River bank vegetation ecological health dynamic evaluation system and method

Abstract

The invention discloses a river bank vegetation ecological health dynamic evaluation system and method, and relates to the technical field of river bank vegetation ecological health, and the method specifically comprises the following steps: judging whether each image unit grid has a condition that a shrub color is close to a background color according to color features and texture features; marking the image unit grid with the condition as a to-be-identified area; boundary identification feature information of each to-be-identified area is extracted, analysis is carried out after extraction, the boundary identification fuzzy degree of each to-be-identified area under the condition that the shrub color is close to the background color is evaluated, and classification is carried out according to an evaluation result; and according to a classification result, selecting corresponding shrub vegetation spatial distribution area extraction modes for different types of to-be-identified areas so as to extract shrub spatial distribution areas in the to-be-identified areas. According to the method, the problem that the shrub boundary is fuzzy and difficult to identify is solved, and accurate extraction of shrub space distribution and accurate evaluation of the ecological state under the complex background are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of riverbank vegetation ecological health, and in particular to a riverbank vegetation ecological health dynamic assessment system and method. Background Art

[0002] The ecological health of riparian vegetation refers to the overall ecological status of riparian vegetation communities, as reflected in their structural integrity, functional stability, and ecological service capacity. It encompasses not only static indicators such as species composition, biodiversity, and cover, but also their responsiveness to environmental disturbances, self-repair capacity, and long-term support for ecological functions such as soil and water conservation, microclimate regulation, and habitat maintenance. Because riparian areas are sensitive zones where land and water meet, the ecological status of their vegetation has a direct impact on flood control, pollution control, water quality purification, and regional ecological security. However, under the combined influence of natural changes (such as climate change and hydrological fluctuations) and human disturbances (such as engineering construction and agricultural development), the ecological health of riparian vegetation is highly dynamic and vulnerable. Therefore, static monitoring at a single point in time cannot fully reflect its ecological evolution and potential risks. Therefore, a multi-temporal, multi-dimensional, and quantifiable dynamic assessment method is urgently needed. Through continuous data collection and indicator analysis, it is possible to accurately understand the health trends and fluctuations of vegetation systems, thereby providing scientific and forward-looking decision-making for ecological protection and restoration management.

[0003] Existing dynamic assessment technologies for riparian vegetation ecological health typically rely on multi-source data acquisition methods, such as remote sensing imagery, drone aerial photography, or ground-based sensors. These methods regularly or continuously collect vegetation information from riparian areas. These technologies then quantify indicators such as vegetation cover, greenness index (such as NDVI), species diversity, and biomass changes through image recognition, spectral analysis, and Geographic Information System (GIS) technology. After data acquisition, the assessment system typically includes four key steps: data preprocessing (such as image denoising, registration, and rectification), indicator extraction (selecting appropriate ecological health indicators based on ecological models), time series analysis (comparing data trends at different time points), and comprehensive evaluation. This comprehensive evaluation often uses established health classification criteria, combined with expert knowledge or a multi-indicator weighting method, to score or grade the current and historical ecological status. Furthermore, some advanced technologies incorporate machine learning algorithms, training models to identify unusual changes or predict future ecological trends, thereby enhancing the automation and intelligence of the assessment.

[0004] The existing technology has the following deficiencies:

[0005] In the complex terrain of riverbank footslopes, shrubs are often interspersed with herbs, trees, and other vegetation types. Shrubs typically have low canopies, dense branches and foliage, and a narrow color range. Especially near exposed sand or rocky backgrounds, shrubs appear very close to the surrounding background in images, resulting in unclear edge contours. Due to the lack of significant color transitions and structural contrast in the images, shrub boundaries are difficult to accurately segment. Current dynamic assessment techniques for riparian vegetation ecological health typically rely on edge detection and color segmentation methods in image analysis. These methods rely heavily on edge sharpness and color difference, making them inadequate for images with slow transitions between shrub boundaries and backgrounds. Consequently, they cannot accurately extract the spatial distribution of shrubs based on the degree of boundary ambiguity when shrub colors are close to the background. This bias in identification can lead to persistent underestimation of shrub cover, inaccurate extraction of subsequent structural integrity indicators, and ultimately misrepresent the ecological health of the area. This can ultimately lead to unnecessary or even destructive interventions in areas that should otherwise be stable, severely undermining the scientific and precise nature of ecological governance.

[0006] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0007] The purpose of the present invention is to provide a system and method for dynamically evaluating the ecological health of riverbank vegetation to solve the problems in the above-mentioned background technology.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a method for dynamically assessing the ecological health of riverbank vegetation, comprising the following steps:

[0009] Acquire image data containing the riverbank area, perform preprocessing on the image data, including image denoising, registration, and multi-channel alignment, extract the riverbank area image in combination with terrain information, and divide the riverbank area image into several image unit grids;

[0010] Judging whether the shrub color is close to the background color in each image unit grid based on the color feature and texture feature, and marking the image unit grid with such a situation as the area to be identified;

[0011] Extract the boundary recognition feature information of each area to be identified, analyze it after extraction, evaluate the degree of boundary recognition fuzziness of each area to be identified when the shrub color is close to the background color, and classify it according to the evaluation results;

[0012] According to the classification results, the corresponding shrub vegetation spatial distribution area extraction method is selected for the different categories of areas to be identified, so as to extract the shrub spatial distribution area in each area to be identified;

[0013] The image range statistics of the extracted shrub spatial distribution area are performed, and the output display of the ecological health status of the river bank area is completed based on the statistical results.

[0014] Preferably, judging whether the color of the shrubs in each image unit grid is close to the background color based on the color characteristics and texture characteristics, and marking the image unit grid with such a situation as a to-be-identified area, includes:

[0015] Analyze the color distribution differences between shrub vegetation areas and adjacent non-vegetation areas in each image unit grid;

[0016] Analyze the distribution differences of texture arrangement direction and density between shrub vegetation areas and adjacent non-vegetation areas in each image unit grid;

[0017] When the color distribution difference between the shrub vegetation area and the adjacent non-vegetation area is lower than the color judgment standard set based on the pixel color value cluster difference threshold, and the distribution difference of the texture arrangement direction and density between the shrub vegetation area and the adjacent non-vegetation area is lower than the texture judgment standard set based on the texture direction histogram overlap ratio, it is determined that the shrub color is close to the background color in the image unit grid;

[0018] The image unit grids where the shrub color is close to the background color are marked as the area to be identified.

[0019] Preferably, the boundary recognition feature information of each area to be identified is extracted and analyzed after extraction to evaluate the degree of boundary recognition fuzziness of each area to be identified when the shrub color is close to the background color, and classification is performed based on the evaluation results, specifically:

[0020] Extracting boundary recognition feature information of each area to be identified and performing preprocessing operations after extraction;

[0021] Extract boundary structure contrast information and image texture and color fusion information from the preprocessed boundary recognition feature information, and perform normalization processing after extraction;

[0022] Based on the normalized boundary structure contrast information and image texture and color fusion information, the boundary transition perception coefficient and boundary recognition obstacle index of each area to be identified are generated respectively;

[0023] Generate a boundary recognition fuzzy index for each area to be recognized based on the generated boundary transition perception coefficient and boundary recognition obstruction index of each area to be recognized;

[0024] Determine the pre-set boundary recognition fuzzy index threshold range, and compare it with the generated boundary recognition fuzzy index of each area to be identified after determination. According to the comparison results, evaluate the boundary recognition fuzziness of each area to be identified when the shrub color is close to the background color, and classify it according to the evaluation results.

[0025] Preferably, the logic for obtaining the boundary transition perception coefficient of each area to be identified is as follows:

[0026] The boundary structure contrast information is extracted from the preprocessed boundary recognition feature information, specifically including the average grayscale gradient amplitude, grayscale change transition length and gradient direction standard deviation of the edge of each area to be identified. These three types of data are divided by their corresponding preset maximum values ​​to obtain the normalized average grayscale gradient amplitude, grayscale change transition length and gradient direction standard deviation of the edge of each area to be identified, and are calibrated as 、 and , Represents the normalized The average gray gradient amplitude of the edge of the area to be identified, Represents the normalized The grayscale change transition length of the edge of the area to be identified, Represents the normalized The standard deviation of the gradient direction of the edge of the area to be identified, , is a positive integer;

[0027] Calculate the boundary transition perception coefficient of each area to be identified The specific calculation method is as follows: the normalized grayscale change transition length of each edge of the area to be identified is The exponential power is divided by the normalized average gray gradient amplitude of the edge of each area to be identified The natural logarithm of the ratio obtained by adding one and the normalized standard deviation of the gradient direction of the edge of each area to be identified Take the square root and add them to get the boundary transition perception coefficient of each area to be identified .

[0028] Preferably, the logic for obtaining the boundary recognition obstacle index of each area to be identified is as follows:

[0029] The image texture and color fusion information is extracted from the preprocessed boundary recognition feature information, which specifically includes three types of data: the color mean Euclidean distance of each area to be identified, the texture direction histogram cross ratio, and the pixel color variance ratio. These three types of data are divided by their corresponding preset maximum values ​​to obtain the normalized color mean Euclidean distance, texture direction histogram cross ratio, and pixel color variance ratio of each area to be identified, and are calibrated as 、 and , Represents the normalized The Euclidean distance of the color mean of the area to be identified, Represents the normalized The texture direction histogram cross ratio of the area to be identified, Represents the normalized The pixel color variance ratio of the area to be identified, , is a positive integer;

[0030] Calculate the boundary recognition obstacle index of each area to be identified The specific calculation method is as follows: the normalized texture direction histogram cross ratio of each area to be identified Take the square to get the first term, and normalize the pixel color variance ratio of each area to be identified Take the square root to get the second term, and normalize the Euclidean distance of the color mean of each area to be identified Add one and take the reciprocal, then add one and take the natural logarithm as the third term. Add the first, second and third terms to get the boundary recognition obstacle index of each area to be identified. .

[0031] Preferably, based on the generated boundary transition perception coefficient of each area to be identified and boundary identification barrier index , the boundary recognition fuzzy index of each area to be identified is generated by weighted summation, specifically: Determine the boundary transition perception coefficient of each area to be identified and boundary identification barrier index The first non-zero weight coefficient and the second non-zero weight coefficient corresponding to each other are preset, and the sum of the first non-zero weight coefficient and the second non-zero weight coefficient is one; the first non-zero weight coefficient is multiplied by the boundary transition perception coefficient of each area to be identified The result of multiplying the second non-zero weight coefficient by the boundary recognition obstacle index of each area to be identified The results are summed to obtain the boundary recognition fuzzy index of each area to be identified .

[0032] Preferably, a preset boundary recognition fuzzy index threshold interval is determined , and after determination, the generated boundary recognition fuzzy index of each area to be identified Compare and evaluate the degree of boundary fuzziness of each area to be identified when the shrub color is close to the background color based on the comparison results. Then classify the areas based on the evaluation results. The specific comparison analysis and classification are as follows:

[0033] like , when the color of the shrub is close to the background color, the boundary recognition fuzziness of the area to be identified is low, and the area to be identified is divided into the first category area;

[0034] like , when the color of the shrub is close to the background color, the boundary recognition fuzziness of the area to be identified is medium fuzzy, then the area to be identified is divided into the second category area;

[0035] like When the color of the bush is close to the background color, the boundary recognition fuzziness of the area to be identified is high, and the area to be identified is divided into the third category area.

[0036] Preferably, according to the classification results, corresponding shrub vegetation spatial distribution area extraction methods are selected for different categories of areas to be identified, specifically:

[0037] For the areas to be identified that are classified as the first category, a fixed threshold segmentation method based on the pixel value distribution of the green band among the three red, green and blue bands is used to extract the spatial distribution area of ​​shrubs.

[0038] For the areas to be identified that are classified as the second category, the Laplacian edge convolution operation is first performed on the image to enhance the boundary information. Then, the red, green, and blue band images are channel-joined to form a fused image. The spatial distribution area of ​​shrubs is extracted based on the edge response intensity and color difference in the fused image and the combined threshold judgment.

[0039] For the area to be identified that is classified as the third category area, the current image and the image of the same area to be identified at a historical time point are obtained, and a change difference map of the green band pixels in the two phase images is generated. Then, in the change difference map, the area with enhanced green pixel values ​​and edge gradient amplitude change lower than the preset boundary change threshold is identified as the shrub spatial distribution area.

[0040] Preferably, a dynamic assessment system for ecological health of riverbank vegetation includes an image preprocessing module, a heterochromatic interference recognition module, a boundary fuzzy assessment module, a hierarchical extraction decision module, and an ecological display generation module;

[0041] The image preprocessing module acquires image data containing the riverbank area and performs preprocessing on the image data, including image denoising, registration, and multi-channel alignment. It also extracts the riverbank area image in combination with terrain information and divides the riverbank area image into several image unit grids.

[0042] The heterochromatic interference recognition module determines whether the shrub color is close to the background color in each image unit grid based on color and texture features, and marks the image unit grid with such a situation as the area to be recognized;

[0043] The boundary fuzzy assessment module extracts the boundary recognition feature information of each area to be identified, analyzes it after extraction, evaluates the boundary recognition fuzziness of each area to be identified when the shrub color is close to the background color, and classifies it according to the assessment results;

[0044] The hierarchical extraction decision module selects the corresponding shrub vegetation spatial distribution area extraction method for different categories of areas to be identified according to the classification results, so as to extract the shrub vegetation spatial distribution area in each area to be identified;

[0045] The ecological display generation module performs image range statistics on the extracted shrub spatial distribution area and completes the output display of the ecological health status of the river bank area based on the statistical results.

[0046] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0047] 1. This paper quantitatively analyzes the blurriness of shrub boundaries by constructing two high-dimensional parameter models: the "boundary transition perception coefficient" and the "boundary recognition obstruction index," which fuse image gradient structure and texture color fusion features. This effectively overcomes the traditional image segmentation method's heavy reliance on color transitions and edge clarity. Furthermore, by grading the blurriness and implementing a layered recognition strategy, the method selects the most adaptable spatial distribution region extraction method for complex scenes where shrubs and background colors are very close and the boundary outlines are unclear. This significantly improves the accuracy and stability of shrub extraction and avoids spatial bias caused by misidentification.

[0048] 2. Based on the technical concept of "hierarchical extraction of unidentified regions," this invention matches regions of varying blur levels to three extraction paths: fixed threshold segmentation, fused channel edge enhancement, and multi-temporal image change detection, thereby establishing a flexible regional adaptation mechanism. This multi-strategy fusion extraction framework not only improves the method's adaptability to varying image clarity, background complexity, and temporal dynamics, but also significantly enhances the system's generalizability to multiple image sources, varying observation conditions, and diverse vegetation habitats, resulting in a wider range of engineering applications and greater robustness.

[0049] 3. By using image grids as the minimum analysis unit and mapping the extracted results to the entire riverbank area for statistical output, this proposed technical solution enables structural-level assessment of shrub cover distribution, providing fundamental data support for ecological health with higher spatial resolution and clearer structural expression. This closed-loop analysis process, from pixel-level identification to regional-level statistics, not only improves the accuracy of ecological assessment results, but also avoids erroneous interventions caused by shrub identification bias. This significantly enhances the scientific nature, precision, and controllability of ecological governance, and has promising practical application value and widespread application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction to the drawings required for use in the embodiments will be given below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0051] Figure 1 This is a flow chart of a system and method for dynamically assessing the ecological health of riverbank vegetation according to the present invention.

[0052] Figure 2 This is a module diagram of a system and method for dynamically assessing the ecological health of riverbank vegetation according to the present invention. DETAILED DESCRIPTION

[0053] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of this disclosure will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art.

[0054] The present invention provides Figure 1 A dynamic assessment method for the ecological health of riverbank vegetation is shown, which specifically includes the following steps:

[0055] Acquire image data containing the riverbank area, perform preprocessing on the image data, including image denoising, registration, and multi-channel alignment, extract the riverbank area image in combination with terrain information, and divide the riverbank area image into several image unit grids;

[0056] Image data of riverbank areas can be collected through a variety of remote sensing platforms, including high-resolution satellite imagery (such as Sentinel-2 and Landsat-8), low-altitude aerial imagery from drones, and close-up photographic images collected by ground-based mobile devices. Upon receiving the raw image data, the software system first performs image denoising, typically using image processing algorithms such as median filtering, Gaussian filtering, or wavelet denoising, to reduce pixel-level anomalies caused by sensor noise, weather interference, and other factors. Subsequently, spatial alignment is performed on multi-temporal or multi-platform images using feature-based image registration methods (such as SIFT / SURF feature extraction and affine transformation matching) or grayscale-based template matching, aligning images acquired at different times or angles. After image registration, multi-channel alignment is performed, synchronously overlaying data from multiple bands or sensor channels, such as visible light, near-infrared, and elevation, with pixel-level accuracy to create a composite image with consistent structure and multispectral characteristics for subsequent shrub identification and health assessment.

[0057] Topographic information can be imported into the system using DEM (digital elevation model), DOM (digital orthophoto), or geographic vector boundary files (such as river channel vector lines and shoreline extent area maps) for fusion processing. The software first generates preliminary analysis blocks based on the river channel centerline and the left and right bank buffer zones. Elevation data is then combined to determine the slope foot, slope crest, and water edge characteristics. Non-riverbank highlands or water surface pixels are removed to accurately extract riparian image areas demonstrating topographic continuity and potential for vegetation growth. Next, based on image or geographic coordinates, the extracted riparian area is divided into a grid of image cells, either using a fixed-size grid (e.g., 50m×50m) or adaptive partitioning (e.g., based on image texture variations). Each image cell grid serves as the minimum analysis unit for subsequent assessment, facilitating block-by-block image recognition, index calculation, and health classification, supporting dynamic analysis and spatial comparison.

[0058] Image data preprocessing and gridding are necessary because raw images often suffer from noise, temporal dislocation, and spectral inconsistencies. Without registration, filtering, and multi-channel alignment, shrub information at the same location in the image may drift, blur, or even be missing, seriously affecting the accuracy of subsequent feature extraction and recognition algorithms. Furthermore, riverbanks themselves have irregular spatial distributions, fuzzy boundaries, and complex terrain variations. Failure to incorporate terrain information to eliminate non-analyzed areas will lead to non-target areas interfering with assessment results, leading to the erroneous extraction of non-vegetation objects such as farmland, bare land, water bodies, or man-made structures. Image gridding provides a stable spatial index unit for subsequent analysis, enabling the model to independently analyze the fuzziness of shrub boundaries and their health status on a grid-by-grid basis. This supports regional-level indicator aggregation, time series comparison, and visualization, effectively improving the stability, scalability, and interpretability of the overall system.

[0059] Judging whether the shrub color is close to the background color in each image unit grid based on the color feature and texture feature, and marking the image unit grid with such a situation as the area to be identified;

[0060] In this embodiment, the following steps are used to determine whether the color of the shrubs in each image unit grid is close to the background color based on the color and texture features, and mark the image unit grid with such a situation as a to-be-identified area:

[0061] Analyze the color distribution differences between shrub vegetation areas and adjacent non-vegetation areas in each image unit grid;

[0062] Analyzing the color distribution differences between shrub vegetation areas and adjacent non-vegetated areas within each image grid cell can be accomplished through software methods such as image partitioning, color statistics, and distribution modeling. First, based on image preprocessing and segmentation, a preliminary partitioning is performed between the suspected shrub vegetation areas and the surrounding non-vegetated areas within each image grid cell. Candidate shrub areas and background areas can be identified within each image grid cell using established Normalized Difference Vegetation Indices (such as NDVI) or color-based K-means clustering. Subsequently, the distribution of pixels in each region is statistically analyzed in RGB, HSV, or Lab color space, and the color distribution is modeled using a histogram or Gaussian mixture model. After modeling, the degree of color differentiation is assessed by calculating the statistical difference between the color distribution of the shrub area and the adjacent non-vegetated areas, using methods such as the Bhattacharyya distance, KL divergence, or mean-variance comparison. A high degree of overlap (i.e., low difference) between the two color distributions indicates that the shrub area and background area are similar in color, providing a basis for subsequent identification of areas with blurred boundaries. The entire process can be implemented through image processing software or deep learning image feature extraction tool chain, which supports batch processing of image grid data.

[0063] Analyze the distribution differences of texture arrangement direction and density between shrub vegetation areas and adjacent non-vegetation areas in each image unit grid;

[0064] To analyze the differences in texture arrangement direction and density between shrub vegetation areas and adjacent non-vegetated areas within each image unit grid, image processing methods for texture feature extraction and distribution analysis can be used. First, within each image unit grid, shrub vegetation areas and adjacent non-vegetated areas are preliminarily demarcated using NDVI or color clustering results. Subsequently, direction-sensitive texture analysis methods, such as gray-level co-occurrence matrix (GLCM), Gabor filter banks, or local binary patterns (LBP), are used to extract texture direction distribution and texture density features for each area. In direction extraction, the dominant direction of texture arrangement in each area and its directional concentration can be determined by comparing energy responses in multiple directions (e.g., 0°, 45°, 90°, and 135°), reflecting the orderliness and directionality of the texture structure. In density analysis, the frequency of texture pattern changes per unit area, or texture energy, can be calculated to measure the richness of detail within the area. After extraction, statistical modeling is performed on the texture features of the shrub and non-vegetated areas (e.g., constructing texture orientation histograms and density distribution curves). The degree of difference in orientation and density between the two areas is calculated, quantified using methods such as cross-entropy, histogram overlap ratio, or mean difference. Low differences indicate similar texture structures in the two areas, but there is a risk of blurred boundary identification. This analysis can be performed using image processing and pattern recognition software tools such as OpenCV, the Matlab Image Toolbox, or the Python Image Processing Library, making it suitable for batch, automated regional texture identification and comparison tasks.

[0065] When the color distribution difference between the shrub vegetation area and the adjacent non-vegetation area is lower than the color judgment standard set based on the pixel color value cluster difference threshold, and the distribution difference of the texture arrangement direction and density between the shrub vegetation area and the adjacent non-vegetation area is lower than the texture judgment standard set based on the texture direction histogram overlap ratio, it is determined that the shrub color is close to the background color in the image unit grid;

[0066] To achieve the goal of "determining that the shrub color and background color are similar in an image unit grid when the color distribution difference between the shrub vegetation area and the adjacent non-vegetation area is lower than the color judgment standard set by the pixel color value cluster difference threshold, and the texture arrangement direction and density distribution difference between the shrub vegetation area and the adjacent non-vegetation area is lower than the texture judgment standard set by the texture direction histogram overlap ratio," two difference judgment processes can be established and automated analysis is performed separately. In the color dimension, the pixels in the shrub area and the adjacent non-vegetation area are first clustered (using clustering methods such as K-means and MeanShift) to form two main color feature vectors. Based on these feature vectors, the distance between their cluster centers or the degree of distribution overlap in a color space (such as Lab space) is calculated. If the distance is lower than a preset threshold, the color difference is small. In the texture dimension, the texture direction histogram of each area is extracted and the overlap area ratio between the two area histograms is calculated. When this ratio exceeds a set limit, the texture distribution is similar. When both conditions are met, the system automatically marks the image grid as a "color-similar area" for subsequent fine-tuning. The entire judgment process can be completed automatically with the help of image recognition and statistical analysis software without the need for human intervention, and can be executed in parallel on multiple images.

[0067] The "color judgment criterion based on a pixel color value cluster difference threshold" involves counting the dominant color features of a large number of shrubs and non-vegetation areas across multiple image units. The typical distribution patterns of color value cluster distances between regions with clear and blurred boundaries are analyzed, and an empirical threshold is selected to distinguish between similar and dissimilar colors. For example, when the distance between cluster centers is less than the statistical average at a certain percentile, the colors are considered similar. The "texture judgment criterion based on the overlap ratio of texture direction histograms" extracts texture direction distribution curves of shrubs and non-vegetation areas from a large number of training samples, calculates the ratio of histogram intersection areas in samples with clear and blurred boundaries, and sets an overlap threshold as the basis for judgment. For example, when the overlap ratio of the texture direction histograms of two regions exceeds a certain set ratio (e.g., 85%), the textures are considered similar. Both of these criteria can be pre-trained offline through data statistics and sample annotation and embedded in the evaluation program to ensure consistency in subsequent processing and objectivity of the judgment results.

[0068] The image unit grids where the shrub color is close to the background color are marked as the area to be identified.

[0069] To achieve the goal of "marking image grids where shrubs are close in color to the background as areas to be identified," software can be used to introduce logical judgment and spatial mapping mechanisms into the image processing process. First, after completing color and texture difference analysis for each image grid, the system assigns a status flag variable to each grid. This variable records whether the grid meets both the color difference and texture difference criteria: "to be identified" and "to be identified." If both criteria are met, the status variable is set to "1," or marked as "to be identified." Otherwise, it is set to "0," or marked as "not to be identified." Based on this status variable, the system logically marks the corresponding grid position in a spatial mapping data structure (such as a two-dimensional array or raster index table). The marking results can be visualized as a mask image, visually displaying the location and extent of all areas to be identified on the image. Furthermore, to facilitate subsequent processing and classification, each image grid marked as to be identified can be assigned an index number or coordinate index information, which is stored in a task queue or recognition list for subsequent spatial distribution extraction steps. The entire labeling process does not rely on manual operation and can be completed automatically by the algorithm in real time during the image analysis process. It is suitable for batch processing needs of large-scale image data.

[0070] Extract the boundary recognition feature information of each area to be identified, analyze it after extraction, evaluate the degree of boundary recognition fuzziness of each area to be identified when the shrub color is close to the background color, and classify it according to the evaluation results;

[0071] In this embodiment, the boundary recognition feature information of each area to be identified is extracted and analyzed after extraction to evaluate the degree of boundary recognition fuzziness of each area to be identified when the shrub color is close to the background color. Classification is performed based on the evaluation results, specifically:

[0072] Extracting boundary recognition feature information of each area to be identified and performing preprocessing operations after extraction;

[0073] Extracting boundary recognition feature information for each target region typically uses the target region mask generated after image segmentation as the positioning basis, combining the image's brightness distribution, pixel gradients, and texture response values ​​for spatial analysis. The specific extraction method involves first applying the Sobel or Canny edge detection operator in the image grayscale space to extract boundary pixels for each target region and construct a boundary pixel sequence. Then, within this boundary pixel band, feature dimensions such as horizontal and vertical brightness gradients, edge amplitude, gradient direction, edge thickness, and texture response distribution are extracted to form a multidimensional boundary feature vector set. Furthermore, texture analysis methods such as gray-level co-occurrence matrices, Gabor filters, or local binary patterns (LBP) are used to further extract texture direction and density information within the boundary band. This process is generated pixel by pixel or block by block using a sliding window or region masking approach. All extracted information is bound to the region index to form a complete set of boundary recognition feature information for subsequent processing.

[0074] The primary purpose of preprocessing is to improve the comparability and stability of boundary recognition feature information and reduce interference caused by factors such as noise, brightness drift, and regional scale differences. First, the extracted feature information is standardized. For example, brightness gradient values ​​and directional angle values ​​are range-normalized to ensure that each dimension has a uniform dimension, thus avoiding weight shifts caused by scale inconsistencies in subsequent analysis. Second, to eliminate interference from local outliers, the feature data can be subjected to median filtering or Gaussian smoothing to enhance the continuity and interpretability of the overall boundary features. Furthermore, introducing principal component analysis (PCA) or feature selection algorithms into texture features can compress dimensions and eliminate redundant indicators, ensuring that the input data is both compact and discriminative. The entire preprocessing process is automatically performed using image processing libraries and mathematical statistical tools, requiring no human intervention. It is widely adaptable to multi-resolution and multi-source image data, providing a solid foundation for subsequent boundary fuzzy assessment.

[0075] Extract boundary structure contrast information and image texture and color fusion information from the preprocessed boundary recognition feature information, and perform normalization processing after extraction;

[0076] To extract boundary structure contrast information and image texture and color fusion information from preprocessed boundary recognition feature information, multi-dimensional statistical analysis and feature coupling processing are typically required for the boundary neighborhood of each region to be identified. First, to extract boundary structure contrast information, the standardized gradient magnitude map and gradient direction map can be used to calculate the grayscale gradient change amplitude, edge direction consistency, and edge transition distance for each pixel band along the boundary of the region to be identified. The grayscale gradient change amplitude is calculated by the average grayscale difference of 3×3 pixel blocks on both sides of the boundary line, the direction consistency is obtained by calculating the concentration of gradient directions, and the transition distance is quantified based on the distance curve of the edge intensity decay from the center to the outside. These three factors together constitute the boundary structure contrast information of the region. Second, to extract image texture and color fusion information, the color channel mean and standard deviation, color variance ratio, and the degree of distribution overlap of texture histograms (such as LBP or Gabor texture) can be calculated within and outside the boundary band. By weightedly combining color and texture differences, a comprehensive data representation describing the clarity of the region boundary is formed. These extraction processes are automatically performed through image analysis and matrix operations, without relying on human intervention, and can efficiently extract boundary blur related information on high-resolution images.

[0077] Based on the normalized boundary structure contrast information and image texture and color fusion information, the boundary transition perception coefficient and boundary recognition obstacle index of each area to be identified are generated respectively;

[0078] Generate a boundary recognition fuzzy index for each area to be recognized based on the generated boundary transition perception coefficient and boundary recognition obstruction index of each area to be recognized;

[0079] Determine the pre-set boundary recognition fuzzy index threshold range, and compare it with the generated boundary recognition fuzzy index of each area to be identified after determination. According to the comparison results, evaluate the boundary recognition fuzziness of each area to be identified when the shrub color is close to the background color, and classify it according to the evaluation results.

[0080] To determine the threshold range for the boundary recognition fuzzy index using software, offline statistics and distribution modeling can be performed based on annotated areas of known shrub boundary recognition quality in a historical image sample library. Specifically, a representative image dataset is first selected. For multiple shrub areas with varying degrees of manually verified boundary recognition difficulty, the corresponding boundary recognition fuzzy index values ​​are calculated. These values ​​are then categorized into three groups based on recognition performance: "easy to identify," "medium," and "difficult to identify." Then, based on the fuzzy index distributions of these three groups, statistical methods such as K-means clustering, the natural breaks method (Jenks), or the Gaussian mixture model (GMM) are used to segment the index values ​​and automatically define three non-overlapping intervals corresponding to low, medium, and high fuzziness, respectively. Once the intervals are segmented, the upper and lower bounds of each interval are used as preset thresholds for the boundary recognition fuzzy index and fixed as callable configuration parameters for subsequent real-time comparison and classification operations. The entire threshold setting process can be automatically integrated into the model training or system initialization phase, eliminating the need for manual adjustment and ensuring consistent and objective evaluation criteria.

[0081] In this embodiment, the logic for obtaining the boundary transition perception coefficient of each area to be identified is as follows:

[0082] The boundary structure contrast information is extracted from the preprocessed boundary recognition feature information, specifically including the average grayscale gradient amplitude, grayscale change transition length and gradient direction standard deviation of the edge of each area to be identified. These three types of data are divided by their corresponding preset maximum values ​​to obtain the normalized average grayscale gradient amplitude, grayscale change transition length and gradient direction standard deviation of the edge of each area to be identified, and are calibrated as 、 and , Represents the normalized The average gray gradient amplitude of the edge of the area to be identified, Represents the normalized The grayscale change transition length of the edge of the area to be identified, Represents the normalized The standard deviation of the gradient direction of the edge of the area to be identified, , is a positive integer;

[0083] Three types of data are obtained for the edges of each region to be identified: the average grayscale gradient amplitude, grayscale transition length, and gradient direction standard deviation. This can be achieved by scanning the edge regions pixel by pixel in the software using image processing algorithms. Specifically, the Sobel or Prewitt operator is first used to identify edge locations in each region to extract a set of edge pixels. The average grayscale gradient amplitude for the region is then calculated by calculating the gradient amplitude of each edge pixel and averaging the gradient amplitudes of all edge pixels in the region. Second, in the gradient distribution map, the continuous transition range of grayscale values ​​from high to low or low to high is calculated along an axis perpendicular to the edge direction, with the center point of each edge as the reference point. This is the distance over which the slope of the grayscale change curve reaches a plateau. This is used to calculate the edge grayscale transition length. Finally, the standard deviation of the gradient direction angles (typically 0° to 180°) of all edge pixels is calculated to indicate the degree of boundary direction dispersion, thus obtaining the gradient direction standard deviation for the region. All of these operations can be automatically performed in batches within the software using the image analysis toolkit.

[0084] Dividing each of these three data types by their respective preset maximum values ​​is intended to normalize them, eliminating differences in numerical scale and physical units. This prevents one data type from dominating the overall calculation due to its excessively large absolute value during the subsequent calculation of the boundary transition perception coefficient, ensuring that all structural information is equally weighted in the index. The respective preset maximum values ​​can be obtained through sample statistics. During the model construction phase, a large number of edge sample data are extracted from multiple images of representative riverbank areas, and the maximum observed value of each of the three data types across all samples is recorded. Alternatively, the upper quantile (e.g., the 95th percentile) after removing extreme values ​​is set as the maximum value, thereby improving the model's stability and interference resistance. This statistical threshold setting can be implemented by software during training, where parameters are learned and fixed as a baseline configuration. This can then be applied as needed during runtime, ensuring the scientific and universal nature of the normalization process.

[0085] Calculate the boundary transition perception coefficient of each area to be identified The specific calculation method is as follows: the normalized grayscale change transition length of each edge of the area to be identified is The exponential power is divided by the normalized average gray gradient amplitude of the edge of each area to be identified The natural logarithm of the ratio obtained by adding one and the normalized standard deviation of the gradient direction of the edge of each area to be identified Take the square root and add them to get the boundary transition perception coefficient of each area to be identified ;

[0086] The specific calculation formula is as follows: , where For the The boundary transition perception coefficient of the area to be identified.

[0087] This formula is used to calculate the boundary transition perception coefficient for each region to be identified. Its core purpose is to comprehensively assess the blur level of shrub boundaries within an image by combining and modeling the multi-dimensional features of the image's edge structure. The three parameters in the formula reflect boundary clarity in terms of boundary orientation consistency, grayscale transition width, and edge sharpness. These parameters are then fused using a nonlinear function to enhance the discernibility of boundary blur perception.

[0088] Specifically: the first It represents the square root of the normalized standard deviation of the edge gradient direction. Its physical meaning is to reflect the degree of disturbance in the edge direction. The larger the standard deviation, the more inconsistent and irregular the edge contour direction is, and the stronger the blur. Using the square root function can weaken the dominance of the maximum value, so that the influence of the boundary direction distribution is more smoothly integrated into the final index. The exponential operation representing the normalized grayscale transition length reflects the continuous blur transition effect caused by the slow change of grayscale values ​​in the boundary area. The longer the grayscale transition, the smoother the pixel grayscale change from the shrub to the background in the boundary area. The exponential function can amplify the weight of such boundary blur areas in the index, making them occupy a more prominent position in the boundary blur perception. The normalized average gradient amplitude of edge pixels plus 1 represents a modulating factor for edge sharpness. A larger gradient indicates a sharper boundary, thus suppressing the blur exponent in the denominator; the addition of 1 prevents the denominator from being zero and controls the numerical scale. Dividing the exponential term by the gradient term and taking its natural logarithm helps compress the resulting numerical range and enhance the model's ability to distinguish between blur and sharpness under different boundary conditions. The entire formula structure embodies a balancing act: directional confusion and slow transitions enhance blur, while edge sharpness suppresses it. Through a weighted combination of these factors, the index effectively captures the indistinguishability between shrub boundaries and backgrounds in situations of similar color, providing a quantifiable basis for downstream classification and region extraction strategies.

[0089] No. Boundary transition perception coefficient of the area to be identified This coefficient reflects the degree of blurriness of the boundary features between the shrubs and the background in the image. A larger value indicates that the boundary is more difficult to segment using conventional image analysis methods. Specifically, this coefficient is composed of factors such as the discreteness of edge directions, the width of grayscale transitions, and edge sharpness. A more dispersed edge direction, a slower grayscale transition, and a weaker edge gradient all contribute to a higher coefficient. Because these factors are crucial for edge recognition accuracy, the value of the boundary transition perception coefficient can be used as a direct indicator of the degree of boundary blurriness. A smaller coefficient indicates a clear boundary structure, short transitions, and a concentrated direction, making the boundary easy to identify. A larger coefficient indicates weak contrast, weak structure, and weak directionality, making it difficult to accurately segment the true boundary when the shrubs and background colors are similar. Therefore, this coefficient is highly explanatory and operational in assessing boundary blurriness and is an essential parameter for quantifying "difficult-to-identify boundaries."

[0090] In this embodiment, the logic for obtaining the boundary recognition obstacle index of each area to be recognized is as follows:

[0091] The image texture and color fusion information is extracted from the preprocessed boundary recognition feature information, which specifically includes three types of data: the color mean Euclidean distance of each area to be identified, the texture direction histogram cross ratio, and the pixel color variance ratio. These three types of data are divided by their corresponding preset maximum values ​​to obtain the normalized color mean Euclidean distance, texture direction histogram cross ratio, and pixel color variance ratio of each area to be identified, and are calibrated as 、 and , Represents the normalized The Euclidean distance of the color mean of the area to be identified, Represents the normalized The texture direction histogram cross ratio of the area to be identified, Represents the normalized The pixel color variance ratio of the area to be identified, , is a positive integer;

[0092] Obtaining three types of data for each region to be identified: the Euclidean distance of color mean, the texture direction histogram intersection ratio, and the pixel color variance ratio. This can be achieved through software based on image processing and feature extraction algorithms. The color mean Euclidean distance is calculated by calculating the mean vector of the pixel colors in RGB or Lab color space between the shrub vegetation area and the adjacent non-vegetated area. The Euclidean distance formula is then used to measure the difference in color mean between the two, reflecting their overall color differentiation. The texture direction histogram intersection ratio measures the similarity of texture directions between the two regions by extracting the gradient direction distribution of the two regions (for example, using the directional gradient histogram or Gabor filter to construct a histogram) and then calculating their intersection ratio. The pixel color variance ratio is calculated by calculating the standard deviation of the pixel colors of the two regions and taking their ratio. This value is used to describe the consistency and variation of color within the region. All three types of data can be automatically obtained by analyzing the pixel attributes of different regions within each image unit grid and have clear numerical expressions.

[0093] The purpose of dividing the above three types of data by their respective preset maximum values ​​is to complete the normalization process, unify the numerical range to eliminate the deviation caused by different dimensions and scales in subsequent calculations, and improve the comparability between parameters and the accuracy of weight control. The preset maximum value can be obtained by counting the historical maximum observation value or the upper limit of the 95% confidence interval of this type of data in a large number of representative sample images. It can also be determined by combining manual experience to set an upper limit that does not exceed the maximum value that may occur in the system to ensure that the normalized value is stable between 0 and 1. This normalization method facilitates the subsequent integration of various indicators into the index calculation model, enhancing the stability and versatility of the overall evaluation.

[0094] Calculate the boundary recognition obstacle index of each area to be identified The specific calculation method is as follows: the normalized texture direction histogram cross ratio of each area to be identified Take the square to get the first term, and normalize the pixel color variance ratio of each area to be identified Take the square root to get the second term, and normalize the Euclidean distance of the color mean of each area to be identified Add one and take the reciprocal, then add one and take the natural logarithm as the third term. Add the first, second and third terms to get the boundary recognition obstacle index of each area to be identified. ;

[0095] The specific calculation formula is as follows: , where For the The boundary recognition obstacle index of the area to be identified.

[0096] The calculation method of the boundary recognition obstruction index aims to integrate the boundary demarcation features of the shrub vegetation area and the adjacent non-vegetation area in three dimensions: color, texture direction, and color distribution difference, to measure the clarity of the boundary of the image area when the shrub and background colors are close. It represents the square of the normalized texture direction histogram cross ratio, reflecting the consistency or overlap of regional texture directions. The square treatment enhances the discrimination power of this index under similar texture conditions. The larger the value, the closer the texture directions of the two regions are and the less obvious the boundary is. It is the square root of the pixel color variance ratio, which is used to measure the uniformity of the color distribution difference within the region. The larger the value, the smaller the internal color difference and the less prominent the boundary. This is a nonlinear mapping of the Euclidean distance of color means, used to smooth the impact of small differences on the overall result, significantly increasing the contrast index of regions with similar color means. Overall, when all three indicators are high, it indicates that the region's boundary color differences are small, texture direction overlap is high, the internal color distribution is uniform, the transition between regions is natural, and the boundaries are blurred. Consequently, the Boundary Identification Hinderation Index value increases, indicating that the region's boundaries are difficult to identify. Conversely, a low index value indicates that the boundary features between regions are clear and easy to accurately identify.

[0097] No. Boundary recognition obstacle index of the area to be identified The value of is positively correlated with the degree of boundary ambiguity in the assessment of the area when the shrub color is close to the background color. Specifically, when the value of this index is large, it means that the boundary contrast between the area to be identified and the adjacent non-vegetated area is low in multiple dimensions such as color mean, texture direction, and color distribution. In other words, the difference between regions is weak, the transition is gentle, and the texture is converging, making it difficult to clearly segment the boundary in the image, and it belongs to a high-ambiguity area; when the index value is low, it means that the shrub area and the adjacent area have significant differences in color and texture, and the boundary is clearly distinguishable, which belongs to an area with low boundary ambiguity. Therefore, the larger the value of the boundary identification hindering index, the higher the degree of boundary ambiguity, and the more complex the extraction strategy needs to be to avoid recognition bias.

[0098] In this embodiment, based on the generated boundary transition perception coefficients of each area to be identified and boundary identification barrier index , the boundary recognition fuzzy index of each area to be identified is generated by weighted summation, specifically: Determine the boundary transition perception coefficient of each area to be identified and boundary identification barrier index The first non-zero weight coefficient and the second non-zero weight coefficient corresponding to each other are preset, and the sum of the first non-zero weight coefficient and the second non-zero weight coefficient is one; the first non-zero weight coefficient is multiplied by the boundary transition perception coefficient of each area to be identified The result of multiplying the second non-zero weight coefficient by the boundary recognition obstacle index of each area to be identified The results are summed to obtain the boundary recognition fuzzy index of each area to be identified .

[0099] The specific calculation formula is as follows: , where For the The boundary recognition fuzzy index of the area to be identified, and are the boundary transition perception coefficients of each area to be identified and boundary identification barrier index respectively corresponding to a first non-zero weight coefficient and a second non-zero weight coefficient, and .

[0100] In the actual implementation process, the boundary transition perception coefficient of each area to be identified can be calculated by software. Boundary Identification Obstacle Index Read uniformly and assign the first non-zero weight coefficient ( ) and the second non-zero weight coefficient ( ), and then use the weighted summation formula to calculate the boundary recognition fuzzy index These two weight coefficients are used to measure and The relative contribution in the fuzziness assessment, The larger the value, the more emphasis is placed on the boundary transition characteristics. The larger the value, the more emphasis is placed on the fusion of color and texture. The two coefficients can be preset by statistically fitting the historical image training data set: in the known labeled vegetation area samples, extract and The consistency measure of the degree of fuzziness assessed manually is used to determine a set of optimal and ; Empirical models can also be used to assign weights in different scenarios to ensure adaptability and model stability. The above settings can be configured through graphical interface parameters or pre-set model configuration files loaded and called by the software, without manual intervention.

[0101] In this embodiment, the preset boundary recognition fuzzy index threshold range is determined , and after determination, the generated boundary recognition fuzzy index of each area to be identified Compare and evaluate the degree of boundary fuzziness of each area to be identified when the shrub color is close to the background color based on the comparison results. Then classify the areas based on the evaluation results. The specific comparison analysis and classification are as follows:

[0102] like , when the color of the shrub is close to the background color, the boundary recognition fuzziness of the area to be identified is low, and the area to be identified is divided into the first category area;

[0103] This indicates that the boundary features between the shrub vegetation and the adjacent background in this area are clearly visible in the image, with strong contrast in color, texture, and structure. The boundaries of these areas are easily and accurately identified by current edge detection or segmentation methods in image analysis, providing excellent recognition conditions. Therefore, in the subsequent extraction of shrub spatial distribution areas, conventional methods based on color segmentation or edge detection can be directly used to achieve efficient segmentation, avoiding excessive processing while ensuring data processing efficiency and discrimination accuracy, facilitating the rapid construction of a reliable and reliable shrub cover map.

[0104] like , when the color of the shrub is close to the background color, the boundary recognition fuzziness of the area to be identified is medium fuzzy, then the area to be identified is divided into the second category area;

[0105] This indicates that the boundary between shrub vegetation and the background in this area appears partially blurred in the image, indicating a degree of color fusion or texture similarity, but not complete indistinguishability. Relying solely on single-image feature processing methods to extract shrub space in such areas is prone to omission or misclassification. Therefore, multidimensional processing techniques (such as edge enhancement and multi-band fusion) are necessary to improve identification accuracy. The quality of processing in such areas directly affects the accurate calculation of structural integrity and coverage in ecological health indicators and is crucial for ensuring the reliability of assessment results.

[0106] like When the color of the bush is close to the background color, the boundary recognition fuzziness of the area to be identified is high, and the area to be identified is divided into the third category area.

[0107] This indicates that the boundary contrast between the shrubs and the background in the image is extremely weak, and the color, texture, and edge transitions are highly similar, making it difficult for current common segmentation methods to accurately distinguish shrub vegetation. Direct extraction without processing will lead to a serious underestimation of shrub cover, affecting the effectiveness of ecological structure indicator evaluation. Therefore, when processing such areas, more complex techniques such as multi-temporal image comparison and trend analysis should be adopted to enhance discrimination and ensure that such shrub areas are properly identified and restored, avoiding biased conclusions about ecological status and thus affecting the scientific nature of subsequent intervention decisions.

[0108] According to the classification results, the corresponding shrub vegetation spatial distribution area extraction method is selected for the different categories of areas to be identified, so as to extract the shrub spatial distribution area in each area to be identified;

[0109] In this embodiment, according to the classification results, corresponding shrub vegetation spatial distribution area extraction methods are selected for different categories of areas to be identified, specifically:

[0110] For the areas to be identified that are classified as the first category, a fixed threshold segmentation method based on the pixel value distribution of the green band among the three red, green and blue bands is used to extract the spatial distribution area of ​​shrubs.

[0111] For the areas to be identified that are classified as the first category, since their boundary identification fuzziness is low, the shrub vegetation areas in the image have obvious differences in color and structure from the surrounding background, especially in the pixel values ​​of the green band. Therefore, they can be extracted through a fixed threshold segmentation method based on the distribution of green band pixel values ​​in the three red, green and blue bands. The specific implementation method is as follows: first, the pixel value matrix of the green band in the image unit is extracted, and the pixel histogram analysis of the entire image is performed. Combined with sample statistics or historical analysis, a fixed threshold interval for the green band is determined. This interval corresponds to the pixel intensity range formed by the reflection of typical shrub leaves. Then, each pixel in the green band image is traversed, and any pixel whose pixel value falls within this fixed interval is marked as a suspected shrub pixel. Further, through morphological operations such as opening and closing operations and connected domain analysis, isolated noise points are removed and scattered areas are connected, ultimately generating a mask layer for the spatial distribution of shrubs. This method is suitable for the first category of areas because the boundaries of this type of areas are clear and the green features are significant. Color-dominated extraction can quickly and efficiently complete the identification of distribution areas without introducing complex fusion or time series analysis operations, and has both accuracy and computational efficiency.

[0112] For the areas to be identified that are classified as the second category, the Laplacian edge convolution operation is first performed on the image to enhance the boundary information. Then, the red, green, and blue band images are channel-joined to form a fused image. The spatial distribution area of ​​shrubs is extracted based on the edge response intensity and color difference in the fused image and the combined threshold judgment.

[0113] For the areas to be identified, classified as the second category, the boundary blur is moderate. While the color difference and structural boundary information between the shrubs and the background in the image are relatively weak, they still provide a basis for identification. Therefore, a combined fusion edge enhancement and color difference judgment method is used to improve extraction accuracy. Specifically, the image in this area is first subjected to a Laplacian edge convolution operation. This significantly increases the edge response strength of the shrub boundary by enhancing areas with prominent gradient changes. The red, green, and blue bands are then normalized and channel-wise concatenated to construct a three-channel fused image that preserves both color information and edge features. For each pixel in the fused image, the edge response strength (e.g., based on the absolute value of the convolution result) and the color difference with surrounding pixels (e.g., based on Euclidean distance or cosine similarity) are calculated. A joint threshold is then set to determine whether a pixel belongs to the shrub area. A pixel is considered to be in the target area only if both its edge response strength and color difference exceed the edge threshold. Finally, a connectivity analysis is performed on the pixels that meet these conditions, and a continuous region is extracted as the shrub distribution area. This method is adopted because the second category area has a certain degree of fuzziness in color and edge features, which can easily lead to misjudgment or missed detection through single-dimensional segmentation. The use of a two-dimensional joint threshold strategy in the fused image can effectively utilize the remaining feature difference information to achieve more robust area recognition.

[0114] For the area to be identified that is classified as the third category area, the current image and the image of the same area to be identified at a historical time point are obtained, and a change difference map of the green band pixels in the two phase images is generated. Then, in the change difference map, the area with enhanced green pixel values ​​and edge gradient amplitude change lower than the preset boundary change threshold is identified as the shrub spatial distribution area.

[0115] For areas classified as the third category, their boundaries are highly ambiguous, meaning that the shrubs and background in the current image have almost no discernible differences in color, texture, and edges. Conventional image segmentation and edge detection methods are unable to accurately distinguish the spatial distribution of shrubs. To extract the spatial distribution of shrubs in areas classified as the third category while maintaining identifiable boundaries despite high image boundary ambiguity, a combined approach of two-temporal image difference analysis and boundary gradient assessment is employed. First, the software selects the current image and an image acquired at a historical time point. After ensuring consistent georegistration accuracy between the two images, pixel values ​​in the green band of these two images are extracted, and a green channel change map for each pixel at the two time points is constructed. By calculating the pixel difference in the green band between the two images, a green band change difference map is generated. An edge detection algorithm (such as the Sobel, Laplacian, or Canny operator) is then applied to this difference map to calculate the edge gradient magnitude for each pixel. The edge gradient map after the change is then subtracted from the edge gradient map before the change to obtain an edge gradient magnitude change map. Based on this, a boundary change assessment threshold is set. Regions where green pixel values ​​significantly increase but edge gradient changes fall below this threshold are identified as areas showing increased vegetation growth over time but stable structural boundaries. These regions can be inferred to be shrub areas that were previously difficult to clearly segment from single-temporal images. This method has the advantage of combining the changing trends of time series analysis with the stability characteristics of spatial structure to effectively improve the accuracy of shrub extraction in situations where boundary identification is ambiguous. It is suitable for scenarios where the shrub color in an image is very close to the background, making traditional boundary extraction difficult. The entire process can be embedded in image analysis software using a streamlined algorithm, supporting automatic identification and batch processing.

[0116] The image range statistics of the extracted shrub spatial distribution area are performed, and the output display of the ecological health status of the river bank area is completed based on the statistical results.

[0117] To achieve "conducting image range statistics on the extracted shrub spatial distribution areas, and completing the output display of the ecological health status of the riverbank area based on the statistical results", we can first conduct a pixel-level range analysis on the identified shrub spatial distribution areas. Specifically, using the pixel labels marked as shrub areas in the image segmentation results, the total number of pixels in each image unit grid and the entire riverbank area is accumulated, and converted into area units based on the spatial resolution of the image. This statistical operation can be automatically completed through raster image processing methods, and the statistical results represent the spatial range of shrub distribution. In order to facilitate regional comparisons, the riverbank can also be divided into zones according to the longitudinal or cross-sectional directions, and the shrub coverage area of ​​each area can be summarized separately to observe its spatial distribution pattern and relative concentration.

[0118] After completing the above statistics, in order to demonstrate the ecological health status, the statistically obtained shrub distribution range can be compared with the set regional expected vegetation range, historical average distribution value or related standard parameters to construct a regional evaluation level based on spatial coverage ratio. For example, by setting different area coverage rate thresholds, each image unit or area is assigned a low, medium, and high ecological status label. The assessment results are then visualized using color grading or grade labeling in combination with graphic rendering methods, giving the ecological assessment results an intuitive expression. This display method not only assists researchers or decision makers in judging the status of shrub habitats, but also serves as an input interface to provide basic data for subsequent processing links such as ecological restoration simulation and dynamic evolution monitoring. At the software level, the entire process can be achieved through the linkage of image statistical analysis, geographic information system mapping components and graphic display interfaces, with a coherent process and friendly interaction.

[0119] like Figure 2 The system for dynamic assessment of riverbank vegetation ecological health shown in the figure includes an image preprocessing module, a heterochromatic interference recognition module, a boundary fuzzy assessment module, a hierarchical extraction decision module, and an ecological display generation module;

[0120] The image preprocessing module acquires image data containing the riverbank area and performs preprocessing on the image data, including image denoising, registration, and multi-channel alignment. It also extracts the riverbank area image in combination with terrain information and divides the riverbank area image into several image unit grids.

[0121] The heterochromatic interference recognition module determines whether the shrub color is close to the background color in each image unit grid based on color and texture features, and marks the image unit grid with such a situation as the area to be recognized;

[0122] The boundary fuzzy assessment module extracts the boundary recognition feature information of each area to be identified, analyzes it after extraction, evaluates the boundary recognition fuzziness of each area to be identified when the shrub color is close to the background color, and classifies it according to the assessment results;

[0123] The hierarchical extraction decision module selects the corresponding shrub vegetation spatial distribution area extraction method for different categories of areas to be identified according to the classification results, so as to extract the shrub vegetation spatial distribution area in each area to be identified;

[0124] The ecological display generation module performs image range statistics on the extracted shrub spatial distribution area and completes the output display of the ecological health status of the river bank area based on the statistical results.

[0125] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0126] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in the embodiments of this application are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0127] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0128] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0129] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. 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. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0130] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0131] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0132] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A dynamic assessment method for the ecological health of riverbank vegetation, characterized in that: The specific steps include: Acquire image data containing the riverbank area, perform preprocessing on the image data, including image denoising, registration, and multi-channel alignment, extract the riverbank area image in combination with terrain information, and divide the riverbank area image into several image unit grids; Judging whether the shrub color is close to the background color in each image unit grid based on the color feature and texture feature, and marking the image unit grid with such a situation as the area to be identified; Extract the boundary recognition feature information of each area to be identified, analyze it after extraction, evaluate the degree of boundary recognition fuzziness of each area to be identified when the shrub color is close to the background color, and classify it according to the evaluation results; According to the classification results, the corresponding shrub vegetation spatial distribution area extraction method is selected for the different categories of areas to be identified, so as to extract the shrub spatial distribution area in each area to be identified; The image range statistics of the extracted shrub spatial distribution area are performed, and the output display of the ecological health status of the river bank area is completed based on the statistical results.

2. A dynamic assessment method for riverbank vegetation ecological health according to claim 1, characterized in that: Based on the color and texture features, it is determined whether the shrub color of each image unit grid is close to the background color, and the image unit grid with such a situation is marked as the area to be identified, including: Analyze the color distribution differences between shrub vegetation areas and adjacent non-vegetation areas in each image unit grid; Analyze the distribution differences of texture arrangement direction and density between shrub vegetation areas and adjacent non-vegetation areas in each image unit grid; When the color distribution difference between the shrub vegetation area and the adjacent non-vegetation area is lower than the color judgment standard set based on the pixel color value cluster difference threshold, and the distribution difference of the texture arrangement direction and density between the shrub vegetation area and the adjacent non-vegetation area is lower than the texture judgment standard set based on the texture direction histogram overlap ratio, it is determined that the shrub color is close to the background color in the image unit grid; The image unit grids where the shrub color is close to the background color are marked as the area to be identified.

3. A dynamic assessment method for riverbank vegetation ecological health according to claim 2, characterized in that: Extract the boundary recognition feature information of each area to be identified, analyze it after extraction, evaluate the degree of boundary recognition fuzziness of each area to be identified when the shrub color is close to the background color, and classify it according to the evaluation results, specifically as follows: Extracting boundary recognition feature information of each area to be identified and performing preprocessing operations after extraction; Extract boundary structure contrast information and image texture and color fusion information from the preprocessed boundary recognition feature information, and perform normalization processing after extraction; Based on the normalized boundary structure contrast information and image texture and color fusion information, the boundary transition perception coefficient and boundary recognition obstacle index of each area to be identified are generated respectively; Generate a boundary recognition fuzzy index for each area to be recognized based on the generated boundary transition perception coefficient and boundary recognition obstruction index of each area to be recognized; Determine the pre-set boundary recognition fuzzy index threshold range, and compare it with the generated boundary recognition fuzzy index of each area to be identified after determination. According to the comparison results, evaluate the boundary recognition fuzziness of each area to be identified when the shrub color is close to the background color, and classify it according to the evaluation results.

4. A dynamic assessment method for riverbank vegetation ecological health according to claim 3, characterized in that: The logic for obtaining the boundary transition perception coefficient of each area to be identified is as follows: The boundary structure contrast information is extracted from the preprocessed boundary recognition feature information, specifically including the average grayscale gradient amplitude, grayscale change transition length and gradient direction standard deviation of the edge of each area to be identified. These three types of data are divided by their corresponding preset maximum values ​​to obtain the normalized average grayscale gradient amplitude, grayscale change transition length and gradient direction standard deviation of the edge of each area to be identified, and are calibrated as 、 and , Represents the normalized The average gray gradient amplitude of the edge of the area to be identified, Represents the normalized The grayscale change transition length of the edge of the area to be identified, Represents the normalized The standard deviation of the gradient direction of the edge of the area to be identified, , is a positive integer; The specific calculation formula is as follows: , where For the The boundary transition perception coefficient of the area to be identified.

5. A dynamic assessment method for riverbank vegetation ecological health according to claim 4, characterized in that: The logic for obtaining the boundary recognition obstacle index of each area to be identified is as follows: The image texture and color fusion information is extracted from the preprocessed boundary recognition feature information, which specifically includes three types of data: the color mean Euclidean distance of each area to be identified, the texture direction histogram cross ratio, and the pixel color variance ratio. These three types of data are divided by their corresponding preset maximum values ​​to obtain the normalized color mean Euclidean distance, texture direction histogram cross ratio, and pixel color variance ratio of each area to be identified, and are calibrated as 、 and , Represents the normalized The Euclidean distance of the color mean of the area to be identified, Represents the normalized The texture direction histogram cross ratio of the area to be identified, Represents the normalized The pixel color variance ratio of the area to be identified, , is a positive integer; The specific calculation formula is as follows: , where For the The boundary recognition obstacle index of the area to be identified.

6. A method for dynamic assessment of riverbank vegetation ecological health according to claim 5, characterized in that: Based on the generated boundary transition perception coefficients of each area to be identified and boundary identification barrier index , the boundary recognition fuzzy index of each area to be identified is generated by weighted summation. The specific calculation formula is as follows: , where For the The boundary recognition fuzzy index of the area to be identified, and are the boundary transition perception coefficients of each area to be identified and boundary identification barrier index respectively corresponding to a first non-zero weight coefficient and a second non-zero weight coefficient, and .

7. A dynamic assessment method for riverbank vegetation ecological health according to claim 6, characterized in that: Determine the pre-set boundary recognition fuzzy index threshold range , and after determination, the generated boundary recognition fuzzy index of each area to be identified Compare and evaluate the degree of boundary fuzziness of each area to be identified when the shrub color is close to the background color based on the comparison results. Then classify the areas based on the evaluation results. The specific comparison analysis and classification are as follows: like , when the color of the shrub is close to the background color, the boundary recognition fuzziness of the area to be identified is low, and the area to be identified is divided into the first category area; like , when the color of the shrub is close to the background color, the boundary recognition fuzziness of the area to be identified is medium fuzzy, then the area to be identified is divided into the second category area; like When the color of the bush is close to the background color, the boundary recognition fuzziness of the area to be identified is high, and the area to be identified is divided into the third category area.

8. A dynamic assessment method for riverbank vegetation ecological health according to claim 7, characterized in that: According to the classification results, the corresponding shrub vegetation spatial distribution area extraction method is selected for the different categories of areas to be identified, specifically: For the areas to be identified that are classified as the first category, a fixed threshold segmentation method based on the pixel value distribution of the green band among the three red, green and blue bands is used to extract the spatial distribution area of ​​shrubs. For the areas to be identified that are classified as the second category, the Laplacian edge convolution operation is first performed on the image to enhance the boundary information. Then, the red, green, and blue band images are channel-joined to form a fused image. The spatial distribution area of ​​shrubs is extracted based on the edge response intensity and color difference in the fused image and the combined threshold judgment. For the area to be identified that is classified as the third category area, the current image and the image of the same area to be identified at a historical time point are obtained, and a change difference map of the green band pixels in the two phase images is generated. Then, in the change difference map, the area with enhanced green pixel values ​​and edge gradient amplitude change lower than the preset boundary change threshold is identified as the shrub spatial distribution area.

9. A system for dynamically evaluating the ecological health of riverbank vegetation, for implementing a method for dynamically evaluating the ecological health of riverbank vegetation according to any one of claims 1 to 8, characterized in that: It includes image preprocessing module, heterochromatic interference recognition module, boundary fuzzy assessment module, hierarchical extraction decision module and ecological display generation module; The image preprocessing module acquires image data containing the riverbank area and performs preprocessing on the image data, including image denoising, registration, and multi-channel alignment. It also extracts the riverbank area image in combination with terrain information and divides the riverbank area image into several image unit grids. The heterochromatic interference recognition module determines whether the shrub color is close to the background color in each image unit grid based on color and texture features, and marks the image unit grid with such a situation as the area to be recognized; The boundary fuzzy assessment module extracts the boundary recognition feature information of each area to be identified, analyzes it after extraction, evaluates the boundary recognition fuzziness of each area to be identified when the shrub color is close to the background color, and classifies it according to the assessment results; The hierarchical extraction decision module selects the corresponding shrub vegetation spatial distribution area extraction method for different categories of areas to be identified according to the classification results, so as to extract the shrub vegetation spatial distribution area in each area to be identified; The ecological display generation module performs image range statistics on the extracted shrub spatial distribution area and completes the output display of the ecological health status of the river bank area based on the statistical results.

Images