Image-based method and apparatus for determining suitable growth area for riparian vegetation

Abstract

Provided are an image-based method and apparatus for determining a suitable growth area for riparian vegetation. The method comprises: collecting multi-source image data of a target riparian zone, and performing fusion processing to acquire high-quality image data; extracting, on the basis of the high-quality image data, feature indices related to vegetation growth, and assessing a vegetation growth state by means of the feature indices; dividing, on the basis of the vegetation growth state, the target riparian zone into a suitable growth area for vegetation and an unsuitable growth area for vegetation, and performing multi-scale segmentation on the suitable growth area for vegetation to acquire multi-level suitable areas; and fusing the multi-level suitable areas with the high-quality image data to output an image of the multi-level suitable growth areas for vegetation in the target riparian zone. The method can realize high-precision mapping and dynamic monitoring of a suitable growth area for riparian vegetation, thereby providing a scientific basis for vegetation protection and ecological restoration.

Description

A method and apparatus for determining suitable vegetation growth zones in riparian zones based on imagery. Technical Field

[0001] This invention relates to the field of coastal vegetation growth environment technology, and in particular to a method and apparatus for determining suitable areas for coastal vegetation growth based on imagery. Background Technology

[0002] Determining suitable areas for riparian vegetation growth using multi-source imagery data presents technical challenges in data fusion and processing. First, imagery data from different sources differ in spatial, spectral, and temporal resolution. Effectively fusing these heterogeneous data to fully leverage their respective advantages is a pressing issue. Second, riparian vegetation growth is influenced by various factors, such as topography, soil, and hydrology. Relying solely on a single spectral index is insufficient to accurately characterize vegetation growth; a more comprehensive and reliable indicator system is needed, taking into account multiple factors. Third, determining reasonable thresholds and classification methods for suitable area delineation requires a complex trade-off between considering the ecological characteristics of vegetation growth and the interpretability and operability of the classification results. Finally, in practical applications, factors such as data acquisition costs, processing efficiency, and the timeliness of results must be considered. Improving efficiency while maintaining accuracy for operational application is also a crucial issue. Therefore, this invention proposes a method and apparatus for determining suitable riparian vegetation growth areas based on imagery. Summary of the Invention

[0003] The purpose of this invention is to address the aforementioned technical problems by providing a method and apparatus for determining suitable areas for riparian vegetation growth based on imagery, enabling high-precision mapping and dynamic monitoring of suitable riparian vegetation growth areas, and providing a scientific basis for vegetation protection and ecological restoration.

[0004] To achieve the above objectives, the present invention provides the following solution:

[0005] A method for determining suitable vegetation growth zones in riparian zones based on imagery, comprising:

[0006] Collect multi-source image data of the target beach area and perform fusion processing to obtain high-quality image data;

[0007] Based on the high-quality image data, vegetation growth-related feature indicators are extracted, and the vegetation growth status is evaluated through the feature indicators.

[0008] Based on the vegetation growth status, the target beach zone is divided into a suitable vegetation growth zone and an unsuitable vegetation growth zone. The suitable vegetation growth zone is then divided into multiple scales to obtain a multi-level suitable zone.

[0009] The multi-level suitable zones are fused with high-quality image data to output a multi-level suitable zone image for vegetation growth in the target beach area.

[0010] To further achieve the above objectives, the present invention also provides an image-based device for determining suitable areas for riparian vegetation growth, comprising:

[0011] Image data acquisition equipment is used to acquire multi-source image data of the target beach zone and perform fusion processing to obtain high-quality image data;

[0012] A vegetation growth assessment device is used to extract vegetation growth-related feature indicators based on the high-quality image data, and to assess the vegetation growth status through the feature indicators.

[0013] The suitable growth zone delineation device is used to divide the target beach zone into a suitable vegetation growth zone and an unsuitable vegetation growth zone according to the vegetation growth status, and to perform multi-scale segmentation of the suitable vegetation growth zone to obtain a multi-level suitable zone.

[0014] A suitable area image generation device is used to fuse the multi-level suitable area with high-quality image data and output a multi-level suitable area image of vegetation growth in the target beach zone.

[0015] The beneficial effects of this invention are as follows:

[0016] This invention first fuses multi-source heterogeneous remote sensing image data using a combination of wavelet transform and principal component analysis to extract vegetation growth-related characteristic indicators. Then, it constructs a multivariate regression model to obtain comprehensive evaluation indicators of vegetation growth status and uses maximum entropy threshold segmentation and fuzzy logic reasoning to delineate suitable growth zones. Furthermore, it employs a support vector machine algorithm for refined classification of suitable zones, introducing a cost-sensitive learning mechanism to balance classification accuracy and cost. Finally, it applies multi-scale segmentation and object-oriented classification methods to extract vegetation patch features, analyze spatial patterns, and realize online publishing and dynamic updating of refined mapping products through a WebGIS platform. This invention achieves high-precision mapping and dynamic monitoring of suitable vegetation growth zones in riparian zones, providing a scientific basis for vegetation protection and ecological restoration. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments 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 drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 is a flowchart of a method for determining suitable areas for riparian vegetation growth based on images according to an embodiment of the present invention.

[0019] Figure 2 is a schematic diagram of an image-based device for determining suitable vegetation growth zones in riparian zones according to an embodiment of the present invention, wherein 1-image data acquisition device, 2-vegetation growth assessment device, 3-suitable growth zone division device, and 4-suitable zone image generation device. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] This embodiment provides a method for determining suitable vegetation growth zones in coastal areas based on images, including: acquiring multi-source image data of the target coastal zone and performing fusion processing to obtain high-quality image data; extracting vegetation growth-related feature indicators based on the high-quality image data and evaluating the vegetation growth status through the feature indicators; dividing the target coastal zone into suitable vegetation growth zones and unsuitable vegetation growth zones according to the vegetation growth status, and performing multi-scale segmentation of the suitable vegetation growth zones to obtain multi-level suitable zones; fusing the multi-level suitable zones with the high-quality image data to output a multi-level suitable vegetation growth zone image of the target coastal zone.

[0023] Specifically, this embodiment first fuses multi-source heterogeneous remote sensing image data by combining wavelet transform and principal component analysis to extract vegetation growth-related feature indicators; then, it constructs a multivariate regression model to obtain a comprehensive evaluation index of vegetation growth status, and uses maximum entropy threshold segmentation and fuzzy logic reasoning to divide suitable growth areas; further, it uses the support vector machine algorithm to perform refined classification of suitable areas, and introduces a cost-sensitive learning mechanism to balance classification accuracy and cost; finally, it applies multi-scale segmentation and object-oriented classification methods to extract vegetation patch features, analyze spatial patterns, and realize the online publication and dynamic updating of refined mapping products through the WebGIS platform.

[0024] The following describes in detail, with reference to Figure 1, a method for determining suitable vegetation growth zones in riparian zones based on images, as proposed in this embodiment. The method includes the following steps:

[0025] Step S101: Acquire multi-source image data of the target beach zone and perform fusion processing to obtain high-quality image data, including:

[0026] Multi-source image data of the target beach zone was acquired, including image data with different spatial, spectral, and temporal resolutions. The multi-source image data was scaled and normalized, and wavelet transform was used to decompose the normalized multi-source image data into multi-scale features to obtain features at each scale. Principal component analysis was used to reduce the dimensionality of the features at each scale, and then inverse wavelet transform was performed on the dimensionality-reduced features at each scale to obtain high-quality reconstructed image data.

[0027] Specifically, in this embodiment, the multi-source image data comes from different sensors and has different spatial, spectral, and temporal resolutions. These data are fused to obtain a high-resolution image data that is clear, rich in color information, and has a relatively fast time update—that is, high-quality image data. First, scale normalization is performed. Common methods include interpolation, such as bilinear interpolation and cubic convolution interpolation, or scale matching by constructing a point spread function. After normalization, the image data are at the same scale, equivalent to unifying measuring cups of different diameters to a standard, facilitating subsequent mixing processing. Next, wavelet transform is used to perform multi-scale decomposition on the normalized image. Specifically, the image can be decomposed into multiple layers, each layer yielding a low-frequency component and several high-frequency components. The first layer of decomposition yields a low-frequency component (approximate component) and three high-frequency components (horizontal, vertical, and diagonal detail components), representing the detail information of the image in different directions. Then, the low-frequency component is further decomposed to obtain the second layer of decomposition results, and so on, resulting in a multi-scale image representation containing various information from the overall picture to the details. After obtaining the wavelet coefficients, since wavelet decomposition generates a large amount of data, principal component analysis (PCA) is needed for dimensionality reduction. This not only reduces the amount of data but also highlights the main features and removes the influence of noise. Then, inverse wavelet transform is performed on the features processed by PCA to reconstruct high-resolution image data. Inverse wavelet transform is essentially the reverse process of wavelet transform, converting the dimensionality-reduced principal components back from the wavelet domain to the spatial domain, resulting in the fused high-resolution image. Finally, the quality of the reconstructed image data is evaluated to determine if the fusion result meets the requirements. This can be done by calculating indicators such as root mean square error (RMSE), peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) between the fused image and the original high-resolution and multispectral images. If these indicators reach the preset quality threshold, the fusion result is considered satisfactory, and high-resolution fused image data is output. Otherwise, the wavelet transform parameters (such as wavelet basis, number of decomposition levels), the number of principal components retained by PCA, etc., need to be readjusted, and the fusion process is repeated until the fusion result meets the quality requirements.

[0028] For example, principal component analysis yields a feature matrix containing 10 principal components, each representing the main information of the original image data in different directions. These principal component data are then input into an inverse wavelet transform algorithm, using wavelet basis functions (such as the Daubechies wavelet) for inverse transform to recover wavelet coefficients at various scales. Using the wavelet coefficients obtained from the inverse wavelet transform, a pyramid reconstruction algorithm is employed to progressively superimpose high-frequency and low-frequency wavelet coefficients to reconstruct the complete image data. A weighted average fusion algorithm is then used to weight the reconstructed image data from different sources, with the weights determined based on the signal-to-noise ratio and sharpness of each image data point.

[0029] Step S102: Extract vegetation growth-related feature indicators based on high-quality image data, and evaluate the vegetation growth status through these feature indicators, including:

[0030] Based on high-quality image data, normalized vegetation index, enhanced vegetation index, leaf area index, chlorophyll content, topographic factors, soil factors, and hydrological factors are extracted. The characteristic indicators are substituted into a pre-set multiple linear regression model to predict the distribution of vegetation growth status in the target beach zone and to evaluate the overall vegetation growth status. The independent variables of the multiple linear regression model are normalized vegetation index, enhanced vegetation index, leaf area index, chlorophyll content, topographic factors, soil factors, and hydrological factors, and the dependent variables are vegetation biomass and yield.

[0031] Specifically, in this embodiment, the Normalized Difference Vegetation Index (NDVI) and Enhanced Vegetation Index (EDI) are calculated based on high-quality image data; the Leaf Area Index (LAI) and chlorophyll content are retrieved from the high-quality image data using a radiative transfer model combined with ground measurement data; Digital Elevation Model (DEM) data of the target shoreline is acquired, and topographic factors such as slope, aspect, and elevation are extracted from the DEM data; Soil texture, organic matter content, and nutrient content are extracted based on soil type maps and soil property data; and the water stress index is extracted from the target shoreline using a hydrological model based on the high-quality image data.

[0032] The Normalized Difference Vegetation Index (NDVI) is calculated using the reflectance of red and near-infrared wavelengths, using the following formula:

[0033] NIR stands for near-infrared reflectance, typically ranging from 0.75 to 1.30 micrometers; RED stands for red reflectance, typically ranging from 0.60 to 0.75 micrometers.

[0034] NDVI reflects vegetation cover and health status. The NDVI value usually ranges from -1 to +1. Specifically: NDVI≈1 indicates healthy, lush green vegetation with high vegetation cover and vigorous photosynthesis; NDVI≈0 indicates bare soil, desert, or non-vegetated areas lacking greenery; NDVI≈-1 indicates water bodies, clouds, snow, or other non-vegetated areas.

[0035] The Enhanced Vegetation Index (EVI) further considers atmospheric effects, and its formula is as follows:

[0036] Wherein, SR is the surface reflectivity, RED is the reflectivity of the red light band, BLUE is the reflectivity of the blue light band, G is the gain factor, which is usually taken as 2.5; C1 and C2 are constants for the red and blue light bands, respectively, which are usually taken as 6 and 7.5; L is an offset value of the ground reflectivity, which is usually taken as 10000.

[0037] EVI is applicable to areas with high vegetation cover. The value of EVI is usually between -1 and +1. However, because EVI is more sensitive to high vegetation cover, it can better reflect the growth status of high-density vegetation than NDVI. Specifically: EVI≈1 indicates healthy and lush vegetation; EVI≈0 indicates bare soil, desert, or areas without vegetation; EVI≈-1 indicates water bodies or non-vegetated areas, especially under arid or extreme conditions.

[0038] Using the PROSAIL model, by inputting surface reflectance, soil type, vegetation structure parameters, etc., the leaf area index (LAI) and chlorophyll content can be retrieved through iterative optimization.

[0039] Using ASTERGDEM data, GIS software was used to calculate topographic factors such as slope, aspect, and elevation, which affect vegetation distribution and growth. Areas with steeper slopes experience faster water loss, limiting vegetation growth; aspect affects sunlight conditions, with vegetation typically growing better on sunny slopes than on shady slopes.

[0040] Soil survey data and soil type maps were used to identify soil types such as loam and sandy soil in the study area, and data on their organic matter and nutrient content were obtained. Soil texture affects the ability to retain water and nutrients, while organic matter and nutrient content are directly related to the nutrient supply to vegetation.

[0041] Using the SWAT hydrological model, input data such as rainfall, soil, and topography, simulate hydrological processes such as surface runoff and soil moisture, calculate the water stress index, and reflect the degree to which vegetation is limited by water.

[0042] Specifically, in this embodiment, a multiple linear regression model is used to spatially predict the vegetation growth status of the target beach zone and obtain the distribution of vegetation growth status. The multiple linear regression model uses normalized vegetation index, enhanced vegetation index, leaf area index, chlorophyll content, topographic factors, soil factors and hydrological factors as independent variables, and vegetation biomass and yield as dependent variables.

[0043] First, the independent variable data, including NDVI, EVI, LAI, chlorophyll content, topographic factors, soil factors, and hydrological factors, are preprocessed, including data cleaning, missing value handling, and outlier handling, to ensure data quality. For example, missing NDVI data can be imputed using interpolation methods (such as linear interpolation or Kriging interpolation); outliers can be identified and removed through box plot analysis.

[0044] Preprocessing of the dependent variable data, vegetation biomass and yield, includes data standardization and normalization to eliminate the influence of dimensions. For example, the Z-score standardization method is used to convert vegetation biomass and yield data into standard scores with a mean of 0 and a standard deviation of 1. The Z-score standardization formula is: Z = (X - μ) / σ, where X is the original data, μ is the mean, and σ is the standard deviation.

[0045] The preprocessed independent and dependent variable data are randomly divided into training and test sets. The training set is used for model training, and the test set is used for model evaluation. Specifically, in this embodiment, the data is divided into a 70% training set and a 30% test set to ensure that the data distribution of the training and test sets is consistent.

[0046] A multiple linear regression algorithm is employed, using the independent variable data from the training set as input and the dependent variable data as output to train a multiple linear regression model and obtain model parameters. This embodiment utilizes the least squares method to solve for the regression coefficients, establishing a linear relationship model between vegetation biomass or yield and NDVI, EVI, LAI, chlorophyll content, topographic factors, soil factors, and hydrological factors. The independent variable data from the test set are input into the trained multiple linear regression model to predict the corresponding vegetation biomass and yield, which are then compared with the actual values ​​in the test set to calculate the model's evaluation indicators, including mean squared error (MSE) and coefficient of determination (R²). 2 (e.g., to evaluate model performance.)

[0047] The specific calculations are as follows:

[0048] in, Let y be the predicted value of the i-th sample. i Let n be the true value of the i-th sample, and n be the number of samples.

[0049] Where SSE is the residual sum of squares and SST is the total sum of squares. R is the mean of the true values. 2 The closer it is to 1, the stronger the model's explanatory power.

[0050] Specifically, in this embodiment, a comprehensive evaluation index of vegetation growth status is calculated based on the distribution of vegetation growth status to quantitatively assess the overall status of vegetation growth in the study area.

[0051] By using methods such as weighted averaging, a comprehensive score for vegetation growth is derived by integrating various indicators, providing a scientific basis for ecological protection and agricultural production. This step allows for a systematic assessment of vegetation growth in the study area, providing strong support for ecological environment monitoring and agricultural management.

[0052] Step S103: Divide the target shoreline into suitable and unsuitable vegetation growth zones based on vegetation growth status, and further segment the suitable vegetation growth zones at multiple scales to obtain multi-level suitable zones, including:

[0053] By setting a suitable area division threshold, the distribution of vegetation growth status in the target beach zone is judged, and a preliminary division of suitable and unsuitable vegetation growth areas is made. The suitable area division threshold is determined based on the overall vegetation growth status. A fuzzy logic reasoning mechanism is used to perform fuzzy classification on the preliminary vegetation growth suitable areas, and then the fuzzy classification results are converted into the final vegetation growth suitable area division results through defuzzification.

[0054] Environmental characteristic data of suitable vegetation growth areas are input into a pre-set classification model to obtain the suitability classification results of the suitable vegetation growth areas. The classification model is obtained by training a support vector machine. During the classification process, the classification model introduces a cost-sensitive learning mechanism to balance the data acquisition cost and classification accuracy, and obtains the suitability classification results, which include highly suitable areas, moderately suitable areas, and lowly suitable areas. The highly suitable areas, moderately suitable areas, and lowly suitable areas are segmented at multiple scales to obtain the vegetation patch distribution in each suitable area, generating multi-level suitability area classification results.

[0055] Specifically, this embodiment employs a maximum entropy threshold segmentation algorithm to adaptively determine the threshold for dividing suitable vegetation growth zones in the target riparian zone. Based on the vegetation growth suitability assessment results of the target riparian zone, a fuzzy rule base is established through a fuzzy logic reasoning mechanism to optimize the division results of suitable vegetation growth zones. A fuzzy membership function is used to map various growth suitability indicators of the target riparian zone vegetation to the [0, 1] interval, serving as the input variable for fuzzy logic reasoning. According to the established fuzzy rule base, fuzzy reasoning methods, such as Mamdani reasoning or Sugeno reasoning, are used to perform logical operations on the input fuzzy membership degrees, obtaining the fuzzy classification results of suitable vegetation growth zones in the target riparian zone. Through defuzzification, the fuzzy classification results are converted into determined suitable vegetation growth zone division results for the target riparian zone, resulting in a binary distribution map of suitable and unsuitable zones. The suitable zone division results obtained from maximum entropy threshold segmentation are compared with the division results optimized by fuzzy logic reasoning to evaluate the optimization effect. If necessary, the fuzzy rule base and membership function are adjusted to improve the interpretability and operability of the classification.

[0056] Based on environmental characteristic data of suitable growth areas, a support vector machine (SVM) algorithm is used to train a classification model. Different classification error cost weight parameters are set for highly suitable, moderately suitable, and poorly suitable areas. Environmental characteristic data is input into the trained classification model, and the model predicts the suitability classification results for each area. The classification results are post-processed, and the classification error cost weights are adjusted according to a cost-sensitive learning mechanism to balance data acquisition costs and classification accuracy. Based on the balanced classification results, the final suitability of each area is determined, generating the division results of highly suitable, moderately suitable, and poorly suitable areas.

[0057] A multi-scale segmentation method is employed to assess the suitability of the classification results. By setting different scale parameters, similar pixels or objects are gradually merged from bottom to top to obtain segmentation results at different levels. For each segmentation level, spectral, shape, and texture features of the segmented objects are extracted to construct feature vectors. Based on a pre-established expert knowledge rule base, rule-based reasoning is used to initially classify the segmented objects, resulting in preliminary vegetation patches. A support vector machine algorithm is then used to further classify these preliminary vegetation patches to obtain final vegetation patches. For each vegetation patch, its geometric and topological features are extracted, including patch area, perimeter, shape index, and fractal dimension. The spatial relationships between adjacent vegetation patches are analyzed, and the distance matrix and connectivity index between patches are calculated to construct a spatial topological network of patches. Based on the spatial distribution characteristics and topological network structure of the vegetation patches, spatial statistical methods are used to analyze the spatial pattern and ecological processes of vegetation growth, revealing the interaction and succession patterns between different patches.

[0058] Step S104: Fuse the multi-level suitability zone with high-quality image data to output a multi-level suitability zone image of vegetation growth in the target beach zone, including:

[0059] Multi-level suitable zones are fused with high-quality image data through spatial registration; based on the fused data, mapping software is used to create multi-level suitable zone images of vegetation growth in the target beach zone.

[0060] Specifically, this embodiment spatially registers multi-level suitable areas with high-quality image data, and achieves accurate fusion and mapping through methods such as pixel resampling. The fused mapping product undergoes quality control processes such as topology checking and attribute editing to ensure its accuracy and aesthetics. Spatial database technology is used to organize and manage the mapping product, constructing the geographic database required by the WebGIS platform. Map services for the mapping product are published on the WebGIS platform, and a user-friendly front-end interface is designed to enable online browsing and querying of the mapping product. A back-end data processing and update mechanism is established. When the vegetation of the target riparian zone changes, the latest remote sensing imagery is periodically acquired and steps S101-S104 are repeated to achieve dynamic updates of the mapping product.

[0061] To further optimize the technical solution, this embodiment also provides an image-based device for determining suitable areas for riparian vegetation growth, as shown in Figure 2, comprising:

[0062] Image data acquisition device 1 is used to acquire multi-source image data of the target beach zone and perform fusion processing to obtain high-quality image data;

[0063] Vegetation growth assessment device 2 is used to extract vegetation growth-related feature indicators based on high-quality image data and to assess vegetation growth status through feature indicators.

[0064] The suitable growth zone division device 3 is used to divide the target beach zone into a suitable vegetation growth zone and an unsuitable vegetation growth zone according to the vegetation growth status, and to perform multi-scale segmentation of the suitable vegetation growth zone to obtain a multi-level suitable zone.

[0065] Suitable area image generation device 4 is used to fuse multi-level suitable areas with high-quality image data and output multi-level suitable area images of vegetation growth in the target beach zone.

[0066] Furthermore, the image data acquisition device 1 includes:

[0067] Image data acquisition component, used to acquire multi-source image data of the target beach zone, including image data with different spatial resolution, spectral resolution and temporal resolution;

[0068] The first image data processing component is used to normalize the scale of multi-source image data and use wavelet transform to decompose the normalized multi-source image data into multiple scales to obtain features at each scale.

[0069] The second image data acquisition component is used to perform dimensionality reduction on features at each scale using principal component analysis, and then perform inverse wavelet transform on the dimensionality-reduced features at each scale to obtain high-quality reconstructed image data.

[0070] Furthermore, the vegetation growth assessment device 2 includes:

[0071] Feature index extraction component is used to extract normalized vegetation index, enhanced vegetation index, leaf area index, chlorophyll content, topographic factors, soil factors and hydrological factors based on high-quality image data;

[0072] The vegetation growth assessment component is used to input characteristic indicators into a preset multiple linear regression model to predict the distribution of vegetation growth status in the target beach zone and assess the overall vegetation growth status. The independent variables of the multiple linear regression model are normalized vegetation index, enhanced vegetation index, leaf area index, chlorophyll content, topographic factors, soil factors and hydrological factors, and the dependent variables are vegetation biomass and yield.

[0073] Furthermore, the suitable growth zone delineation device 3 includes:

[0074] The first suitable area division component is used to judge the distribution of vegetation growth status of the target beach zone by means of a preset suitable area division threshold, and to initially divide the suitable vegetation growth zone and the unsuitable vegetation growth zone. The suitable area division threshold is determined according to the overall vegetation growth status.

[0075] The second suitable area division component is used to perform fuzzy classification of the initially divided suitable vegetation growth areas using a fuzzy logic reasoning mechanism, and then convert the fuzzy classification results into the final suitable vegetation growth area division results through defuzzification.

[0076] The first suitable area segmentation component is used to input environmental feature data of suitable vegetation growth areas into a preset classification model to obtain the suitability classification results of suitable vegetation growth areas. The classification model is obtained by training a support vector machine. During the classification process, the classification model introduces a cost-sensitive learning mechanism to balance the data acquisition cost and classification accuracy to obtain the suitability classification results. The suitability classification results include highly suitable areas, moderately suitable areas, and low suitable areas.

[0077] The second suitability zone segmentation component is used to perform multi-scale segmentation of highly suitable zones, moderately suitable zones, and lowly suitable zones, obtain the vegetation patch distribution in each suitable zone, and generate multi-level suitability zone classification results.

[0078] Furthermore, the suitable area image generation device 4 includes:

[0079] Fusion components are used to fuse multi-level suitable areas with high-quality image data through spatial registration;

[0080] A mapping component is used to create multi-level suitable areas for vegetation growth in a target beach zone using mapping software based on the fused data.

[0081] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for determining suitable vegetation growth zones in riparian zones based on imagery, characterized in that, include: Collect multi-source image data of the target beach area and perform fusion processing to obtain high-quality image data; Based on the high-quality image data, vegetation growth-related feature indicators are extracted, and the vegetation growth status is evaluated through the feature indicators. Based on the vegetation growth status, the target beach zone is divided into a suitable vegetation growth zone and an unsuitable vegetation growth zone. The suitable vegetation growth zone is then divided into multiple scales to obtain a multi-level suitable zone. The multi-level suitable zones are fused with high-quality image data to output a multi-level suitable zone image for vegetation growth in the target beach area.

2. The method for determining suitable areas for riparian vegetation growth based on imagery according to claim 1, characterized in that, Acquiring multi-source image data of the target beach area and performing fusion processing to obtain the high-quality image data includes: Collect multi-source image data of the target beach area; The multi-source image data is scale-normalized, and wavelet transform is used to decompose the normalized multi-source image data into multiple scales to obtain features at each scale. Principal component analysis was used to reduce the dimensionality of the features at each scale, and then inverse wavelet transform was performed on the reduced features to obtain high-quality reconstructed image data.

3. The method for determining suitable areas for riparian vegetation growth based on imagery according to claim 1, characterized in that, Assessing vegetation growth status using the aforementioned characteristic indicators includes: The characteristic indicators are substituted into a preset multiple linear regression model to predict the distribution of vegetation growth status in the target beach zone and to evaluate the overall vegetation growth status. The independent variables of the multiple linear regression model are normalized vegetation index, enhanced vegetation index, leaf area index, chlorophyll content, topographic factors, soil factors and hydrological factors, and the dependent variables are vegetation biomass and yield.

4. The method for determining suitable areas for riparian vegetation growth based on imagery according to claim 3, characterized in that, Based on the vegetation growth status, the target shoreline zone is divided into a suitable vegetation growth zone and a non-suitable vegetation growth zone, including: By using a preset suitable area division threshold, the distribution of vegetation growth status in the target beach zone is judged, and a preliminary division of suitable vegetation growth areas and unsuitable vegetation growth areas is made. The suitable area division threshold is determined based on the overall vegetation growth status. A fuzzy logic reasoning mechanism is used to perform fuzzy classification on the initially delineated suitable vegetation growth areas, and then the fuzzy classification results are converted into the final suitable vegetation growth area delineation results through defuzzification.

5. The method for determining suitable vegetation growth zones in riparian zones based on imagery according to claim 1, characterized in that, The suitable vegetation growth zone is segmented at multiple scales to obtain multi-level suitable zones, including: The environmental characteristic data of the suitable vegetation growth area are input into a preset classification model to obtain the suitability classification result of the suitable vegetation growth area. The highly suitable area, moderately suitable area and lowly suitable area are segmented at multiple scales to obtain the distribution of vegetation patches in each suitable area and generate multi-level suitable area classification results.

6. A device for determining suitable areas for riparian vegetation growth based on imagery, characterized in that, include: Image data acquisition equipment is used to acquire multi-source image data of the target beach zone and perform fusion processing to obtain high-quality image data; A vegetation growth assessment device is used to extract vegetation growth-related feature indicators based on the high-quality image data, and to assess the vegetation growth status through the feature indicators. The suitable growth zone delineation device is used to divide the target beach zone into a suitable vegetation growth zone and an unsuitable vegetation growth zone according to the vegetation growth status, and to perform multi-scale segmentation of the suitable vegetation growth zone to obtain a multi-level suitable zone. A suitable area image generation device is used to fuse the multi-level suitable area with high-quality image data and output a multi-level suitable area image of vegetation growth in the target beach zone.

7. The image-based device for determining suitable vegetation growth zones in riparian zones according to claim 6, characterized in that, The image data acquisition device includes: The image data acquisition component is used to acquire multi-source image data of the target beach zone, including image data with different spatial resolutions, spectral resolutions, and temporal resolutions; The first image data processing component is used to perform scale normalization on the multi-source image data and to perform multi-scale decomposition on the normalized multi-source image data using wavelet transform to obtain features at each scale. The second image data acquisition component is used to perform dimensionality reduction processing on the features at each scale using principal component analysis, and then perform inverse wavelet transform on the dimensionality-reduced features at each scale to obtain high-quality reconstructed image data.

8. The image-based device for determining suitable vegetation growth zones in riparian zones according to claim 6, characterized in that, The vegetation growth assessment equipment includes: The feature index extraction component is used to extract normalized vegetation index, enhanced vegetation index, leaf area index, chlorophyll content, topographic factors, soil factors and hydrological factors based on the high-quality image data. The vegetation growth assessment component is used to input the characteristic indicators into a preset multiple linear regression model to predict the distribution of vegetation growth status in the target beach zone and assess the overall vegetation growth status. The independent variables of the multiple linear regression model are normalized vegetation index, enhanced vegetation index, leaf area index, chlorophyll content, topographic factors, soil factors and hydrological factors, and the dependent variables are vegetation biomass and yield.

9. The image-based device for determining suitable vegetation growth zones in riparian zones according to claim 8, characterized in that, The suitable growth zone delineation device includes: The first suitable area division component is used to judge the distribution of vegetation growth status of the target beach zone by means of a preset suitable area division threshold, and to initially divide the suitable vegetation growth zone and the unsuitable vegetation growth zone, wherein the suitable area division threshold is determined according to the overall vegetation growth status. The second suitable area division component is used to perform fuzzy classification of the initially divided suitable vegetation growth areas using a fuzzy logic reasoning mechanism, and then convert the fuzzy classification results into the final suitable vegetation growth area division results through defuzzification. The first suitable area segmentation component is used to input the environmental characteristic data of the suitable vegetation growth area into a preset classification model to obtain the suitability classification result of the suitable vegetation growth area. The second suitable area segmentation component is used to perform multi-scale segmentation on the highly suitable area, the moderately suitable area and the lowly suitable area respectively, obtain the vegetation patch distribution in each suitable area, and generate multi-level suitable area classification results.

10. The image-based device for determining suitable vegetation growth zones in riparian zones according to claim 6, characterized in that, The suitable area image generation device includes: A fusion component is used to fuse the multi-level suitable area with high-quality image data through spatial registration; A mapping component is used to create a multi-level suitable area image of vegetation growth in the target beach zone based on the fused data using mapping software.

Images