A method for intelligent extraction of forest gap parameters and quantitative analysis of structural characteristics thereof

Abstract

A method for intelligent extraction of forest gap parameters and quantitative analysis of structural characteristics, comprising the following steps: based on the CHM generated by the airborne laser radar data, the forest gap area pixels are segmented through the height threshold setting based on the percentile, and the unsupervised extraction of the forest gap is realized by combining the area connectivity judgment; then the multi-dimensional forest gap structural characteristics of the forest gap area, shape, spatial distribution, network topology structure and ecological connectivity are fused through the modular parameter extraction technology flow. It does not rely on multispectral features or deep learning segmentation models, but fully excavates the spatial geometric features reflecting the crown distribution in the CHM image to realize the full-automatic and low-power structured parameter extraction. It makes up for the defects of insufficient automation level, low precision, small application scope and insufficient multi-dimensional parameter extraction and analysis in the current forest gap structure extraction, and its precision fully meets the application requirements of primary forestry.

Description

Technical Field

[0001] The invention belongs to the technical field of forest remote sensing and forest structure parameter extraction, and relates to a method for intelligent extraction of forest gap parameters and quantitative analysis of their structural characteristics. Background Art

[0002] Canopy gaps, a crucial component of forest canopy structure, are a key factor contributing to stand spatial heterogeneity and light environment complexity, influencing key ecological processes such as forest community dynamics, species diversity, carbon flux, and regeneration mechanisms. The size, morphology, and spatial distribution of canopy gaps are closely related to their ecological impacts. Therefore, the precise and efficient identification, extraction, and parameterized description of these gaps are crucial for improving forestry management decision-making and the accuracy of ecological process modeling.

[0003] Traditional methods rely heavily on field observations, ground-based analysis, transect analysis, and extraction methods based on visual selection from aerial photographs and stereoscopes. These extraction strategies suffer from low automation, poor accuracy in forest gap extraction, and a lack of image-driven pixel-level analysis mechanisms. Previous methods are only suitable for extracting basic metrics such as gap area, proportion, and number within small forest areas, making it difficult to systematically assess their geometric complexity, spatial distribution patterns, spatial topological connectivity, and overall ecological connectivity. Furthermore, most methods suffer from fragmented processing flows and a single indicator system, making them difficult to adapt to the modern forestry industry's demand for multi-dimensional, automated, and batch-based forest gap structure extraction. Summary of the Invention

[0004] In response to the existing technical problems, the present invention provides a method for intelligently extracting forest gap parameters and quantitatively analyzing their structural characteristics.

[0005] A method for intelligently extracting forest gap parameters and quantitatively analyzing their structural characteristics includes the following steps: based on the CHM generated by airborne lidar data, the forest gap area pixels are segmented by setting a height threshold based on percentiles, and combined with regional connectivity judgment to achieve unsupervised extraction of forest windows; then, through a modular parameter extraction technology flow, the multi-dimensional forest gap structural characteristics of forest window area, shape, spatial distribution, network topology and ecological connectivity are fused and extracted.

[0006] The advantages of the present application are: to realize the standardized, automated processing and visualization of the whole process from CHM data input to forest gap structure identification and parameter output, to provide a new technical support for efficient and accurate extraction of regional scale forest gap structure information, to realize efficient and accurate identification and spatial significance segmentation of forest gap structure (forest gap) through the input of canopy height model (C), and to further realize systematic extraction and quantitative expression of multi-dimensional parameters such as forest gap geometric characteristics, shape complexity, spatial pattern and ecological connectivity through modular parameter extraction analysis. The method does not rely on multi-spectral features or deep learning segmentation model, but fully excavates the spatial geometric features reflecting the tree crown distribution in the CHM image, realizes the automatic and low-power structured parameter extraction. It makes up for the defects of insufficient automation level, low precision, small application scope and insufficient multi-dimensional parameter extraction and analysis in the current forest gap structure extraction, and its precision fully meets the application requirements of basic forestry.

[0007] The present application realizes the whole automatic and standardized processing process from forest gap structure identification, labeling to parameter output and visualization by importing the canopy height model (C), which significantly improves the efficiency and accuracy of forest gap parameter extraction. Through the spatial significance segmentation of gap structure, the key gaps in the forest stand can be accurately identified, and the spatial distribution and morphological characteristics of forest gap can be accurately evaluated, thereby improving the accuracy and reliability of forest gap extraction. In addition, the present application can also comprehensively quantitatively express multi-dimensional parameters such as shape complexity, spatial pattern and ecological connectivity of forest gap. Unlike traditional methods, this method does not rely on multi-spectral image features or deep learning segmentation model, but fully excavates the spatial geometric features of tree crown distribution in the CHM image, simplifies the data processing process, and reduces the demand for high-performance computing resources. BRIEF DESCRIPTION OF DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor. As shown in the drawings:

[0009] Figure 1 The flowchart of the present application.

[0010] Figure 2 (a) is a schematic diagram of the test area of Shangxia Experimental Forest Farm in Fenyi County, Jiangxi Province, and the green arrow indicates the position for forest gap extraction. The right side is the corresponding detailed area display.

[0011] Figure 2 (b) is a schematic diagram of the test area of the large sample plot of arbor forest in Sigou of Qilian Mountain National Park in Qinghai Province, and the green arrow indicates the position for forest gap extraction. The right side is the corresponding detailed area display.

[0012] Figure 3(a) shows the pixel points of the forest gap in the mixed coniferous and broadleaf forest.

[0013] Figure 3(b) shows the pixel points of the forest gap in the pure coniferous forest.

[0014] Figure 3(c) shows the results of the forest gap area extraction in the mixed coniferous and broadleaf forest.

[0015] Figure 3(d) shows the results of the forest gap area extraction in the pure coniferous forest.

[0016] Figure 4(a) shows the distribution of the shape index of the forest gap in the mixed coniferous and broadleaf forest.

[0017] Figure 4(b) shows the distribution of the circularity of the forest gap in the mixed coniferous and broadleaf forest.

[0018] Figure 4(c) shows the distribution of the fractal dimension of the forest gap in the mixed coniferous and broadleaf forest.

[0019] Figure 4(d) shows the distribution of the shape index of the forest gap in the pure coniferous forest.

[0020] Figure 4(e) shows the distribution of the circularity of the forest gap in the pure coniferous forest.

[0021] Figure 4(f) shows the distribution of the fractal dimension of the forest gap in the pure coniferous forest.

[0022] Figure 5(a) shows the spatial distribution of the nearest neighbor distance of the forest gap centroid in the pure coniferous forest.

[0023] Figure 5(b) shows the histogram of the nearest neighbor distance of the forest gap centroid in the pure coniferous forest.

[0024] Figure 5(c) shows the spatial distribution of the nearest neighbor distance of the forest gap centroid in the mixed coniferous and broadleaf forest.

[0025] Figure 5(d) shows the histogram of the nearest neighbor distance of the forest gap centroid in the mixed coniferous and broadleaf forest.

[0026] Figure 6(a) shows the Ripley's L function in the mixed coniferous and broadleaf forest.

[0027] Figure 6(b) shows the Ripley's L function in the pure coniferous forest.

[0028] Figure 6(c) shows the Ripley's K function in the mixed coniferous and broadleaf forest.

[0029] Figure 6(d) shows the Ripley's K function in the pure coniferous forest.

[0030] Figure 7IIC weighted network graph (spatial distribution map of ecological connectivity of needle-broad mixed forest gap, each point represents the position of the gap, and the size of the point indicates the size of the ecological connectivity).

[0031] Figure 8 IIC weighted network graph (spatial distribution map of ecological connectivity of needle-broad mixed forest gap, each point represents the position of the gap, and the size of the point indicates the size of the ecological connectivity). DETAILED DESCRIPTION

[0032] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0033] Embodiment 1: as shown in Figure 1 , Fig. 2(a), Fig. 2(b), Fig. 3(a), Fig. 3(b), Fig. 3(c), Fig. 3(d), Fig. 4(a), Fig. 4(b), Fig. 4(c), Fig. 4(d), Fig. 4(e), Fig. 4(f), Fig. 5(a), Fig. 5(b), Fig. 5(c), Fig. 5(d), Fig. 6(a), Fig. 6(b), Fig. 6(c), Fig. 6(d), Figure 7 and Figure 8 A method for intelligent extraction of gap parameters and quantitative analysis of structural characteristics, related to forest gap structure recognition and quantitative parameter automatic extraction and analysis technology based on canopy height model (C), includes two aspects: one is to realize the accurate identification of the gap area through the preset of canopy height threshold and 8-neighbor connectivity strategy; the other is to design a modular process to efficiently extract the multi-dimensional structural characteristics of the gap area, shape, spatial distribution, network topology and ecological connectivity.

[0034] As shown in Figure 1 A method for intelligent extraction of gap parameters and quantitative analysis of structural characteristics, comprising the following steps:

[0035] Step 1, input of normalized canopy height model,

[0036] Step 2, gap structure extraction, including three parameter setting steps of area, pixel parameter and height threshold, elevation pixel two-dimensional array reading step, threshold segmentation (percentile) gap pixel extraction step, 8-neighbor connectivity gap pixel classification step, gap area labeling step with marking and numbering,

[0037] Step 3: Modular extraction and output of structural parameters, including basic indicator steps: area extraction, perimeter extraction, total number of forest windows, forest window density and forest gap ratio; shape characteristic indicator steps: fractal dimension, circularity and shape index; spatial distance structure steps: nearest forest window distance, maximum nearest neighbor distance and minimum nearest neighbor distance; Ripley's K function value step for spatial aggregation; and overall connectivity index step for ecological connectivity.

[0038] The specific steps are as follows:

[0039] Step 1: Extract forest gap structure using canopy height threshold segmentation and 8-neighborhood connectivity strategy.

[0040] (1) Canopy height model (CHM) image reading and parameter setting.

[0041] The image reading library (rasterio) was used to read .tif format CHM images and generate the corresponding two-dimensional floating-point elevation matrix. Invalid pixels were masked to prevent error propagation in subsequent analyses. To improve the accuracy of forest gap structure extraction and the method's wide applicability and adjustability across different forest types, the algorithm allows for autonomous parameter adjustments based on forest type and terrain. These include a percentile-based threshold for forest gap height (PERCENTILE), grid resolution (PIXEL_SIZE), region area (REGION_AREA), maximum analysis scale (MAX_DISTANCE), and step size (DIST_STEP). The latter two parameters serve as parameters for adjusting the Ripley's K function in spatial pattern analysis. This parameter adjustment capability enables this forest gap structure extraction method to have excellent generalization and expansion potential.

[0042] (2) Forest gap mask generation and connectivity identification.

[0043] Based on tree species canopy height characteristics or regional forest type statistics, a forest window height threshold is set in the parameter settings section. Pixels in the CHM image with heights below this threshold are marked as potential forest windows, generating a binary mask image. Eight-neighborhood connectivity analysis is then used to extract connected components from low-elevation areas, removing isolated noisy pixels to ensure spatial continuity of the identified forest windows. Finally, the extracted forest window areas are uniquely numbered, and a structural attribute table is generated for each forest window area. A number-attribute mapping relationship is established, and a regional mask map is created to provide a spatial reference for subsequent structural quantification, graph analysis, and visualization.

[0044] Step 2: Modular extraction of multi-dimensional forest window structural parameters based on forest window pixel area division.

[0045] (1) Forest gap basic indicator extraction module.

[0046] The area of the gap is obtained by calculating the total number of pixels and the pixel resolution, and the perimeter is obtained by extracting the outer contour length of the gap region according to the boundary pixels of the gap. The total number of gaps is directly counted by the number of connected regions identified in the image, and the gap density is calculated by dividing the total number of gaps by the total area of the region. The gap rate is defined as the ratio of the total area of the gap to the total area of the region.

[0047] (6) Gap shape feature index extraction module.

[0048] The shape index (SIi) is used to measure the closeness of the gap region to an ideal circle. The value tends to 1, and the shape is closer to the standard circle. The calculation formula is as follows:

[0049]

[0050] In the formula, A i represents the area size of the i-th gap, P i represents the perimeter of the i-th gap.

[0051] The circularity index (C i ) is used to measure the closeness of the gap region to an ideal circle. The value tends to 1, and the shape is closer to the standard circle. The calculation formula is as follows:

[0052]

[0053] In the formula, A i represents the area size of the i-th gap, P i represents the perimeter of the i-th gap.

[0054] The fractal dimension (FD i ) is used to calculate the complexity of the gap boundary. The higher the value, the more irregular the gap boundary. The calculation formula is as follows:

[0055]

[0056] In the formula, P i represents the perimeter of the i-th gap, and A i represents the corresponding area.

[0057] (7) Gap spatial distance structure index extraction module.

[0058] The centroid (centroid (row，col) ) of all gap regions is used as the object of spatial analysis, and the Euclidean distance matrix (D ij ) is constructed, and the distance of each gap to its nearest neighbor is extracted The maximum nearest neighbor distance (d max = max d i ) and the minimum nearest neighbor distance (d max= min d i These indicators can effectively evaluate the tightness of the distribution between forest gaps, and identify the aggregated, discrete or extremely isolated forest gap structure. In addition, the centroid coordinates of the forest gap area are calculated as follows:

[0059]

[0060] In the formula, r i , c i represents the sum of the row and column coordinates of each forest gap pixel in the area, and N represents the total number of pixels in the area.

[0061] (8) Forest gap spatial aggregation index extraction module.

[0062] Point pattern evaluation at different scales is performed using the set of forest gap centroid points. This method gradually increases the analysis radius, calculates the spatial relationship density between forest gaps at each scale, and takes the K function value at the largest scale as the quantitative indicator of the degree of aggregation.

[0063]

[0064] In the formula, A represents the area of the region, n represents the total number of forest gaps, I represents the indicator function, which takes 1 when d ij ≤r, otherwise 0, d ij is the Euclidean distance between the i-th and j-th forest gap centroid, and r is the analysis scale radius.

[0065] In order to enhance the contrast, intuitiveness and linearity of interpretation, the K function is converted to L function form as follows:

[0066]

[0067] In the formula, when L (r) >0, it indicates that the forest gap shows aggregated distribution at scale r; when L (r) <0, it indicates uniform distribution; if L (r) ≈0, it means that the forest gap centroid is close to completely random distribution.

[0068] (9) Forest gap overall ecological connectivity index extraction module.

[0069] The overall connectivity index (IIC) in landscape ecology is used to measure the ecological accessibility between forest gaps. The larger the index value, the higher the connectivity between forest gaps; the smaller the value, the greater the discrete distribution of forest gaps. The calculation formula is as follows:

[0070]

[0071] In the formula, a i , a jArea of two forest gaps, respectively, d ij Euclidean distance of two.

[0072] Example 2: as Figure 1 , Figure 2(a), Figure 2(b), Figure 3(a), Figure 3(b), Figure 3(c), Figure 3(d), Figure 4(a), Figure 4(b), Figure 4(c), Figure 4(d), Figure 4(e), Figure 4(f), Figure 5(a), Figure 5(b), Figure 5(c), Figure 5(d), Figure 6(a), Figure 6(b), Figure 6(c), Figure 6(d), Figure 7 And Figure 8 As shown in Figure 2(a), Figure 2(b), Figure 3(a), Figure 3(b), Figure 3(c), Figure 3(d), Figure 4(a), Figure 4(b), Figure 4(c), Figure 4(d), Figure 4(e), Figure 4(f), Figure 5(a), Figure 5(b), Figure 5(c), Figure 5(d), Figure 6(a), Figure 6(b), Figure 6(c), Figure 6(d), a kind of forest gap parameter intelligent extraction and its structural feature quantitative analysis method, two forest types forest gap structure extraction examples of Jiangxi Shannan Experimental Forest Farm and Qinghai Qilian Mountain National Forest Park.

[0073] 1. Data collection

[0074] The airborne laser radar platform (drone: DJI M350; lens: Chan Si L2+RTK) is used to obtain the airborne radar data (ground simulation 130 meters; air speed 10 meters / second; 5 echoes) of the local area of Jiangxi Shannan Experimental Forest Farm and the Qingsonggou Qinghai spruce tree forest of Qinghai Qilian Mountain National Forest Park. The two types of forest have typical differences, the former belongs to subtropical evergreen coniferous and broad-leaved mixed forest, and the latter belongs to Qinghai spruce pure forest in alpine mountain climate area. After the pretreatment process, the canopy height model (C) of the two test areas is extracted. As shown in Figure 2(a), Figure 2(b): test area schematic diagram, green arrow indicates the position for forest gap extraction, and the right side is the corresponding detailed area display, 2. Forest gap area extraction and marking

[0075] The CHM data of the two types of forest is imported into the algorithm respectively, and the parameters are set as follows:

[0076] (1) Coniferous and broad-leaved mixed forest (361m×267m)

[0077] PERCENTILE=10% (extract forest gap height threshold percentile)

[0078] REGION_AREA=361*267 (unit m 2 )

[0079] PIXEL_SIZE=1.0 (m / pixel)

[0080] MAX_DISTANCE=90 (m)

[0081] DIST_STEP=1 (m)

[0082] (2) Coniferous pure forest (240m×240m)

[0083] PERCENTILE = 15% (extraction of forest gap height threshold percentile)

[0084] REGION_AREA = 257 * 257 (unit m 2 )

[0085] PIXEL_SIZE = 1.0 (m / pixel)

[0086] MAX_DISTANCE = 80 (m)

[0087] DIST_STEP = 1 (m)

[0088] The algorithm first marks the pixels with a height lower than the threshold as potential forest gaps, then extracts the forest gap region based on 8-neighbor connectivity, numbers each region and records its centroid and boundary, and finally outputs a visualization image and establishes a mapping between the forest gap number and attributes (area, shape, spatial coordinates, etc.). The forest gap extraction is shown in FIG. 3(c) and FIG. 3(d), and the extraction effect can be analyzed by comparing with FIG. 3(a) and FIG. 3(b). The extraction of forest gap regions for two types of stands is shown in FIG. 3(c) and FIG. 3(d), with yellow lines representing the boundaries of forest gaps and red dots representing the positions of the centroids of forest gaps.

[0089] 3. Modular extraction of forest gap structure parameters

[0090] Based on the above forest gap calibration and number-attribute relationship establishment, the forest gap structure parameter extraction process is further modularized, and independent functional modules are constructed for different types of forest gap structure characteristics to improve computational efficiency, enhance algorithm adaptability, and meet the needs of multi-dimensional structure information extraction. The overall parameter extraction is divided into the following sub-modules:

[0091] (1) Basic structure index extraction module

[0092] The area, perimeter, number of forest gaps, forest gap density, and forest gap rate of each forest gap are automatically calculated. The area and perimeter are calculated based on the number of pixels and the length of the boundary contour, respectively; the number of forest gaps is obtained by counting the connected domains; and the forest gap density and forest gap rate are normalized expressions of the number and area of forest gaps in the region, respectively. The parameter extraction results are shown in Table 1 and Table 2.

[0093] (2) Shape feature extraction module

[0094] The shape index, circularity, and fractal dimension of each forest gap are calculated to quantify the regularity of its geometric shape and the complexity of its boundary. The closer the circularity and shape index are to 1, the closer the forest gap shape is to a regular circle; the higher the fractal dimension, the higher the boundary complexity, representing the spatial heterogeneity of the forest gap. The automatic extraction and visualization of the parameters are shown in FIG. 4(a), FIG. 4(b), FIG. 4(c), FIG. 4(d), FIG. 4(e), FIG. 4(f), and Table 1 and Table 2.

[0095] (3) Spatial distance structure extraction module

[0096] Based on the forest gap centroid coordinates, the Euclidean distance between the nearest neighbor forest gap is calculated, the average, maximum and minimum nearest neighbor distance is counted, and the density and spatial isolation of the forest gap distribution are evaluated. The parameter automatic extraction and visual analysis results are shown in Figures 5(a), 5(b), 5(c) and 5(d) and Tables 1 and 2.

[0097] (4) Spatial aggregation evaluation module

[0098] The following are the evaluation results of Ripley's K function and L function of the two types of forest stands at different scales on the spatial aggregation pattern of forest gap centroid, so as to determine the distribution pattern (uniform, random, aggregated) of the forest gap. See Figures 6(a), 6(b), 6(c) and 6(d) for the fitting function graphs of Ripley's K function and Ripley's L function of the two types of forest stands, wherein L (r) greater than 0 indicates that the forest gap is in an aggregated distribution at the corresponding scale, and if L (r) close to 0, it indicates a random distribution, and if less than 0, it indicates a discrete distribution

[0099] (5) Ecological connectivity index extraction module

[0100] The overall connectivity index (IIC) is introduced to integrate the forest gap area and distance information to measure the ecological accessibility and accessibility of the forest gap in the region. The output results are shown in Figure 7 and Figure 8 .

[0101] Figure 7 IIC weighted network graph (spatial distribution graph of ecological connectivity of mixed coniferous and broad-leaved forest gap, each point represents the position of the forest gap, and the size of the point represents the size of the ecological connectivity).

[0102] Figure 8 IIC weighted network graph (spatial distribution graph of ecological connectivity of coniferous pure forest gap, each point represents the position of the forest gap, and the size of the point represents the size of the ecological connectivity).

[0103] Among the above five modules, the basic structure index extraction module, the shape feature extraction module, the spatial distance structure extraction module, the spatial aggregation evaluation module and the ecological connectivity index extraction module can automatically output statistical results graphs. In addition, the present application also realizes tabular output of the detailed structure parameters of each forest gap and the average level of the regional forest gap parameters (Tables 1 and 2).

[0104] Table 1:

[0105]

[0106] Table 2:

[0107]

[0108] From the results, the forest gap parameter intelligent extraction and structure feature quantitative analysis method proposed in the application shows high universality and regional adaptation ability, can automatically and accurately label the forest gap area and extract the forest gap structure parameters in different tree species and different canopy density forest types, and significantly reduces the manual intervention cost. The method only relies on a single data source (C-FLUX data), has high calculation efficiency, can completely express the forest gap structure features, and ensures accurate extraction of forest gap features of various forests. The application does not need manual participation, the parameter setting is simple, has the ability of fast response, multi-scale adaptation and multi-dimensional structure comprehensive analysis, significantly improves the extraction efficiency of forest gap structure information, and can meet the application requirements of basic forestry.

[0109] The above only describes the preferred embodiments of the application and is not intended to limit the application. Any modification, equivalent replacement and improvement made within the spirit and principle of the application shall be included in the protection scope of the application.

Claims

1. A method for intelligent extraction of forest gap parameters and quantitative analysis of their structural characteristics, characterized by: Contains the following modules: Forest gap basic indicator extraction module: The forest window area is calculated by the total number of pixels and pixel resolution. The perimeter is obtained by extracting the outer contour length of the boundary pixels of the forest window area. The total number of forest windows is directly counted by the number of connected regions identified in the image. The forest window density is calculated by dividing the total number of forest windows by the total area of ​​the region. The forest gap ratio is defined as the ratio of the total area of ​​forest windows to the area of ​​the entire region. Forest gap shape feature index extraction module: Using shape index SI i Measures the closeness of the forest gap area to the ideal circle. The closer the value is to 1, the closer the shape is to the standard circle. The calculation formula is as follows: ； Where, A i Indicates the i The size of the forest gap, P i Indicates the i The perimeter of a forest window, Circularity Index C i Measures the closeness of the forest gap area to the ideal circle. The closer the value is to 1, the closer the shape is to the standard circle. The calculation formula is as follows: ； Where, A i Indicates the i The size of the forest gap, P i Indicates the i The perimeter of a forest window, Using fractal dimension FD i Calculate the complexity of the forest window boundary. The higher the value, the more irregular the forest window boundary. The calculation formula is as follows: ； Where, P i Indicates the i The perimeter of a forest window, A i represents the corresponding area, Forest gap spatial distance structure index extraction module: the centroid of all forest gap areas centroid (row，col) As a spatial analysis object, construct the Euclidean distance matrix D ij , and extract the distance from each forest window to its nearest neighbor , and calculate the maximum nearest neighbor distance Minimum nearest neighbor distance These indicators can effectively evaluate the distribution density between forest gaps and identify clustered, discrete or extremely isolated forest gap structures. The centroid coordinates of the forest gap area are calculated as follows: ； Where, r i , c i Represents the row and column coordinates of each forest window pixel in the area, N Represents the total number of pixels in the area, Forest gap spatial aggregation index extraction module: The forest gap centroid point set is used to evaluate the point pattern at different scales. By gradually increasing the analysis radius, the spatial relationship density between forest windows at each scale is calculated, and the K function value at the maximum scale is used as a quantitative indicator of the degree of aggregation. ； Where, A Indicates the area of ​​the region, n represents the total number of forest gaps, I represents the indicator function, when d ij ≤ r 1 when , otherwise 0. d ij For the i and j The Euclidean distance between the centroids of the forest windows, r is the analysis scale radius, In order to enhance the intuitiveness of the comparison and the linear interpretation, the K function is converted into the L function form in the following way: ； In the formula, when L (r) When >0, it means that the forest gap shows a clustered distribution at scale r; when L (r) <0, it indicates uniform distribution; if L (r) ≈0, it means that the centroids of forest gaps are almost completely randomly distributed. Forest gap overall ecological connectivity index extraction module: using the overall connectivity index in landscape ecology IIC , which is used to measure the ecological accessibility between forest windows. The larger the index value, the higher the degree of connectivity between forest windows; the smaller the value, the greater the discreteness of the forest window distribution. The calculation formula is as follows: ； Where, a i , a j are the areas of the two forest windows, d ij is the Euclidean distance between the two, The method includes the following steps: based on the CHM generated by airborne lidar data, the pixels in the forest window area are segmented by setting a height threshold based on percentiles, and the unsupervised extraction of forest windows is achieved by combining regional connectivity judgment; then, the multi-dimensional forest window structural characteristics of forest window area, shape, spatial distribution, network topology and ecological connectivity are extracted by integrating the modular parameter extraction technology flow. It also contains the following steps: Step 1: Normalize the model input of canopy height, Step 2: Extracting forest window structures, including setting three parameters: area, pixel parameters, and height threshold, reading the two-dimensional array of elevation pixels, extracting forest window pixels using the threshold segmentation percentile, classifying forest window pixels using the 8-neighborhood connectivity, and demarcating forest window areas with labels and numbers. Step 3: Modular extraction and output of structural parameters, including basic indicator steps: area extraction, perimeter extraction, total number of forest windows, forest window density and forest gap ratio; shape characteristic indicator steps: fractal dimension, circularity and shape index; spatial distance structure steps: nearest forest window distance, maximum nearest neighbor distance and minimum nearest neighbor distance; Ripley's K function value step for spatial aggregation; overall connectivity index step for ecological connectivity. The forest gap structure extraction using canopy height threshold segmentation and 8-neighborhood connectivity strategy includes the following steps: Image reading and parameter setting of canopy height model (CHM): The image reading library (rasterio) was used to read CHM images in .tif format and generate the corresponding two-dimensional floating-point elevation matrix. Invalid pixels were shielded to avoid the propagation of errors in subsequent analysis. Parameters were independently adjusted according to forest type and terrain, including the forest window height threshold setting (PERCENTILE) based on percentile control, grid resolution (PIXEL_SIZE), regional area (REGION_AREA), maximum analysis scale (MAX_DISTANCE) and step interval (DIST_STEP). The latter two items were used as the basis for adjusting the parameters of Ripley's K function in spatial pattern analysis. Forest gap mask generation and connectivity identification: According to the canopy height characteristics of tree species or regional forest type statistics, the forest window height threshold is set based on percentile regulation in the parameter setting part, and the pixel areas with height less than the threshold in the CHM image are marked as potential forest window areas. A binary mask image is generated, and then the 8-neighborhood connectivity analysis is used to extract connected components in low-elevation areas, and isolated noise pixels are removed to ensure that the identified forest windows have spatial continuity. Finally, the extracted forest window areas are assigned unique numbers, and a structural attribute table is formed for each forest window area. A number-attribute mapping relationship is established, and a regional mask map is established to provide a spatial reference benchmark for subsequent structural quantification, graph analysis and visualization.

2. The method for intelligent extraction of forest gap parameters and quantitative analysis of their structural characteristics according to claim 1, characterized in that: The threshold segmentation percentile extraction step of forest window pixels includes the modular extraction of multi-dimensional forest window structure parameters based on the forest window pixel area division.

Images