A feature analysis method for regional vegetation restoration

Abstract

The present application relates to the technical field of remote sensing image, and more particularly to a feature analysis method for regional vegetation restoration. The method comprises the following steps: obtaining multi-temporal remote sensing gray images; determining pixel gray difference values according to the multi-temporal remote sensing gray images; determining a set of vegetation candidate pixel points based on the pixel gray difference values; constructing vegetation candidate connected domains according to the set of vegetation candidate pixel points; identifying effective vegetation regions by using the vegetation candidate connected domains; collecting vegetation change data in the effective vegetation regions; determining vegetation restoration stability according to the vegetation change data; dividing vegetation restoration stages based on the vegetation restoration stability, and determining an evolution weight coefficient set for the vegetation restoration stages; and calculating vegetation restoration feature values according to the evolution weight coefficient set. The present application realizes phased quantitative evaluation of the vegetation restoration process based on remote sensing image technology, and effectively improves the reliability and evaluation accuracy of vegetation restoration grade determination.

Description

Technical Field

[0001] This invention relates to the field of remote sensing image technology, and in particular to a feature analysis method for regional vegetation restoration. Background Technology

[0002] Regional vegetation restoration assessment, as a key technical link in ecological and environmental monitoring, plays an important role in evaluating the effectiveness of ecological restoration, identifying degraded areas, and making resource management decisions. Due to the significant spatiotemporal dynamics, multi-stage evolutionary characteristics, and multi-factor coupling influences of the vegetation restoration process, existing technologies still have significant shortcomings in analyzing regional vegetation restoration characteristics: First, traditional methods are mostly based on static assessments using single-temporal remote sensing images or single indicators, lacking a continuous depiction of multi-temporal changes and failing to accurately reflect the dynamic evolutionary characteristics of vegetation restoration; second, existing technologies mostly focus on overall regional scale analysis, lacking refined modeling of vegetation spatial structure, making it difficult to effectively identify local restoration differences and spatial heterogeneity; third, existing restoration assessment methods typically do not divide the vegetation restoration process into stages, ignoring the characteristic differences under different evolutionary stages such as expansion, contraction, and splitting, resulting in assessment results lacking specificity and interpretability. Summary of the Invention

[0003] Therefore, it is necessary for the present invention to provide a characteristic analysis method for regional vegetation restoration in order to solve at least one of the above-mentioned technical problems.

[0004] To achieve the above objectives, a characteristic analysis method for regional vegetation restoration includes the following steps:

[0005] Step S1: Acquire multi-temporal remote sensing grayscale images; determine pixel grayscale difference values ​​based on multi-temporal remote sensing grayscale images; determine a set of candidate vegetation pixels based on pixel grayscale difference values;

[0006] Step S2: Construct candidate connected components for vegetation based on the candidate pixel set; identify effective vegetation areas using the candidate connected components; collect vegetation change data in the effective vegetation areas;

[0007] Step S3: Determine vegetation restoration stability based on vegetation change data; divide vegetation restoration into stages based on vegetation restoration stability, and determine the evolution weight coefficient set for each vegetation restoration stage;

[0008] Step S4: Calculate vegetation restoration characteristic values ​​based on the evolution weight coefficient set; assess the vegetation restoration level of the effective vegetation area based on the vegetation restoration characteristic values.

[0009] The beneficial effects of this invention are as follows:

[0010] 1. It realizes the step-by-step extraction from raw remote sensing data to vegetation spatial units. Compared with the traditional method of directly judging based on a single vegetation index, it can effectively reduce noise interference and the probability of misjudgment, improve the accuracy and spatial integrity of vegetation identification, and provide a reliable data foundation for subsequent restoration analysis.

[0011] 2. Differentiated features such as vegetation cover, leaf senescence, and patch fragmentation were introduced for the expansion, contraction, and splitting stages, respectively, enabling refined modeling of the vegetation restoration process. Compared with the overall assessment method that does not distinguish stages, it can more accurately reflect the ecological change characteristics under different evolutionary stages and improve the pertinence and interpretability of restoration analysis.

[0012] 3. It achieves dynamic weighting and comprehensive quantification of multi-stage features. Compared with the traditional fixed weight or simple superposition method, it can adaptively adjust the contribution ratio according to the importance of different stages, thereby more accurately assessing the degree of regional vegetation restoration and improving the reliability of assessment results and decision-making reference value. Attached Figure Description

[0013] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0014] Figure 1 This is a schematic diagram of the steps of a feature analysis method for regional vegetation restoration according to the present invention;

[0015] Figure 2 This is an example of vegetation tilt angle point cloud imaging;

[0016] Figure 3 This is an example of vegetation vertical angle point cloud imaging;

[0017] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] The technical method of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0019] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0020] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0021] To achieve the above objectives, please refer to Figures 1 to 3 This invention provides a characteristic analysis method for regional vegetation restoration, the method comprising the following steps:

[0022] Step S1: Acquire multi-temporal remote sensing grayscale images; determine pixel grayscale difference values ​​based on multi-temporal remote sensing grayscale images; determine a set of candidate vegetation pixels based on pixel grayscale difference values;

[0023] In one embodiment, grayscale image data of the same area at different times are acquired using remote sensing satellites (such as Sentinel-2 or Landsat 8) at 30-day intervals, for a total of six image data periods. Radiometric correction and geometric registration are then performed on the multi-temporal images. Using the first temporal image as a reference image, pixel-level alignment is performed on the remaining images, and the grayscale value change sequence of each pixel over time is extracted. For each pixel, the grayscale difference ΔG(t) = |G(t+1)−G(t)| between adjacent temporal phases is calculated, and all ΔG values ​​are used to construct a grayscale difference sequence. Statistical analysis is performed on the grayscale difference sequence to calculate its mean μ and standard deviation σ, and a threshold T = μ + 0.5σ is set. When the grayscale difference value of a pixel is greater than the threshold T, it is determined to be a candidate pixel for vegetation, thus constructing a set of candidate vegetation pixels.

[0024] In another embodiment, assume that four grayscale images of a certain region are acquired, with a resolution of 512×512 pixels and a total of approximately 260,000 pixels. The average grayscale difference is calculated. Standard deviation Then the threshold Statistical results show that approximately 32,000 pixels have grayscale differences greater than the threshold, accounting for 12.3% of the total pixels. These pixels were identified as candidate vegetation pixels and used for subsequent connected component construction.

[0025] Step S2: Construct candidate connected components for vegetation based on the candidate pixel set; identify effective vegetation areas using the candidate connected components; collect vegetation change data in the effective vegetation areas;

[0026] In one embodiment, an 8-neighborhood connectivity analysis is performed based on the candidate vegetation pixel set to merge adjacent candidate pixels point by point, constructing multiple connected regions. The area of ​​each connected region is calculated, and if the area of ​​a connected region is greater than a preset threshold (e.g., 100 pixels), the region is retained as a valid vegetation region. Within the valid vegetation region, a multi-temporal grayscale value change sequence for each pixel is extracted, and the average grayscale change trend and rate of change within the region are calculated, thereby forming vegetation change data.

[0027] In another embodiment, it is assumed that the candidate pixel set forms approximately 150 connected regions, of which 45 regions have an area greater than 100 pixels. The largest connected region has an area of ​​3200 pixels, and the smallest has an area of ​​120 pixels. After filtering these 45 regions, their time-series grayscale change data is extracted. For example, if the average grayscale value of a certain region increases from 80 to 110, the corresponding rate of change is +5 / time phase, which is used to characterize the vegetation growth trend.

[0028] Of particular importance, step S2, which involves constructing candidate connected components for vegetation based on the set of candidate pixels, includes:

[0029] Acquire target pixel data and determine the neighborhood range based on the target pixel data; within the neighborhood range, perform a neighborhood search based on the vegetation candidate pixel set and record the neighborhood search results;

[0030] In one embodiment, target pixel data is extracted from preprocessed remote sensing images or high-resolution images, the target pixel data including pixel coordinates. The target pixel and its corresponding spectral feature values ​​(e.g., NDVI or RGB values) are used. Based on the pixel's spatial location, a fixed window method or distance constraint method is used to determine the neighborhood range, for example, constructing a 3×3 or 5×5 neighborhood window centered on the target pixel. A neighborhood search is performed on the candidate pixel set for vegetation within this neighborhood range. The candidate pixel set is pre-obtained through threshold filtering (e.g., NDVI > 0.3). For each target pixel, all candidate pixels within its neighborhood are traversed, and the indices of neighboring pixels that satisfy similarity conditions (e.g., spectral difference less than threshold ε or color distance less than a set range) are recorded, forming a neighborhood search result set. .

[0031] In another embodiment, assuming 8000 candidate vegetation pixels are extracted from a 1024×1024 resolution image, a 5×5 neighborhood window (corresponding to an actual space of approximately 10m×10m) is constructed centered on each pixel. A spectral similarity threshold is set. During the neighborhood search process, each pixel can match an average of 6 to 12 neighboring candidate pixels. Statistical analysis shows that the maximum number of neighborhood connections in the neighborhood search results is 20, the minimum is 2, and the average is 8.5.

[0032] The search results for neighboring pixels determine adjacent pixel pairs; regional connectivity is established based on adjacent pixel pairs; regional merging is performed based on regional connectivity to form connected regions, and these connected regions are used as candidate connected domains for vegetation.

[0033] In one embodiment, based on the neighborhood search result set, neighbor pixel pairs are constructed for each target pixel and its neighboring pixels that meet the conditions. All adjacent pixel pairs are constructed into an undirected graph structure, where pixels are nodes and adjacency relationships are edges, thus forming a region connectivity graph. A connected component labeling algorithm (such as one based on disjoint-set data structure or depth-first search, DFS) is used to merge regions in this graph structure, dividing the set of pixels with connectivity relationships into several connected regions. Each connected region corresponds to a candidate vegetation connected component, and its area, shape, and boundary features are calculated for subsequent vegetation distribution analysis or change detection.

[0034] In another embodiment, it is assumed that approximately 32,000 pairs of adjacent pixels are constructed based on the aforementioned 8,000 candidate pixels, forming several connected subgraphs after constructing the graph structure. After merging regions using the disjoint-set data structure algorithm, a total of 120 connected regions are obtained. The largest connected region contains 950 pixels, the smallest contains 5 pixels, and the average size of the connected region is approximately 66 pixels. Statistically, there are 18 connected regions with more than 200 pixels, accounting for 62% of the total area; and 40 connected regions with less than 20 pixels, mostly scattered.

[0035] Step S3: Determine vegetation restoration stability based on vegetation change data; divide vegetation restoration into stages based on vegetation restoration stability, and determine the evolution weight coefficient set for each vegetation restoration stage;

[0036] In one embodiment, the area change rate is calculated based on the area change data of the effective vegetation area. The recovery phase is divided according to the range of the area change rate: when When it is determined to be in the expansion phase, When it is determined to be the contraction phase, when The time is determined to be a splitting stage. For different stages, corresponding features (coverage, aging degree, and fragmentation degree) are calculated, and evolution weight coefficients are assigned according to the duration and magnitude of change of each stage, finally forming a set of evolution weight coefficients.

[0037] In another embodiment, assuming that the area of ​​a certain region is 1000, 1200, 900, and 1300 pixels in four consecutive time phases, the corresponding area change rates are respectively... , , Therefore, the region is determined to be in the expansion phase, contraction phase, and splitting phase in sequence. Weights are assigned according to the duration and magnitude of change of each phase; for example, the expansion phase has a weight of 0.35, the contraction phase has a weight of 0.30, and the splitting phase has a weight of 0.35, forming an evolution weight coefficient set.

[0038] Step S4: Calculate vegetation restoration characteristic values ​​based on the evolution weight coefficient set; assess the vegetation restoration level of the effective vegetation area based on the vegetation restoration characteristic values.

[0039] In one embodiment, the evolution weight coefficient set is normalized so that the sum of the weights for each stage is 1. The recovery contribution value for each stage is calculated based on the vegetation recovery amount (e.g., cover change value, aging index, and fragmentation index). The recovery contribution values ​​for each stage are weighted and summed to obtain the cumulative recovery amount; the recovery change rate is calculated, and a recovery scoring function is constructed based on the recovery change rate to obtain the vegetation recovery characteristic value. The vegetation recovery level is evaluated according to a preset level classification rule (e.g., 0–0.3 for low recovery, 0.3–0.7 for medium recovery, and above 0.7 for high recovery).

[0040] In another embodiment, assuming the recovery amount during the expansion phase is 0.6, the recovery amount during the contraction phase is -0.4, and the recovery amount during the splitting phase is 0.3, with corresponding weights of 0.35, 0.30, and 0.35 respectively, the cumulative recovery amount is: The recovery rate of change was calculated to be 0.05 / phase, and the recovery characteristic value was obtained as 0.42 through the scoring function mapping, which corresponds to a medium recovery level.

[0041] It should be noted that you should refer to [link / reference]. Figure 2 The image shows a hilly terrain covered by a dense canopy, with vegetation in varying shades of green to represent the undulating height of the canopy. Light yellow lines represent the network of forest roads, and gray-white areas represent bare ground or sparse grass. Overall, it presents a typical three-dimensional spatial pattern of a mountain vegetation restoration area, providing a data foundation for canopy structure analysis, vegetation coverage estimation, and topographic measurement.

[0042] It should be noted that you should refer to [link / reference]. Figure 3The three-dimensional point cloud data of the vegetation area is presented from a near-two-dimensional map perspective using orthophoto projection, with RGB true color textures directly mapped onto the point cloud surface. Compared to tilted point clouds, the geometric distortion of ground features is smaller from a vertical perspective, and the spatial topological relationship between canopy cover and forest road network is clearer. In the image, dense green vegetation patches are distributed in contiguous areas, light yellow roads divide the woodland in a tree-like pattern, and the gray-white area on the right is a sparsely vegetated bare area.

[0043] Preferably, the step S1 of determining the pixel grayscale difference value based on the multi-temporal remote sensing grayscale image includes:

[0044] Select any one of the multi-temporal remote sensing grayscale images as a reference image; extract the first feature corner point in the reference image and extract the second feature corner point in the multi-temporal remote sensing grayscale image; perform matching on the first feature corner point and the second feature corner point, and record the feature corner point pair;

[0045] In one embodiment, an intermediate time phase (e.g., phase 3) is selected from the acquired multi-temporal remote sensing grayscale images as a reference image. Based on the reference image, the Harris corner detection algorithm is used to extract first feature corners, and stable corners are screened using non-maximum suppression. Simultaneously, the grayscale gradient distribution characteristics within the corner's neighborhood (e.g., a 7×7 window) are calculated as descriptors. In the remaining time-phase images, the FAST algorithm is used to extract second feature corners, and their corresponding BRIEF descriptors are calculated. The first feature corners are matched with the second feature corners in each time phase, using Hamming distance as the matching metric and setting a matching threshold (e.g., 0.3). High-confidence matching point pairs are selected and recorded as a set of feature corner pair sets.

[0046] In another embodiment, it is assumed that the second phase image is selected as the reference image, with a resolution of 1024×1024 pixels. Approximately 250 first feature corner points are extracted using the Harris algorithm, and approximately 300 second feature corner points are extracted from the remaining three phase images. After matching and filtering, approximately 120 pairs of valid feature corner points are retained between each phase and the reference image, totaling approximately 360 matching pairs. The matching accuracy is verified through random sampling, and the average matching error is controlled within 1.5 pixels.

[0047] Perform coordinate transformation on the feature corner point pairs to obtain transformed coordinate data; perform resampling based on the transformed coordinate data to obtain a registered grayscale image;

[0048] In one embodiment, using matched feature corner pairs, the least squares method is employed to estimate the affine transformation model parameters, including rotation, translation, and scaling parameters, thereby establishing a spatial mapping relationship from the image to be registered to the reference image. Based on this affine transformation model, the pixel coordinates in the image to be registered are transformed to obtain transformed coordinate data. Bilinear interpolation is then used to perform grayscale resampling on the transformed coordinates, thereby generating a registered grayscale image spatially aligned with the reference image.

[0049] In another embodiment, assuming that 100 pairs of feature points are matched between a certain image to be registered and a reference image, the affine transformation matrix parameters are obtained through least-squares fitting, where the rotation angle is approximately 2.5° and the translation amount is ( Pixels The scale change is 1.02 pixels. Based on this transformation parameter, the original image is mapped to coordinates and resampled using bilinear interpolation, finally obtaining a registered grayscale image with a mean registration error of 0.8 pixels.

[0050] Extract the position coordinates of each pixel in the registered grayscale image; determine the pixel grayscale value based on the position coordinates, and construct a multi-temporal grayscale value sequence using the pixel grayscale values; calculate the pixel grayscale difference value between adjacent time nodes based on the multi-temporal grayscale value sequence.

[0051] In one embodiment, a uniform coordinate grid is traversed across all registered grayscale images to extract the position coordinates of each pixel. For the same location coordinates, grayscale values ​​are extracted from the registered grayscale images at different times to construct a multi-temporal grayscale value sequence for that pixel. Perform a difference operation on the grayscale value sequence to calculate the grayscale difference between adjacent time points. This results in pixel-level grayscale variation information.

[0052] In another embodiment, assuming a pixel's grayscale values ​​at four time points are [85, 92, 110, 105], the corresponding grayscale difference values ​​are [7, 18, 5]. Across the entire image range (e.g., 512×512 pixels), approximately 260,000 pixel grayscale sequences and their difference values ​​are calculated. Statistical results show that the average grayscale difference value is 10.2, and the maximum is 35.6. Pixels with difference values ​​greater than 15 account for approximately 18% of the total, and these areas typically correspond to regions with significant vegetation changes.

[0053] Preferably, the step S1 of determining the vegetation candidate pixel set based on pixel grayscale difference values ​​includes:

[0054] Sort the pixel grayscale difference values ​​in descending order and record the grayscale difference values ​​of the pixels in descending order; calculate the mean grayscale difference value based on the grayscale difference value of the pixels in descending order and use the mean grayscale difference value as the grayscale threshold.

[0055] In one embodiment, assume a region contains 512×512 pixels, totaling approximately 260,000 pixels, with grayscale difference values ​​ranging from 0 to 40. After sorting all pixel grayscale difference values ​​in descending order, the average of the top 95% of the data is calculated to obtain the average grayscale difference. Compared to the untruncated mean of 14.5, it's clear that truncating reduced the impact of abnormally high values. The final selection... As a grayscale threshold.

[0056] When the pixel grayscale difference value is greater than the grayscale threshold, it is determined as a candidate pixel for vegetation; a set of candidate pixels for vegetation is constructed based on the candidate pixels for vegetation.

[0057] In one embodiment, the grayscale difference values ​​of all pixels are iterated and compared with a grayscale threshold. Compare the grayscale differences of a certain pixel. Greater than the threshold If the condition is met, the pixel is marked as a candidate pixel for vegetation; otherwise, it is marked as a non-candidate pixel. All pixels that meet the conditions are aggregated to construct a set of candidate pixels for vegetation.

[0058] In another embodiment, it is assumed that among the aforementioned 260,000 pixels, approximately 48,000 pixels have a grayscale difference value greater than the threshold of 13.2, accounting for approximately 18.5% of the total pixels. These pixels are marked as vegetation candidate pixels and spatially exhibit several clustered regions, with the largest clustering region containing approximately 3,500 pixels and the smallest approximately 120 pixels. Finally, all candidate pixels are aggregated to form a vegetation candidate pixel set.

[0059] Preferably, step S3, determining vegetation restoration stability based on vegetation change data, includes:

[0060] The area of ​​the vegetation region is calculated based on the vegetation change data, and the area change rate is calculated based on the area of ​​the vegetation region. When the area change rate is 0.15 to 0.30, it is determined to be the expansion stage; when the area change rate is -0.15 to -0.30, it is determined to be the fusion stage; when the area change rate is 0.30 to 0.40, it is determined to be the division stage.

[0061] In one embodiment, area statistics are performed on vegetation area data acquired at different time points, and the vegetation area at each time phase is calculated using either pixel counting or vector boundary integration. Calculate the rate of change of area based on area data at adjacent time points. The calculated area change rate is compared with a preset threshold range: when When the value is in the range of 0.15 to 0.30, this stage is determined to be the expansion stage; when... In When the interval is reached, it is determined to be in the fusion stage; when When the value is in the range of 0.30 to 0.40, it is determined to be in the splitting stage, thus realizing the stage division of the dynamic evolution process of vegetation.

[0062] In another embodiment, suppose the vegetation area of ​​a certain region at four consecutive time points is as follows: The corresponding area change rates are as follows: , , .in, If the value falls within the 0.15 to 0.30 range, it is considered to be in the expansion phase. Approaching the upper limit, it can be determined that it is in the transition stage from expansion to splitting; fall into The interval is determined to be in the fusion phase. Furthermore, if the area during a certain period is... Increase to ,but This can be identified as the splitting stage.

[0063] Vegetation cover is calculated during the expansion phase; leaf senescence is identified during the contraction phase; the degree of fragmentation of remotely sensed patches is assessed during the splitting phase; and the stability of vegetation recovery is evaluated based on vegetation cover, leaf senescence, and the degree of fragmentation of remotely sensed patches.

[0064] In one embodiment, assuming the analysis results for a certain area are as follows: During the expansion phase, the calculated vegetation cover is 0.65; during the contraction phase, the identified leaf senescence degree is 0.40 (range 0-1); and during the fragmentation phase, the calculated remote sensing patch fragmentation degree is 0.55. After normalizing the above parameters, and setting weights of 0.4, 0.3, and 0.3 respectively, the vegetation restoration stability index is: Set a stability grading threshold: when At this time, it is in a highly stable recovery state, when When it is in a moderately stable state, If the region is in a low-stability state, then it can be determined to be in a moderately stable recovery state.

[0065] Preferably, calculating vegetation cover during the expansion phase includes:

[0066] During the expansion phase, three-dimensional point cloud data of vegetation is collected; the stem axis direction is fitted based on the three-dimensional point cloud data of vegetation, and the vegetation growth tilt angle is determined based on the stem axis direction; the canopy horizontal projection coefficient is determined based on the vegetation growth tilt angle.

[0067] In one embodiment, during the expansion phase, the vegetation in the target area is scanned using lidar or structured light equipment to acquire three-dimensional point cloud data of the vegetation. After denoising and filtering the point cloud data, principal component analysis (PCA) is used to fit the orientation of individual or local vegetation point clouds, extracting the principal orientation vector of the point cloud as the stem axis orientation. The angle between the stem axis orientation and the ground normal vector is taken as the vegetation growth tilt angle. Based on growth tilt angle Construct the horizontal projection coefficient of the canopy It is used to characterize the degree of shrinkage of vegetation projected from an inclined state to a horizontal plane.

[0068] In another embodiment, it is assumed that approximately 50,000 points of vegetation point cloud data for a certain area are acquired through laser scanning, and after filtering, 42,000 valid point cloud points are retained. PCA analysis is performed on the individual vegetation point cloud to obtain the principal direction vector, with angles between the vector and the ground normal vector of [10°, 18°, 25°]. Based on this, the corresponding canopy horizontal projection coefficients are calculated as follows: [0.98, 0.95, 0.91]. Statistical results show that the overall vegetation tilt in this area is relatively small, and the canopy projection loss is low, which is beneficial for subsequent cover calculation.

[0069] Calculate the horizontal expansion area of ​​the canopy based on the horizontal projection coefficient of the canopy; obtain the area of ​​the land surface region; and use the ratio of the horizontal expansion area of ​​the canopy to the area of ​​the land surface region as the vegetation coverage.

[0070] In one embodiment, the original three-dimensional area of ​​the canopy is projected and corrected based on the horizontal projection coefficient of the canopy to calculate the horizontal expansion area of ​​the canopy. ,in This refers to the canopy surface area or envelope area obtained through point cloud reconstruction. Simultaneously, the surface area of ​​the corresponding region is acquired. (This can be obtained through region boundaries or grid statistics). Expand the canopy horizontally by area. With the surface area ratio Vegetation cover is used to characterize the degree of vegetation coverage on the earth's surface.

[0071] In another embodiment, it is assumed that the three-dimensional area of ​​the canopy is obtained by point cloud computing for a certain region. Corresponding average projection coefficient The horizontal expansion area of ​​the canopy Meanwhile, the surface area of ​​this region Then vegetation coverage Statistical analysis of multiple sub-regions yielded coverage values ​​of [0.62, 0.71, 0.78], with an average coverage of approximately 0.70, indicating that the region is at a relatively high coverage level.

[0072] Preferably, identifying leaf senescence during the shrinkage phase includes:

[0073] During the shrinkage stage, the red light band reflectance value and the near-infrared band reflectance value of the leaves were extracted; the spectral reflectance ratio was calculated based on the red light band reflectance value and the near-infrared band reflectance value of the leaves; and the chlorophyll sensitivity index was determined based on the spectral reflectance ratio.

[0074] In one embodiment, after multi-temporal remote sensing registration and vegetation region extraction have been completed, the remote sensing images corresponding to the shrinkage phase are decomposed into bands, selecting the red band (R, wavelength approximately 630–690 nm) and the near-infrared band (NIR, wavelength approximately 760–900 nm). The red reflectance value is then extracted for each vegetation pixel. Near-infrared reflectance And the spectral reflectance ratio is calculated based on the following ratio relationship: To enhance sensitivity to chlorophyll changes, the spectral reflectance ratio was normalized, and a chlorophyll sensitivity index was constructed using an empirical model. For all pixels Spatial statistics were performed to obtain the distribution of chlorophyll sensitivity index in this region. Typically, when... A decrease in the value indicates a decline in chlorophyll content, reflecting that the vegetation has entered a stage of decline or shrinkage.

[0075] In another embodiment, assuming 1000 effective vegetation pixels are selected within a certain contraction phase region, with red light reflectance values ​​ranging from 0.12 to 0.28 and near-infrared reflectance values ​​ranging from 0.35 to 0.62, the spectral reflectance ratio is calculated. The values ​​range from 1.8 to 3.5, with an average of 2.6. Calculate... The index ranges from 0.25 to 0.48, with an average of 0.36; approximately 65% ​​of the pixels... A value below 0.40 indicates an overall downward trend in chlorophyll content in the region. This is compared to the expansion phase. The average value is about 0.52, and it decreased by about 30% in this stage, which can clearly reflect the weakening of vegetation physiological activity.

[0076] Leaf water content was measured during the shrinkage stage; leaf epidermal tension was calculated based on leaf water content, and leaf wrinkling density was assessed based on leaf epidermal tension; leaf senescence was calculated based on leaf wrinkling density and chlorophyll sensitivity index.

[0077] In one embodiment, after obtaining the chlorophyll sensitivity index, the leaf water content is estimated by shortwave infrared (SWIR, approximately 1550–1750 nm) or thermal infrared retrieval. Based on the physical relationship between leaf water content and cell turgor pressure, an epidermal tension model was established. ,in This is an empirical coefficient (typically ranging from 0.8 to 1.2). Combining high-resolution image texture features (such as gray-level co-occurrence matrix contrast or local variance), the degree of wrinkling on the leaf surface is characterized, and wrinkling density is defined. (texture roughness, The lower the tension and the coarser the texture, the higher the wrinkle density. The chlorophyll sensitivity index... With shrinkage density We performed weighted fusion to construct a leaf senescence model:

[0078] in , Weighting coefficients (e.g.) , (This is used to comprehensively reflect the degree of physiological decline and morphological degradation of leaves).

[0079] In another embodiment, it is assumed that the leaf water content measured within a certain shrinkage region ranges from 0.42 to 0.68 (volume fraction), with an average value of 0.55; according to The calculated epidermal tension ranged from 0.42 to 0.68 MPa. Wrinkle density was obtained through texture analysis. The values ​​range from 0.30 to 0.72, with an average of 0.51, and approximately 40% of the areas have these values. It exhibits obvious wrinkling characteristics. Combined with... The average value is 0.36. , Calculate leaf senescence Its range is 0.48 to 0.78, with an average value of 0.62; among which The area accounting for approximately 45% can be identified as a moderate to severe aging area, while The area accounts for about 20%, and it still retains a certain degree of physiological activity.

[0080] Most importantly, the calculation of leaf senescence based on leaf wrinkling density and chlorophyll sensitivity index includes:

[0081] Assess the degree of leaf morphological change based on leaf wrinkling density; determine the leaf morphological attenuation value based on the degree of leaf morphological change; calculate the photosynthetic attenuation value based on the chlorophyll sensitivity index.

[0082] In one embodiment, the leaf wrinkling density parameter is first obtained. (Unit: pieces / mm²), and normalized based on a preset morphological mapping function, such as using a linear normalization model. To obtain the morphological change index Based on the degree of morphological change index Construct nonlinear decay functions, such as exponential decay models. ,in The morphological sensitivity coefficient (ranged from 0.8 to 1.5) is used to obtain the leaf morphological attenuation value. Simultaneously, the chlorophyll sensitivity index obtained from the previous steps is used... Construct a photosynthetic capacity decay function, for example ,in The spectral response adjustment coefficient (taken as 1.2~2.0) is used to obtain the photosynthetic attenuation value.

[0083] In another embodiment, it is assumed that leaf wrinkling density is collected in a certain area. The range is 0.8~3.2 particles / mm², take... , After normalization, we get The values ​​are [0.1, 0.35, 0.6, 0.8]; let... The corresponding morphological attenuation value Approximately [0.11, 0.34, 0.55, 0.70]. Meanwhile, assuming a chlorophyll sensitivity index... The value is [0.85, 0.72, 0.60, 0.48]. Let... Then the photosynthetic attenuation value can be calculated. The values ​​are approximately [0.22, 0.39, 0.54, 0.67]. These values ​​correspond to mild, moderate, and severe leaf wrinkling conditions, respectively, and can be used for subsequent grading analysis.

[0084] The leaf senescence coefficient is determined based on the leaf morphology senescence value, and the leaf photosynthetic coefficient is determined based on the photosynthetic senescence value. The leaf senescence coefficient and the leaf photosynthetic coefficient are normalized and weighted to output the leaf senescence degree.

[0085] In one embodiment, the blade morphology attenuation value is... Mapped to blade attenuation coefficient For example, using linear mapping Where γ is the structure weighting coefficient (taken as 0.6~0.9); and the photosynthetic attenuation value is also included. Mapped to leaf photosynthetic coefficient ,For example ,in This is the photosynthetic weighting coefficient (taken as 0.7~1.0). For and Perform normalization processing (such as Min-Max normalization) separately to obtain and And construct a comprehensive aging rate model: ,in (For example , ).

[0086] In another embodiment, it is assumed that the correspondence The values ​​are [0.11, 0.34, 0.55, 0.70]. Given [0.22, 0.39, 0.54, 0.67], take... , Then we get [0.088, 0.272, 0.440, 0.560] [0.198, 0.351, 0.486, 0.603]; after normalization, the following values ​​were obtained: [0.00, 0.33, 0.66, 1.00] [0.00, 0.31, 0.63, 1.00]. Let the weights be... , The final degree of leaf senescence [0.00, 0.32, 0.64, 1.00]. Here, a value close to 0 indicates a healthy leaf condition, and a value close to 1 indicates a highly aged condition, thus enabling a quantitative and graded assessment of the leaf aging process.

[0087] Preferably, the calculation of leaf epidermal tension based on leaf water content and the assessment of leaf wrinkling density based on leaf epidermal tension include:

[0088] Assess the leaf turgor pressure level based on leaf water content; determine the leaf force based on leaf turgor pressure level and calculate leaf skin tension based on leaf force; calculate leaf surface shrinkage amplitude based on leaf skin tension.

[0089] In one embodiment, after obtaining leaf water content data... Based on this, and according to the relationship between plant cell water potential and turgor pressure, a leaf turgor pressure model was constructed. ,in This is a proportionality coefficient (ranging from 0.9 to 1.1). This is a correction term (taken as 0.02–0.05 MPa). Based on the blade tissue mechanics model, the turgor pressure is converted into force per unit area. ,in The equivalent area of ​​cell action. An epidermal tension model is established based on the elastic properties of the leaf epidermis. ,in Characteristic length (e.g., interveinal spacing). The leaf surface shrinkage amplitude is defined based on the inverse relationship between epidermal tension and deformation. ,in For the reference tension of healthy leaves, This is the adjustment coefficient (taken as 0.5 to 0.8). When When the area increases, it indicates that the leaf surface has shrunk significantly, reflecting that the vegetation has entered a decline stage.

[0090] In another embodiment, assume the leaf water content within a certain shrinkage region The range is 0.40 to 0.65, with an average value of 0.52; [The last part is incomplete and requires further context.] , Then the turgor pressure Distributed in Between. Let the equivalent area of ​​the cell be... Then it is subject to force The range is 0.00086 to 0.00136 N. The characteristic length is taken. Epidermal tension was calculated. The range is 1.72–2.72 N / m. (Based on the tension of a healthy blade.) Based on, To obtain the contraction amplitude The value ranges from 0.056 to 0.256, with an average of approximately 0.18; about 50% of the area... This indicates that the leaves have undergone significant structural shrinkage.

[0091] The leaf shrinkage area is marked based on the leaf surface shrinkage amplitude; leaf wrinkling feature points in the leaf shrinkage area are collected; and leaf wrinkling density is calculated based on the wrinkling feature points.

[0092] In one embodiment, the shrinkage amplitude of the blade surface is obtained. Then, set the shrinkage determination threshold. (like ),when At that time, the corresponding pixels are marked as the leaf contraction region. Morphological processing (such as opening operation and connected component analysis) is performed on this region to extract continuous contraction patches. Wrinkle feature points are extracted within the contraction region based on high-resolution texture information, such as using local gradient magnitude, curvature change, or gray-level co-occurrence matrix contrast to screen high-frequency change points, forming a set of wrinkle feature points. Wrinkle density is defined based on the number of wrinkle feature points per unit area. ,in This represents the number of wrinkle feature points. The area of ​​the shrinkage region is used to quantify the degree of degradation of the blade surface structure.

[0093] In another embodiment, suppose there are 2000 pixels in a certain vegetated area, where the following conditions are met: The contracted region contains 850 pixels, accounting for 42.5%. After connected component analysis, 12 main contraction patches were identified, with a total area of ​​approximately 3400 mm². Approximately 680 wrinkling feature points were extracted from these contraction regions, resulting in a wrinkling density. Points / mm². Statistical analysis revealed that among them... The area of ​​high wrinkling per mm² accounts for about 30%, mainly distributed at the leaf margin and the area where leaf veins meet; The low-wrinkle area per mm² accounts for about 25%, and is mostly located in the center of the blade, still maintaining a certain degree of structural stability.

[0094] Preferably, assessing the degree of remote sensing patch fragmentation during the fragmentation phase includes:

[0095] During the splitting phase, remote sensing patch distribution data is calculated, and the number of remote sensing patches is counted based on the distribution data. The rate of change of remote sensing patches is calculated based on the number of patches. The area information of remote sensing patches is extracted from the distribution data.

[0096] In one embodiment, based on the completion of multi-temporal remote sensing image segmentation and vegetation region extraction, connected component analysis is performed on the vegetation regions in the segmentation stage to obtain remote sensing patch distribution data. Each patch corresponds to an independent connected region. Count the number of patches in the current time phase. and the number of patches in the previous time phase. Compare the results and calculate the plaque change rate. This is used to reflect the trend of vegetation evolution from continuous to discrete. Area information is extracted for each patch. (Obtained by multiplying the number of pixels by the spatial resolution), forming an area set. .

[0097] In another embodiment, it is assumed that during a certain regional splitting phase, the number of patches in the previous time phase was... The number of patches in this phase is Then the plaque change rate This indicates a significant increase in the number of patches. The area of ​​each patch ranged from 120 m² to 850 m², with an average area of ​​approximately 420 m². Patches smaller than 300 m² accounted for about 55%, indicating a clear trend towards fragmentation. Furthermore, compared to the average area of ​​approximately 680 m² in the previous time phase, the current area represents an overall decrease of about 38%, preliminarily reflecting a fragmentation in the spatial structure of the vegetation.

[0098] The average remote sensing patch area is calculated based on the remote sensing patch area information; the area reduction rate is calculated based on the average remote sensing patch area; the distance between remote sensing patches is calculated based on the remote sensing patch distribution data, and the spatial dispersion is calculated based on the distance between remote sensing patches; the degree of remote sensing patch fragmentation is determined based on the remote sensing patch change rate, area reduction rate, and spatial dispersion.

[0099] In one embodiment, based on area set Calculate the average remote sensing patch area and the average area of ​​the previous time phase By comparison, the area reduction rate is obtained. Based on the centroid coordinates of each patch Calculate the pairwise Euclidean distance between patches The spatial dispersion is defined by taking the mean or standard deviation of the sample as the spatial distribution characteristic parameter. This is used to describe the dispersion of the spatial distribution of patches. Finally, the patch variation rate is... Area reduction rate and spatial dispersion Weighted fusion is performed to construct a fragmentation index. (in , , (As weighting coefficients, such as 0.4, 0.3, 0.3), to quantitatively assess the degree of fragmentation of remote sensing patches.

[0100] In another embodiment, it is assumed that the average patch area at the current time phase is... And the previous period The area reduction rate The distance between the centroids of the patches was calculated, yielding a mean distance of 75m and a standard deviation of 28m. The spatial dispersion was then determined. Combining Take weight , , Calculate the fragmentation index 0.4×0.68+0.3×0.38²+0.3×0.37≈0.51. Based on a preset threshold (e.g.... The area was identified as highly fragmented; approximately 60% of the patches exhibited the characteristics of "small area + high spacing", indicating a significant decrease in vegetation connectivity and a marked impact on ecological stability.

[0101] Preferably, calculating the distance between remote sensing patches based on the remote sensing patch distribution data, and calculating the spatial dispersion based on the distance between remote sensing patches includes:

[0102] Extract the center coordinates of remote sensing patches from the distribution data of remote sensing patches; calculate the Euclidean distance using the center coordinates of remote sensing patches; and construct a set of remote sensing patch spacings based on the Euclidean distances.

[0103] In one embodiment, based on the preprocessed remote sensing image (resolution, for example, 10m), the patches are first labeled with connected components, and the geometric center coordinates of each remote sensing patch are extracted. ,in For any two patch center coordinates and Calculate its Euclidean distance And combine all distances to form a set of remote sensing patch spacings. In practical calculations, the neighborhood range can be limited (e.g., radius 500m), and only the patch spacing within the local neighborhood can be counted to reduce computational complexity and enhance the expression of spatial correlation.

[0104] In another embodiment, it is assumed that 50 remote sensing patches were identified within the study area, with their center coordinates distributed within a 1000m × 1000m region. 1225 sets of distance data were obtained through pairwise calculations, and 420 sets of effective distances within a 300m neighborhood radius were selected. The minimum distance in this set is 12m, the maximum distance is 285m, the average distance is 96m, and the standard deviation is 38m.

[0105] Record the nearest neighbor distance based on the remote sensing patch spacing set; calculate the distance deviation value using the nearest neighbor distance; determine the degree of distance fluctuation based on the distance deviation value; and evaluate the spatial dispersion based on the degree of distance fluctuation.

[0106] In one embodiment, for the spacing set Perform a traversal, targeting each remote sensing patch. The minimum value among the distances between it and other patches is selected to obtain the nearest neighbor distance. And form the nearest neighbor distance set. Calculate the average of the set. and standard deviation And use this to construct the distance deviation value, for example, define the deviation value. By statistically analyzing the mean or variance of all deviation values, a distance fluctuation index is obtained. (For example .when A larger value indicates significant differences in patch spacing and a relatively discrete spatial distribution; when... A smaller value indicates a more uniform and concentrated patch distribution. Ultimately, spatial dispersion is graded and evaluated based on the degree of distance fluctuation, for example, classified as low dispersion (…). Discrete ( ) ) and highly discrete ( ).

[0107] In another embodiment, it is assumed that the nearest neighbor distance set is calculated among the aforementioned 50 patches. Its numerical range is [10m, 85m], and the average value is... Standard deviation The degree of distance fluctuation was calculated. The distance deviations of each patch were statistically analyzed, with an average deviation of 14m and a maximum deviation of 43m. Based on the preset grading rules, this region was classified as having "moderate dispersion." Assuming another temporal data point... , ,but This corresponds to "high dispersion," indicating that the patch distribution tends to be fragmented and dispersed. Through the above multi-time comparisons, the dynamic evolution characteristics of the spatial structure of vegetation patches can be effectively characterized.

[0108] Preferably, step S4, calculating vegetation restoration characteristic values ​​based on the evolution weight coefficient set, includes:

[0109] Normalization is performed on the evolution weight coefficient set, and the normalization results are recorded; the contribution weight of the recovery stage is determined based on the normalization results; and the vegetation recovery amount at different stages is calculated based on the contribution weight of the recovery stage.

[0110] In one embodiment, the evolutionary weight coefficient set corresponding to different evolutionary stages of vegetation (e.g., expansion stage, stabilization stage, contraction stage, and division stage) is first obtained. The weight coefficients are derived from historical monitoring data or expert experience models. The set of weight coefficients is then normalized, for example using a linear normalization method. This ensures that the normalized weights satisfy... Thus, the normalization result is obtained. Based on the normalization results, the weights of each stage are mapped to the contribution weights of the recovery stages, which are used to characterize the degree of influence of each stage on the overall recovery process. This is combined with vegetation change indicators corresponding to each stage (such as changes in vegetation cover). chlorophyll recovery or changes in biomass ), calculate vegetation restoration at different stages (or other representative quantities), thereby forming a set of phased recovery quantities. This provides basic data for subsequent cumulative recovery analysis.

[0111] In another embodiment, it is assumed that there are four evolutionary stages, with their original weight coefficients being respectively... After normalization, the following was obtained Assume the changes in vegetation cover at each stage are as follows: Then the vegetation restoration amount at each stage can be calculated. The splitting phase contributed the most (7.7%), while the contraction phase contributed the least (0.9%).

[0112] The cumulative restoration amount is calculated based on the vegetation restoration amount at different stages; the restoration change rate is calculated based on the cumulative restoration amount; and a vegetation restoration score is performed based on the restoration change rate to record the vegetation restoration characteristic values.

[0113] In one embodiment, the vegetation restoration amounts at each stage are summed to obtain the cumulative restoration amount. This is used to reflect the overall recovery level. It records the cumulative recovery amount at different time points based on time series (e.g., monthly or quarterly) and calculates the rate of recovery change, for example, defined as... This is used to characterize the dynamic trend of the restoration process. Based on this, a vegetation restoration scoring model is constructed, which can be in linear or piecewise function form, mapping the rate of restoration change to a score value. (e.g., the range of 0 to 100 points) to obtain vegetation restoration characteristic values.

[0114] In another embodiment, it is assumed that the cumulative recovery amount of the above four stages is obtained. And at three consecutive time points ( , , The cumulative recovery rates were recorded as 10.5%, 13.2%, and 16.0% respectively, with a time interval of one month. The corresponding recovery rates were respectively... moon, The average recovery rate per month is approximately Month. Set the scoring function. ,in Taking 0.8, the corresponding score is approximately A score of [score missing] indicates that vegetation recovery in the area is at a high level. If, in another hypothetical scenario, the recovery rate decreases to [score missing], [then the area would be considered unsuitable]. In the month, the score is approximately 55 points, reflecting a slowdown in the recovery process.

[0115] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0116] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A method for analyzing the characteristics of regional vegetation restoration, characterized in that, Includes the following steps: Step S1: Acquire multi-temporal remote sensing grayscale images; determine pixel grayscale difference values ​​based on multi-temporal remote sensing grayscale images; Determine the candidate pixel set for vegetation based on pixel grayscale difference values; Step S2: Construct candidate connected components for vegetation based on the candidate pixel set; identify effective vegetation areas using the candidate connected components; collect vegetation change data in the effective vegetation areas; Step S3: Determine the stability of vegetation restoration based on vegetation change data; Based on the stability of vegetation restoration, vegetation restoration stages are divided, and the set of evolution weight coefficients for each vegetation restoration stage is determined. Step S4: Calculate vegetation restoration characteristic values ​​based on the evolution weight coefficient set; The vegetation restoration level of an effective vegetation area is assessed based on vegetation restoration characteristic values.

2. The characteristic analysis method for regional vegetation restoration according to claim 1, characterized in that, Step S1, which determines the pixel grayscale difference value based on the multi-temporal remote sensing grayscale image, includes: Select any one of the multi-temporal remote sensing grayscale images as a reference image; extract the first feature corner point in the reference image and extract the second feature corner point in the multi-temporal remote sensing grayscale image; perform matching on the first feature corner point and the second feature corner point, and record the feature corner point pair; Perform coordinate transformation on the feature corner point pairs to obtain transformed coordinate data; perform resampling based on the transformed coordinate data to obtain a registered grayscale image; Extract the position coordinates of each pixel in the registered grayscale image; determine the pixel grayscale value based on the position coordinates, and construct a multi-temporal grayscale value sequence using the pixel grayscale values; calculate the pixel grayscale difference value between adjacent time nodes based on the multi-temporal grayscale value sequence.

3. The characteristic analysis method for regional vegetation restoration according to claim 1, characterized in that, Step S1, which determines the candidate set of vegetation pixels based on pixel grayscale difference values, includes: Sort the pixel grayscale difference values ​​in descending order and record the grayscale difference values ​​of the pixels in descending order; calculate the mean grayscale difference value based on the grayscale difference value of the pixels in descending order and use the mean grayscale difference value as the grayscale threshold. When the pixel grayscale difference value is greater than the grayscale threshold, it is determined as a candidate pixel for vegetation; a set of candidate pixels for vegetation is constructed based on the candidate pixels for vegetation.

4. The characteristic analysis method for regional vegetation restoration according to claim 1, characterized in that, Step S3, determining vegetation restoration stability based on vegetation change data, includes: The area of ​​the vegetation region is calculated based on the vegetation change data, and the area change rate is calculated based on the area of ​​the vegetation region. When the area change rate is 0.15 to 0.30, it is determined to be the expansion stage; when the area change rate is -0.15 to -0.30, it is determined to be the fusion stage; when the area change rate is 0.30 to 0.40, it is determined to be the division stage. Vegetation cover is calculated during the expansion phase; leaf senescence is identified during the contraction phase; the degree of fragmentation of remotely sensed patches is assessed during the splitting phase; and the stability of vegetation recovery is evaluated based on vegetation cover, leaf senescence, and the degree of fragmentation of remotely sensed patches.

5. The characteristic analysis method for regional vegetation restoration according to claim 4, characterized in that, Calculating vegetation cover during the expansion phase includes: During the expansion phase, three-dimensional point cloud data of vegetation is collected; the stem axis direction is fitted based on the three-dimensional point cloud data of vegetation, and the vegetation growth tilt angle is determined based on the stem axis direction; the canopy horizontal projection coefficient is determined based on the vegetation growth tilt angle. Calculate the horizontal expansion area of ​​the canopy based on the horizontal projection coefficient of the canopy; obtain the area of ​​the land surface region; and use the ratio of the horizontal expansion area of ​​the canopy to the area of ​​the land surface region as the vegetation coverage.

6. The characteristic analysis method for regional vegetation restoration according to claim 4, characterized in that, Identifying leaf senescence during the contraction phase includes: During the shrinkage stage, the red light band reflectance value and the near-infrared band reflectance value of the leaves were extracted; the spectral reflectance ratio was calculated based on the red light band reflectance value and the near-infrared band reflectance value of the leaves; and the chlorophyll sensitivity index was determined based on the spectral reflectance ratio. Leaf water content was measured during the shrinkage stage; leaf epidermal tension was calculated based on leaf water content, and leaf wrinkling density was assessed based on leaf epidermal tension; leaf senescence was calculated based on leaf wrinkling density and chlorophyll sensitivity index.

7. The characteristic analysis method for regional vegetation restoration according to claim 6, characterized in that, Calculate leaf epidermal tension based on leaf water content, and assess leaf wrinkling density based on leaf epidermal tension, including: Assess the leaf turgor pressure level based on leaf water content; determine the leaf force based on leaf turgor pressure level and calculate leaf skin tension based on leaf force; calculate leaf surface shrinkage amplitude based on leaf skin tension. The leaf shrinkage area is marked based on the leaf surface shrinkage amplitude; leaf wrinkling feature points in the leaf shrinkage area are collected; and leaf wrinkling density is calculated based on the wrinkling feature points.

8. The characteristic analysis method for regional vegetation restoration according to claim 4, characterized in that, Assessing the degree of fragmentation of remote sensing patches during the fragmentation phase includes: During the splitting phase, remote sensing patch distribution data is calculated, and the number of remote sensing patches is counted based on the distribution data. The rate of change of remote sensing patches is calculated based on the number of patches. The area information of remote sensing patches is extracted from the distribution data. The average remote sensing patch area is calculated based on the remote sensing patch area information; the area reduction rate is calculated based on the average remote sensing patch area; the distance between remote sensing patches is calculated based on the remote sensing patch distribution data, and the spatial dispersion is calculated based on the distance between remote sensing patches; the degree of remote sensing patch fragmentation is determined based on the remote sensing patch change rate, area reduction rate, and spatial dispersion.

9. The characteristic analysis method for regional vegetation restoration according to claim 8, characterized in that, The distances between remote sensing patches are calculated based on the distribution data of remote sensing patches, and the spatial dispersion is calculated based on the distances between remote sensing patches, including: Extract the center coordinates of remote sensing patches from the distribution data of remote sensing patches; calculate the Euclidean distance using the center coordinates of remote sensing patches; and construct a set of remote sensing patch spacings based on the Euclidean distances. Record the nearest neighbor distance based on the remote sensing patch spacing set; calculate the distance deviation value using the nearest neighbor distance; determine the degree of distance fluctuation based on the distance deviation value; and evaluate the spatial dispersion based on the degree of distance fluctuation.

10. The characteristic analysis method for regional vegetation restoration according to claim 1, characterized in that, Step S4, which calculates vegetation restoration characteristic values ​​based on the evolutionary weight coefficient set, includes: Normalization is performed on the evolution weight coefficient set, and the normalization results are recorded; the contribution weight of the recovery stage is determined based on the normalization results; and the vegetation recovery amount at different stages is calculated based on the contribution weight of the recovery stage. The cumulative restoration amount is calculated based on the vegetation restoration amount at different stages; the restoration change rate is calculated based on the cumulative restoration amount; and a vegetation restoration score is performed based on the restoration change rate to record the vegetation restoration characteristic values.

Images