening and closing operations, and filters non-vegetation objects based on shape features.
[0042] Furthermore, in a computer vision-based vegetation cover measurement system, the cover of the image after vegetation detection is calculated, and the data is divided according to the cover, specifically as follows: The corrected image is divided into 800×800 pixel squares, and then the 800×800 pixel squares are further divided into several small grids; By performing pixel statistics on the image data of several small grids, the proportion of vegetation pixels in each small grid to the total pixels is calculated, and the coverage evaluation index data is preset. Based on the coverage evaluation index data, the image data of several small grids are divided to generate coverage membership evaluation data, which is then output as coverage partitioning data.
[0043] Furthermore, in a computer vision-based vegetation cover measurement system, early warnings are issued for target areas based on cover segmentation data, specifically as follows: Set up coverage membership evaluation index data information and determine whether the coverage division data is greater than the coverage membership evaluation index data information; If the coverage membership evaluation index data is greater than the coverage membership evaluation index data, an early warning will be issued for the corresponding target area. If the coverage membership evaluation index data is not greater than the coverage membership evaluation index data, then the corresponding target area is considered a normal vegetation growth area.
[0044] It should be noted that the grid division is as follows: the calibrated 800×800 pixel sample plot is divided into 16 small grids of 4×4 grids. Pixel statistics are as follows: the proportion of vegetation pixels in each small grid is counted to the total number of pixels. Grading display is as follows: the coverage is divided into three levels: low (<33%), medium (33-67%), and high (>67%), forming coverage membership evaluation data.
[0045] Furthermore, in the computer vision-based vegetation cover measurement system, relevant vegetation restoration strategies are generated, specifically: Acquire land use, soil, and DEM data in the target area, and generate a spatial distribution layer of regional soil conservation, water conservation, and carbon sequestration ecosystem service value based on the land use, soil, and DEM data in the target area. By spatially overlaying and analyzing vegetation cover decline patches with value spatial distribution layers, a two-dimensional risk matrix is constructed that includes the rate of vegetation cover decline and the service value of the location. Estimate the potential loss of ecosystem service value caused by the continued existence of the current degradation trend for a preset period of time, and set a threshold for the loss of ecosystem service value. When the loss of ecosystem service value caused by the current degradation trend continuing for a preset time exceeds the threshold of ecosystem service value loss, the corresponding target area will be designated as the priority vegetation restoration area, and relevant vegetation restoration strategies will be generated.
[0046] It should be noted that this method can upgrade purely biophysical early warning to ecological function risk early warning, making the early warning information directly linked to management priorities and benefits.
[0047] In addition, this system also includes: Use airborne or vehicle-mounted hyperspectral imagers and sun-induced chlorophyll fluorescence (SIF) detectors to simultaneously acquire hyperspectral cube data and SIF intensity data; The red edge position, red edge slope and photosynthetic pigment sensitivity index are calculated from the hyperspectral cube data. The intensity ratio of SIF intensity data at 687nm and 760nm is calculated, and a random forest or support vector regression model is established. Stress data is constructed based on the red edge position, red edge slope, photosynthetic pigment sensitivity index and intensity ratio. The stress data is used as input to output a vegetation stress coefficient between 0 and 1. A preliminary vegetation cover is calculated based on the classification of hyperspectral data. The cover of high-stress areas is then corrected downward using the stress coefficient, while healthy vegetation is maintained or fine-tuned.
[0048] It should be noted that by introducing chlorophyll fluorescence, an optical signal that directly reflects the physiological state of vegetation, and combining it with high-spectral fine features, physiological corrections are made to the vegetation coverage in an unhealthy state, so that the measurement results are closer to the effective green coverage.
[0049] In addition, this system also includes: When conducting remote sensing photography, satellite data with multi-angle observation capabilities or by planning drones to fly and photograph the same area from multiple directions to obtain a set of multi-angle images of the same ground feature; By introducing the Ross-Li model and using reflectance data observed from multiple angles as input, the isotropic parameters, geometric optical parameters, and volume scattering parameters of the Ross-Li model are derived. A quantitative relationship between the geometric optical parameters and the canopy structure and coverage of vegetation is then constructed. A lookup table for geometric optical parameters and vegetation cover is established based on the quantitative relationship between the geometric optical parameters and the canopy structure and coverage of the vegetation. The vegetation coverage under the current optical parameters is estimated by combining the lookup table of geometric optical parameters and coverage with the real-time geometric optical parameters, and the vegetation coverage under the current optical parameters is obtained. The real-time vegetation coverage is then corrected based on the vegetation coverage under the current optical parameters.
[0050] It should be noted that this method eliminates the influence of observation angle on the coverage at standard angles (such as nadir angle), resulting in better spatiotemporal comparability. Furthermore, by utilizing multi-angle observation and physical optics models, it fundamentally eliminates reflectivity variations caused by observation geometry, making vegetation coverage measurement results more objective and stable, and suitable for large-scale, multi-temporal comparative analysis.
[0051] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.
[0052] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.
[0053] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.
[0054] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0055] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.
[0056] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for measuring vegetation cover based on computer vision, characterized in that, Includes the following steps: Image data information in the target area is acquired, and color space conversion and recognition are performed on the image data information, followed by geometric correction processing. Vegetation detection is performed on the corrected image. Vegetation pixels are identified using a preset HSV threshold, and related shapes are filtered based on shape features. The coverage of the images after vegetation detection is calculated, and the images are divided according to the coverage. The system provides early warnings for target areas based on coverage data and generates relevant vegetation restoration strategies.
2. The method for measuring vegetation cover based on computer vision according to claim 1, characterized in that, The image data information of the target area is collected, specifically as follows: A drone equipped with a multispectral camera takes low-altitude aerial photos of the area to be tested to acquire high spatial resolution image data, and builds an adaptive threshold shadow detection model based on deep learning. The model inputs are the strong reflection characteristics of shadows in the near-infrared band, the solar elevation angle, the shooting time, and the brightness values of RGB images. The shadow areas in the images are used as the prediction results output. The adaptive threshold shadow detection model is used to segment the shadow areas in the image. For the identified shadow area vegetation, the spectral values of the shadow area pixels are compensated and corrected by using the spectral characteristics of the same type of vegetation under sunlight in the same plot. The compensated and corrected image data information is used as the image data information in the target area, and the image data information is output.
3. The method for measuring vegetation cover based on computer vision according to claim 1, characterized in that, The image data information is subjected to color space conversion and recognition, followed by geometric correction processing, specifically as follows: The image data information is converted from RGB to HSV color space, and the color of vegetation and soil is sampled. The HSV threshold range is determined based on the sampled vegetation and soil colors. Based on the HSV threshold range, the image data information is divided into vegetation pixels and non-vegetation pixels, and by selecting the four corner points of the sample plot, the quadrilateral sample plot taken at an angle is corrected into a standard square. The original image coordinates are mapped to standard coordinates using a homography matrix to complete the geometric correction process and form the corrected image data information.
4. The method for measuring vegetation cover based on computer vision according to claim 1, characterized in that, Vegetation detection is performed on the corrected image. Vegetation pixels are identified using a preset HSV threshold, and relevant shapes are filtered based on shape features. Specifically: Vegetation detection is performed on the corrected image. A preset HSV threshold is used to determine whether the relevant pixels are greater than the preset HSV threshold. If they are greater, they are considered as vegetation pixels; otherwise, they are considered as non-vegetation pixels. The system counts vegetation pixels and non-vegetation pixels, removes noise and fills holes through opening and closing operations, and filters non-vegetation objects based on shape features.
5. The method for measuring vegetation cover based on computer vision according to claim 1, characterized in that, The vegetation coverage of the images after detection is calculated, and the images are then divided according to the coverage, specifically as follows: The corrected image is divided into 800×800 pixel squares, and then the 800×800 pixel squares are further divided into several small grids; By performing pixel statistics on the image data of the several small grids, the proportion of vegetation pixels in each small grid to the total pixels is calculated, and the coverage evaluation index data is preset. The image data of the several small grids are divided based on the coverage evaluation index data to generate coverage membership evaluation data, which is then output as coverage division data.
6. The method for measuring vegetation cover based on computer vision according to claim 5, characterized in that, The system provides early warnings for target areas based on coverage-based data segmentation, specifically as follows: Set the coverage membership evaluation index data information, and determine whether the coverage division data is greater than the coverage membership evaluation index data information; If the coverage membership evaluation index data is greater than the coverage membership evaluation index data, then an early warning will be issued for the corresponding target area. If the coverage membership evaluation index data is not greater than the coverage membership evaluation index data, then the corresponding target area is regarded as a normal vegetation growth area.
7. The method for measuring vegetation cover based on computer vision according to claim 1, characterized in that, Generate relevant vegetation restoration strategies, specifically: Acquire land use, soil, and DEM data in the target area, and generate a spatial distribution layer of regional soil conservation, water conservation, and carbon sequestration ecosystem service value based on the land use, soil, and DEM data in the target area. By spatially overlaying the vegetation cover decline patches with the value spatial distribution layer, a two-dimensional risk matrix of the coverage decline rate and the service value of the location is constructed. Estimate the potential loss of ecosystem service value caused by the continued existence of the current degradation trend for a preset period of time, and set a threshold for the loss of ecosystem service value. When the loss of ecosystem service value caused by the current degradation trend continuing for a preset time is greater than the threshold of the loss of ecosystem service value, the corresponding target area will be designated as the priority vegetation restoration area, and a relevant vegetation restoration strategy will be generated.
8. A vegetation cover measurement system based on computer vision, characterized in that, The device includes a memory and a processor. The memory includes a program for a computer vision-based vegetation cover measurement method. When the computer vision-based vegetation cover measurement method is executed by the processor, it implements the steps of the computer vision-based vegetation cover measurement method as described in any one of claims 1-7.