A method, system, and medium for identifying ecologically vulnerable areas based on remote sensing imagery.

The method and system for identifying ecologically vulnerable areas through multi-temporal remote sensing image processing solves the problem of the difficulty in characterizing the evolution process of ecological patches in existing technologies, improves the stability and accuracy of the ecological fragmentation process, and enhances the sensitivity of early identification of ecological degradation.

CN123071818APending Publication Date: 2026-11-10INNER MONGOLIA AUTONOMOUS REGION ACAD OF FORESTRY SCI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610998234.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-06
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing methods for identifying ecologically vulnerable areas based on remote sensing images are insufficient to effectively depict the continuous evolution of ecological patches from intact to fragmented, and are easily affected by noise under complex land cover conditions, resulting in insufficient identification stability and accuracy.

Method used

By acquiring multi-temporal remote sensing images, radiometric correction, geometric registration, and vegetation and bare land index inversion are performed to generate a standardized ecological index image set. Based on the ecological index image set, land cover classification is performed, vegetation and bare land patches are extracted, and pixel-level connected domains are constructed. The patch fragmentation and continuous growth index is confirmed and compared with a preset threshold to identify ecologically vulnerable areas.

Benefits of technology

It enables time-series tracking and structured representation of the evolution of ecological patches from intact to fragmented, improving the identification stability and noise resistance of ecological fragmentation processes, and enhancing the sensitivity and accuracy of early identification of ecological degradation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123071818A_ABST
    Figure CN123071818A_ABST
Patent Text Reader

Abstract

This invention relates to the field of remote sensing image recognition technology, and more particularly to a method, system, and medium for identifying ecologically vulnerable areas based on remote sensing images. The method includes the following steps: acquiring multi-temporal remote sensing images of the target area and performing radiometric correction, geometric registration, and vegetation and bare land index inversion to generate a standardized ecological index image set; classifying land cover based on the ecological index image set, extracting vegetation patches and bare land patches, and constructing pixel-level connected domains to form a spatial distribution map of ecological patches; and confirming the patch fragmentation and continuous growth index based on the spatial distribution maps of ecological patches from adjacent temporal phases. This invention improves the accuracy and stability of ecologically vulnerable area identification by constructing a multi-temporal ecological patch inheritance chain and a fragmentation response persistence determination mechanism to continuously track and quantitatively evaluate the ecological patch fragmentation evolution process.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of remote sensing image recognition technology, and in particular to a method, system, and medium for identifying ecologically vulnerable areas based on remote sensing images. Background Technology

[0002] Ecologically vulnerable areas are typically characterized by vegetation degradation, bare land expansion, and increased surface fragmentation, with their spatial patterns exhibiting clear dynamic evolutionary features in multi-temporal remote sensing imagery. Existing methods for identifying ecologically vulnerable areas based on remote sensing imagery largely rely on single-temporal index thresholding or simple multi-period difference analysis, which struggles to effectively characterize the continuous evolution of ecological patches from intact to fragmented states. Furthermore, these methods are susceptible to noise interference and misclassification under complex land cover conditions. Simultaneously, traditional methods lack refined modeling of the inheritance relationships between patches and the fragmentation process, resulting in insufficient sensitivity to localized, minor fragmentation and expansion processes, thus affecting the stability and accuracy of ecologically vulnerable area identification. Summary of the Invention

[0003] Therefore, it is necessary to provide a method, system, and medium for identifying ecologically vulnerable areas based on remote sensing imagery to solve at least one of the aforementioned technical problems.

[0004] To achieve the above objectives, a method for identifying ecologically vulnerable areas based on remote sensing imagery is provided, the method comprising the following steps:

[0005] Step S1: Acquire multi-temporal remote sensing images of the target area and perform radiometric correction, geometric registration, and vegetation and bare land index inversion to generate a standardized ecological index image set;

[0006] Step S2: Based on the ecological index image set, perform land cover classification, extract vegetation patches and bare land patches, and construct pixel-level connected domains to form a spatial distribution map of ecological patches;

[0007] Step S3: Confirm the continuous growth index of patch fragmentation based on the spatial distribution map of ecological patches in adjacent time phases;

[0008] Step S4: Compare the continuous growth index of patch fragmentation with the preset patch fragmentation growth threshold and output the vulnerability distribution identification result of the target area based on the comparison result.

[0009] This invention provides a system for identifying ecologically vulnerable areas based on remote sensing imagery, used to perform the aforementioned method for identifying ecologically vulnerable areas based on remote sensing imagery. The system includes:

[0010] The image acquisition module is used to acquire multi-temporal remote sensing images of the target area and perform radiometric correction, geometric registration and vegetation and bare land index inversion to generate a standardized ecological index image set.

[0011] The patch analysis module is used to classify land cover based on the ecological index image set, extract vegetation patches and bare land patches, construct pixel-level connected domains, and form a spatial distribution map of ecological patches.

[0012] The growth confirmation module is used to confirm the continuous growth index of patch fragmentation based on the spatial distribution map of ecological patches in adjacent time phases.

[0013] The vulnerable area identification module is used to compare the continuous growth index of patch fragmentation with the preset patch fragmentation growth threshold and output the vulnerable area distribution identification result of the target area based on the comparison result.

[0014] The present invention also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for identifying ecologically vulnerable areas based on remote sensing images.

[0015] The beneficial effects of this invention are that by introducing a multi-temporal ecological patch inheritance control chain and a continuous determination mechanism for the fragmentation response state, it achieves temporal tracking and structured expression of the evolution process of ecological patches from intact to fragmented. Furthermore, by combining the dual-indicator synergistic analysis of patch quantity changes and average area changes, and superimposing the spatiotemporal continuity constraint of the continuous fragmentation chain, it effectively improves the identification stability and noise resistance of the ecological fragmentation process. At the same time, by attributing the same fragmentation process to multiple subsequent inherited patches, it avoids the problem of misjudgment of a single pixel or single temporal phase, thereby achieving refined, continuous, and process-oriented identification of ecologically fragile areas and improving the sensitivity and accuracy of early identification of ecological degradation. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the steps of a method for identifying ecologically vulnerable areas based on remote sensing imagery.

[0017] Figure 2 for Figure 1 A detailed flowchart illustrating the implementation steps of step S3.

[0018] Figure 3 This is a schematic diagram of remote sensing image acquisition and processing according to one embodiment;

[0019] Figure 4 This is a schematic diagram of ecological patch extraction in one embodiment;

[0020] Figure 5 This is a schematic diagram of patch fragmentation and growth in one embodiment;

[0021] Figure 6 This is a vulnerability distribution identification map of one embodiment;

[0022] 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

[0023] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0024] 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.

[0025] 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.

[0026] To achieve the above objectives, please refer to Figures 1 to 6 A method for identifying ecologically vulnerable areas based on remote sensing imagery, the method comprising the following steps:

[0027] Step S1: Acquire multi-temporal remote sensing images of the target area and perform radiometric correction, geometric registration, and vegetation and bare land index inversion to generate a standardized ecological index image set;

[0028] Step S2: Based on the ecological index image set, perform land cover classification, extract vegetation patches and bare land patches, and construct pixel-level connected domains to form a spatial distribution map of ecological patches;

[0029] Step S3: Confirm the continuous growth index of patch fragmentation based on the spatial distribution map of ecological patches in adjacent time phases;

[0030] Step S4: Compare the continuous growth index of patch fragmentation with the preset patch fragmentation growth threshold and output the vulnerability distribution identification result of the target area based on the comparison result.

[0031] In one embodiment, taking the ecological monitoring of a small watershed in a gully region of the Loess Plateau as an example, the area covers approximately 85 km², and includes a mix of sloping farmland, shrubland, and bare slopes. To identify the ecologically vulnerable areas in this region that are continuously degrading, this method was used to process multi-temporal remote sensing images from 2021 to 2023.

[0032] refer to Figure 3 Remote sensing images of the small watershed acquired in the same season of 2021, 2022, and 2023 were obtained, and radiometric correction was performed on each image to eliminate pixel brightness deviations caused by differences in solar radiation intensity at different imaging times. Then, using the 2021 image as the spatial reference image, geometric registration was performed on the 2022 and 2023 images to ensure that the same geographical location in the three images corresponds to the same pixel coordinates. After completing spatiotemporal unification, the vegetation index NDVI and bare ground index BSI corresponding to each pixel were calculated. NDVI represents the degree of vegetation cover, and BSI represents the degree of bare ground. The NDVI and BSI data from each image were then fused according to pixel location to form a standardized ecological index image set for the corresponding year.

[0033] refer to Figure 4 Land cover classification of the target area was performed based on a standardized ecological index image set. Specifically, the NDVI and BSI values ​​corresponding to each pixel were input into a preset classification rule. When a pixel has a high NDVI value and a low BSI value, it was classified as a vegetation pixel; when a pixel has a low NDVI value and a high BSI value, it was classified as a bare land pixel. After pixel classification, connected component analysis was performed on vegetation pixels and bare land pixels respectively, and the eight-neighbor method was used to determine whether adjacent pixels belong to the same spatially continuous region.

[0034] For example, for the central slope area of ​​the watershed, the vegetation pixel set from the corresponding image in 2021 is first extracted. Spatially adjacent and continuously distributed vegetation pixels are merged to generate the first vegetation patch. Subsequently, the vegetation pixel sets of 2022 and 2023 are processed in the same way to generate the vegetation patch sets for the corresponding years. At the same time, the number of pixels, spatial boundary coordinates, and connection relationships with surrounding patches are recorded for each patch, and a spatial distribution map of ecological patches for each year is constructed based on the above information.

[0035] refer to Figure 5Based on the spatial distribution maps of ecological patches from adjacent time phases, the relationship between patch changes within the same location area is analyzed. Specifically, each vegetation patch in the 2021 ecological patch spatial distribution map is first used as the preceding patch object, and subsequent patches that overlap with its spatial location are searched in the 2022 ecological patch spatial distribution map. By calculating the pixel inheritance ratio between patches in the two time phases, the patch correspondence relationship is established.

[0036] After establishing the patch correspondence, the fragmentation process of the same patch in adjacent time phases is further analyzed. For example, when a continuous vegetation patch in 2021 corresponds to multiple spatially separated vegetation sub-patches in 2022, a fragmentation event of the patch is recorded; then, the sub-patches in 2022 are matched with the patches in 2023, and when it is found that the sub-patches have split again or the area has continued to shrink, the corresponding fragmentation process is recorded.

[0037] Based on the number of patch fragmentation events, changes in patch area, and changes in spatial connectivity between patches over consecutive time phases, fragmentation change parameters for each ecological patch are calculated. These parameters from multiple time phases are then arranged chronologically to form a patch fragmentation change sequence. Further analysis of the continuous growth trend within this sequence reveals that when the degree of fragmentation increases across multiple consecutive time periods, a corresponding patch fragmentation continuous growth index is generated.

[0038] refer to Figure 6 The patch fragmentation and continuous growth index corresponding to each spatial region is compared with a preset patch fragmentation and growth threshold. Specifically, the patch fragmentation and continuous growth index is read pixel by pixel, and it is determined whether the index exceeds the preset threshold. When multiple consecutive patches corresponding to a certain spatial region meet the continuous growth condition, the location corresponding to that region is marked as a region to be identified, and the corresponding ecologically vulnerable area distribution identification result is output.

[0039] As an example of the present invention, reference is made to Figure 2 As shown, step S3 in this example includes:

[0040] Step S31: Align the ecological patch spatial distribution maps of adjacent time phases with the same scale grid to obtain the patch set unit of the corresponding time phase;

[0041] Step S32: Identify newly added patch units and reduced patch units in the ecological patch spatial distribution map of adjacent time phases based on patch set units, and mark newly added patch units that meet the preset independent patch conditions as fragmented generated patches.

[0042] Step S33: Using the fragmented patch as the core node, trace its corresponding inherited patch in adjacent time phases along the spatial adjacency relationship, construct the patch inheritance comparison chain, and record the changes in the number of patches and the area splitting ratio within the chain;

[0043] Step S34: Confirm the valid fragmentation and growth events of inherited patches based on the patch inheritance control chain;

[0044] Step S35: Accumulate and count the effective fragmentation growth events, and combine them with the duration of inherited patches in consecutive time phases to generate a patch fragmentation continuous growth index.

[0045] In one embodiment, the spatial distribution maps of ecological patches from adjacent time periods are aligned using raster alignment at the same scale. Specifically, the spatial distribution maps of ecological patches from 2021 and 2022 are input into the analysis module, and a unified raster coordinate system is established based on the original spatial resolution of the remote sensing images. For example, both images are divided into 10m×10m raster units, so that the same geographical location corresponds to the same raster number.

[0046] After the raster unification is completed, the patch set unit corresponding to each time period is extracted according to the raster number. For example, in a certain slope area in 2021, the continuous vegetation area is composed of raster numbers A001-A350. These continuous raster numbers are merged to form the first time phase patch set unit. Then, the raster information of the corresponding location in 2022 is extracted to form the second time phase patch set unit, and a one-to-one spatial correspondence between the two time phases is established.

[0047] Change identification is performed based on patch unit sets from two time phases. Specifically, newly added patch units in 2022 are compared with their corresponding positions in 2021. When a grid cell did not have a vegetation patch identifier in 2021, but a new independent vegetation patch identifier appears in 2022, that area is identified as a newly added patch unit.

[0048] For example, in 2021, a continuous vegetation patch did not contain any independent small patches. However, in 2022, due to the intrusion of local bare land, a new vegetation patch with an area of ​​approximately 0.15 km² was formed within the original vegetation area. Further assessment is needed to determine whether this newly added patch meets the preset conditions for an independent patch, including: no continuous pixel connection between it and the original main patch; the number of independent continuous pixels exceeds a set number; and it has a complete spatial boundary structure.

[0049] When the above conditions are met, the newly added patch unit is marked as a fragmented patch, indicating that the original continuous ecological structure has undergone spatial splitting.

[0050] Using fragmented patches as core nodes, related patches in preceding and following time phases are tracked. Specifically, taking the fragmented patch generated in 2022 as the starting node, the original vegetation patch at the corresponding position in 2021 is searched forward along the eight-neighbor spatial adjacency relationship to determine its source and inheriting patch; at the same time, the ecological patch at the corresponding position in 2023 is searched backward to determine whether the fragmented patch continues to split or its area changes.

[0051] For example, the original vegetation patch P1 in 2021 had an area of ​​2.8 km². In 2022, due to the expansion of bare land within it, it fragmented and formed patch P2, and there is a spatial inheritance relationship between P2 and P1. By 2023, P2 further split into two independent patches, P3 and P4. A patch inheritance control chain was established in chronological order: inherited patch P1 in 2021 → fragmented patch P2 in 2022 → inherited subsequent patches P3 / P4 in 2023.

[0052] During the construction of the patch inheritance control chain, the changes in the number of patches within the chain and the area splitting ratio are recorded simultaneously. For example, the changes in P1 from 1 patch to 2 patches, and then to 4 patches are recorded, and the proportion of inherited patch area to original patch area in each time phase is calculated.

[0053] Valid fragmentation growth events are identified based on a patch inheritance control chain. Specifically, the relationship between patch changes between adjacent time phases in the chain is analyzed, and a valid fragmentation growth event is determined when the following conditions are met:

[0054] On the one hand, multiple patches in subsequent phases can be obtained from the spatial correspondence of a single inherited patch from the preceding phase, that is, there is a clear inheritance relationship;

[0055] On the other hand, the number of patches in the inheritance chain continues to increase, while the area of ​​a single patch continues to decrease.

[0056] For example, if P1 in 2021 corresponds to two sub-patterns, P2 and P3, in 2022, and P2 and P3 continue to correspond to three smaller patches, P4, P5 and P6, in 2023, it indicates that there is a continuous fragmentation evolution process in the region, and this process is recorded as an effective fragmentation growth event.

[0057] All valid fragmentation growth events are cumulatively statistically analyzed, and a fragmentation persistence growth index is generated by combining the persistence length of inherited patches in continuous time phases.

[0058] Specifically, the number of effective fragmentation and growth events occurring within the target area is counted, and the duration of each patch inheritance control chain is obtained. For example, if patch fragmentation occurs in three consecutive monitoring periods in a certain area, its duration is 3; if fragmentation occurs only once in a single time phase, its duration is 1.

[0059] Based on a comprehensive calculation of the cumulative number of fragmentation events, the proportion of patch area splitting, and the duration of the inheritance chain, a corresponding patch fragmentation and continuous growth index is generated to characterize the degree to which ecological patches in the region continuously evolve from an intact state to a fragmented state.

[0060] Preferably, in step S32, marking newly added patch units that meet the preset independent patch conditions as fragmented patches specifically involves:

[0061] The continuous temporal phases based on newly added patch units are divided into newly added patch units in the previous temporal phase and newly added patch units in the subsequent temporal phase;

[0062] A single continuous patch region is identified based on newly added patch units in the previous time phase and newly added patch units in the subsequent time phase;

[0063] The first judgment result is obtained by determining whether a single continuous patch region is replaced by at least two spatially independent new patch units in the current time phase.

[0064] Based on the first judgment result, determine whether the newly added patch units that are independent of each other in space do not satisfy the connectivity constraint, and thus obtain the second judgment result;

[0065] Based on the second judgment result, newly added patch units are selected and marked as fragmented patches.

[0066] In one embodiment, the spatial distribution maps of ecological patches in two adjacent time phases, 2021 and 2022, are obtained. The newly added vegetation patch units in 2022 compared to 2021 are extracted and marked as newly added patch units in the previous time phase and newly added patch units in the next time phase, respectively, according to the time sequence.

[0067] For example, in the 2021 ecological patch spatial distribution map, a continuous vegetation patch A exists on a certain slope area, which is composed of continuous grid cells A001-A500. In the 2022 ecological patch spatial distribution map, due to the expansion of local bare land, several new independent vegetation remnant areas B1, B2, and B3 appear within the original continuous area. The area corresponding to the continuous patch A in 2021 is used as the baseline for the analysis of newly added patches in the previous time phase, and the newly formed patches B1, B2, and B3 in 2022 are used as newly added patch units in the subsequent time phase.

[0068] Using the location coordinates of newly added patch units in a later time phase as an index, we reverse-search the spatial corresponding region in the previous time phase to determine whether multiple newly added patch units in later time phases originate from the same preceding continuous patch region.

[0069] For example, it was found that although the newly added patches B1, B2 and B3 in 2022 were spatially separated, their corresponding locations were all within the continuous vegetation patch A in 2021. Therefore, the preceding regions corresponding to B1, B2 and B3 were uniformly identified as a single continuous patch region A.

[0070] The spatial structure of a single continuous patch region A in two time phases is compared and analyzed. First, the set of connected pixels corresponding to region A in 2021 is obtained, and then the set of newly added patches in the corresponding region in 2022 is obtained. The number of newly added patches corresponding to region A in the current time phase is counted.

[0071] For example, region A in 2021 was a complete connected region with no spatial breaks; while in 2022, three new patch units, B1, B2, and B3, were formed in the corresponding region, and there were no continuous pixel connections between B1, B2, and B3. In this case, it is determined that the original single continuous patch region A has been replaced by at least two independent new patch units, generating the first judgment result.

[0072] Spatial adjacency analysis is performed on B1, B2, and B3 to determine whether there are continuous cell connections between each newly added patch unit according to the preset eight-neighborhood connection rule.

[0073] For example, for B1 and B2, the adjacent pixels in eight directions around the boundary pixel of B1 are detected, and it is found that there are no continuous pixels belonging to B2; at the same time, there is no continuous connection between B2 and B3. Therefore, it is determined that B1, B2 and B3 are spatially independent.

[0074] Calculate the minimum spatial distance between each newly added patch unit and determine whether it exceeds a preset connectivity distance threshold. When there is no cell connection between the newly added patches and the spatial separation condition is met, a second judgment result is generated.

[0075] When the second judgment result indicates that the connectivity constraints between multiple newly added patch units are not met, it is considered that the newly added patch is not an ordinary newly added ecological area, but an independent fragmented unit formed by the internal structural fracture of the original continuous ecological patch.

[0076] For example, the three newly formed patches B1, B2, and B3 in 2022 all originated from continuous vegetation patch A in 2021, and there is no spatial connection between the three. Therefore, B1, B2, and B3 are marked as fragmented patches, and their corresponding source patch A, formation time, and spatial location relationship are recorded.

[0077] Preferably, step S34 includes:

[0078] Traverse each inherited patch in the patch inheritance comparison chain, obtain the number of patches and the area of ​​patches in the corresponding time phase of each inheritance node, and establish a record of changes in inherited patches;

[0079] Based on the inherited patch change records, calculate the change in the number of patches and the change in the average area of ​​a single patch relative to the previous time phase. When the change in the number of patches is greater than zero and the change in the average area of ​​a single patch is less than zero, mark the corresponding inherited patch as a fragmentation response patch.

[0080] Inheritance relationships are traced for fragmentation response patches in continuous time phases to determine whether there is a continuous fragmentation response state in adjacent time phases; if a continuous fragmentation response state exists, the corresponding patch inheritance reference chain is determined as a continuous fragmentation chain.

[0081] The number and duration of the breakage response patches in the continuous breakage chain are counted, and the breakage growth weight is determined according to the proportion of the breakage response patches to the total number of corresponding inherited patches.

[0082] Based on the fragmentation growth weight and duration, a patch fragmentation growth unit value is generated, and inherited patches that meet the preset growth conditions are identified as valid fragmentation growth events.

[0083] In one embodiment, a patch inheritance control chain is read: 2021: original continuous vegetation patch P1 → 2022: inherited patches P2, P3 → 2023: inherited patches P4, P5, P6 → 2024: inherited patches P7, P8, P9, P10, wherein the number and area information of patches corresponding to each time period are obtained.

[0084] Phase Inherited plaque number Number of plaques Total area (km²) Average area of ​​a single patch (km²) 2021 P1 1 4.2 4.2 2022 P2, P3 2 3.6 1.8 2023 P4, P5, P6 3 2.7 0.9 2024 P7, P8, P9, P10 4 2.4 0.6

[0085] The above information is stored chronologically to form a record of inherited patch changes, used to describe the structural changes of the same ecological patch over continuous time. Then, the difference between two adjacent time phases is calculated. For example, in 2022 compared to 2021: the change in the number of patches (2-1=1) indicates that the original single patch split into a new patch. The change in the average area of ​​a single patch (1.8-4.2=-2.4km²) indicates a decrease in the carrying area of ​​a single patch. At this point, the following conditions are met: change in the number of patches > 0; change in the average area of ​​a single patch < 0. Therefore, P2 and P3 in 2022 are marked as fragmentation response patches. Continuing to analyze 2023: the change in the number of patches (3-2=1); the change in the average area of ​​a single patch (0.9-1.8=-0.9km²). The fragmentation criteria are still met, so P4, P5, and P6 are also marked as fragmentation response patches. Through this method, patch nodes exhibiting structural splitting and area reduction over continuous time are identified.

[0086] The spatial correspondence of fragmentation response patches is traced backward along the patch inheritance control chain. For example, P2 and P3 in 2022 have spatial overlap with P4, P5, and P6 in 2023, and P4, P5, and P6 continue to meet the fragmentation response conditions. Subsequently, P4, P5, and P6 in 2023 correspond to P7-P10 in 2024, and continue to show an increase in quantity and a decrease in area. At this point, it is determined that: there is a fragmentation response from 2021 to 2022; there is a continuous fragmentation response from 2022 to 2023; and there is still a fragmentation response from 2023 to 2024. Therefore, there is a continuous fragmentation response state in this patch inheritance control chain, and this inheritance chain is identified as a continuous fragmentation chain.

[0087] The fragmentation response within this ongoing fragmentation chain is statistically analyzed. For example, this inheritance chain experienced three consecutive monitoring periods: 2021-2022: one original patch split into two; 2022-2023: two patches continued to split into three; 2023-2024: three patches continued to split into four. The results show: number of fragmentation response patches = three consecutive change stages; duration = three time periods. The fragmentation growth weight is also calculated as: number of fragmentation response patches / total number of current inheritance patches. For example, if there are four inheritance patches in 2024, and all four continuously exhibit fragmentation response, then the fragmentation growth weight = 4 / 4 = 1. If only some patches continue to split, this weight is reduced accordingly.

[0088] The fragmentation growth weight and duration are combined in the calculation. For example, the fragmentation growth unit value = fragmentation growth weight × duration. In the above case, the fragmentation growth unit value = 1 × 3 = 3. This value is compared with a preset growth condition, for example, setting the effective growth threshold to 2. Since 3 > 2, it is determined that a valid fragmentation growth event has occurred in the region corresponding to this inherited patch.

[0089] Preferably, the inheritance relationship tracing of fracture response patches in continuous time phases to determine whether there is a continuous fracture response state in adjacent time phases specifically involves:

[0090] Using the fragmentation response patch of the current time phase as the reference patch, the inherited patch is matched in the adjacent next time phase patch set based on the spatial overlap relationship to form a patch inheritance correspondence relationship;

[0091] Based on the patch inheritance correspondence, the inherited patches are checked for consistency. When the overlapping area between the inherited patch and the reference patch satisfies the connectivity coverage relationship, the inherited patch is determined to be a valid inherited patch.

[0092] Based on the positional correspondence of effective inherited patches, it is determined whether inherited patches in adjacent time phases simultaneously meet the breakage response state. When they do, it is determined that a continuous breakage response state pair is formed.

[0093] Chain the continuous crushing response state pairs to construct a continuous crushing response inheritance chain;

[0094] A continuous fracture response state is determined to exist when the continuous fracture response state pairs in the continuous fracture response inheritance chain are consecutively true in time.

[0095] In one embodiment, fragmented response patches P2 and P3 identified in 2022 are used as the current time-phase baseline patches. The patch set in the 2023 ecological patch spatial distribution map is read and matched within the corresponding spatial range. For example: 2022:

[0096] Fragmentation response patch P2; Fragmentation response patch P3. 2023: Subsequent patch P4; Subsequent patch P5; Subsequent patch P6. Calculate the spatial overlap ratio between P2, P3, and P4-P6 respectively. Specifically: P2 and P4 have 65% spatial overlap; P2 and P5 have 20% spatial overlap; P3 and P6 have 72% spatial overlap. Therefore, establish the patch inheritance correspondence: P2→P4; P3→P6; . For patches with low overlap ratios and only edge contact relationships, no inheritance relationship is established.

[0097] Further analysis of the spatial coverage between P2 and P4 is conducted. The boundary cell set of P2 is extracted, and it is determined whether P4 covers the main continuous area within the original spatial range of P2. For example, if P2 contains 500 continuous cells, and P4 covers 360 cells within the continuous area of ​​P2, achieving a coverage ratio of 72% with no obvious breaks in the coverage area, then P4 is considered to inherit the spatial structure variation characteristics of P2 and is identified as a valid inherited patch corresponding to P2. Similarly, P3 and P6 satisfy a continuous coverage relationship, therefore P6 is identified as a valid inherited patch of P3. After verification, the valid inheritance relationships are: P2→P4; P3→P6.

[0098] Obtain the patch change information corresponding to P4 and P6 in 2023 and recalculate their fragmentation response parameters. For example: P2 in 2022: formed by the splitting of an original continuous patch; the area of ​​a single patch decreased. P4 in 2023: continued to split relative to P2 in 2022; the corresponding area changed from 1 patch to 2 sub-patches; the average area of ​​a single patch continued to decrease. At this time, both P2 and P4 satisfy: the number of patches increased; the average area of ​​a single patch decreased. Therefore, it is determined that P2→P4 constitutes a continuous fragmentation response state pair. Similarly: P3→P6 also satisfies the above conditions.

[0099] Connect the consecutive breaking response state pairs in chronological order, for example:

[0100] Phase plaque status Patch number Changes in the number of plaques Breaking response state 2021 Original continuous plaque P1 1 continuous plaque Initial unbroken state Phase plaque status Patch number Changes in the number of plaques Breaking response state 2022 First fracture response P2, P3 One plaque split into two plaques The number of patches increases while the area of ​​each patch decreases, leading to the initial fragmentation response. 2023 Inheritance broken response P4, P6 It continued to evolve from two inherited patches. It inherits the fracturing characteristics of the previous phase and continues to meet the fracturing response conditions. 2024 Continuous Fragmentation Response P7, P8, P9, P10 Two inherited plaques further split into four plaques. The trend of fragmentation and growth has been maintained across multiple time phases.

[0101] Then the successive breaking response inheritance chain:

[0102] Inheritance chain number Continuous fracture response path Evolutionary process Chain 1 P1→P2→P4→P7 The original continuous plaque P1 continued to fragment after multi-phase inheritance, forming fragmented plaques. Chain 2 P1→P3→P6→P8 / P9 / P10 Another branch of the original continuous patch P1 continues to undergo spatial splitting, forming multiple independent fragmented patches.

[0103] A continuous fracture response state is determined to exist when the continuous fracture response state pairs in the continuous fracture response inheritance chain are consecutively true in time.

[0104] The temporal transitions in this inheritance chain were examined: 2021-2022: P1 split to form P2 and P3, satisfying the fragmentation response; 2022-2023: P2 and P3 correspondingly formed P4 and P6, continuing to satisfy the fragmentation response; 2023-2024: P4 and P6 continued to split to form multiple sub-patterns, satisfying the fragmentation response. Since a continuous evolutionary relationship of "fractured patch in the previous temporal phase → inherited patch in the next temporal phase → further fragmentation" exists across multiple monitoring periods, it is determined that this region exhibits a continuous fragmentation response state, and the corresponding inheritance chain is considered a persistent fragmentation chain.

[0105] Preferably, determining whether inherited patches in adjacent time phases simultaneously satisfy the fragmentation response state based on the positional correspondence of effective inherited patches includes:

[0106] Obtain the valid inherited patches at both ends of the location correspondence;

[0107] Determine whether there are multiple subsequent valid inherited patches corresponding to the valid inherited patch of the previous time phase;

[0108] When multiple subsequent valid inherited patches correspond to the same valid inherited patch, determine whether all the multiple subsequent valid inherited patches are in a fragmented response state.

[0109] When multiple subsequent valid inherited patches are all in the fragmentation response state, it is determined that the fragmentation state of the valid inherited patches in the previous time phase is continued in the subsequent time phase.

[0110] The positional correspondence of the continued fracture state is determined as a continuous fracture response state pair.

[0111] In one embodiment, the valid inheritance relationship established between 2022 and 2023 is read to obtain the valid inherited patches of the previous time phase and the corresponding inherited patches of the next time phase. For example, two fragmentation response patches were identified in 2022: P2 and P3. After spatial overlap matching, the corresponding relationship is established: P2→P4; P3→P6. Among them, P2 and P3 are the valid inherited patches of the previous time phase, and P4 and P6 are the valid inherited patches of the next time phase.

[0112] Iterate through each valid inherited patch from the previous time phase and count the number of inherited patches from the next time phase. For example: For P2: Spatial matching results: P2→P4; P2→P5. This means that P2 forms two inherited patches, P4 and P5, in the next time phase. For P3: Spatial matching results: P3→P6. This means that P3 corresponds to only one inherited patch in the next time phase.

[0113] Record the above correspondence as follows:

[0114] Previous phase plaque The corresponding plaque at the next moment P2 P4, P5 P3 P6

[0115] Further analysis is needed to determine whether multiple subsequent valid inherited patches correspond to the same valid inherited patch. For example, for P2: in 2022, P2 was a continuously fragmented patch; in 2023, the region further fragmented to form two independent patches, P4 and P5. Therefore, it is determined that P2 has multiple subsequent valid inherited patches. Further analysis is then needed to determine whether P4 and P5 are both in a fragmentation response state.

[0116] Obtain the patch change parameters for P4 and P5 in 2023. For example: P4: Compared to the corresponding region in 2022, the number of patches continued to increase; the average area of ​​a single patch further decreased. Therefore, it meets the fragmentation response condition. P5: Also formed by inheriting the spatial range of P2; subsequently, it continued to show area reduction and spatial fragmentation. Therefore, it also meets the fragmentation response condition. Judgment: Both P4 and P5 are in a fragmentation response state.

[0117] Subsequently, when multiple effectively inherited patches in subsequent time phases are all in a fragmentation response state, it is determined that the fragmentation state of the effectively inherited patches in the previous time phase is continued in the subsequent time phase. For example: P2 in 2022: formed by the splitting of the original continuous patch, belongs to the fragmentation response patch; P4 and P5 in 2023: continue to be inherited by P2 and again exhibit fragmentation response. Therefore, it is considered that the fragmentation state corresponding to P2 has not disappeared, but continues to be passed on in the subsequent time phase. Similarly: in P3→P6, if P6 continues to meet the fragmentation response condition, it is considered that the fragmentation state of P3 is continued. Finally, the positional correspondence of the continued fragmentation state is determined as a continuous fragmentation response state pair.

[0118] For example:

[0119] Previous phase fragmentation response patch Subsequent phase inherited plaques Fragmentation status State determination P2 P4, P5 Both P4 and P5 continue to meet the breaking response. Continuous breakage response state P3 P6 P6 continues to meet the breaking response. Continuous breakage response state

[0120] Preferably, when multiple subsequent valid inherited patches correspond to the same valid inherited patch, determining whether all multiple subsequent valid inherited patches are in a fragmentation response state further includes:

[0121] Calculate the total area corresponding to multiple subsequent valid inherited patches;

[0122] Obtain the patch area of ​​the effective inherited patch from the previous time phase;

[0123] Determine whether the total area corresponding to multiple subsequent valid inherited patches maintains a continuous succession relationship with the area of ​​the valid inherited patch in the previous time phase;

[0124] When the total area corresponding to multiple subsequent valid inherited patches maintains a continuous inheritance relationship, it is determined that multiple subsequent valid inherited patches originated from the same fragmentation process.

[0125] Based on the same fracturing process, multiple subsequent valid inherited patches were determined to be in a fracturing response state.

[0126] In one embodiment, multiple subsequent valid inherited patches corresponding to the same previous valid inherited patch are obtained, and the area information of each inherited patch is read. For example, in 2022, there is a fragmentation response patch P2 with an area of ​​1.8 km². Through spatial inheritance matching, it is found that in 2023, multiple subsequent valid inherited patches correspond to P2: P4 with an area of ​​0.9 km² and P5 with an area of ​​0.75 km². The areas of P4 and P5 are summed: the total area of ​​subsequent valid inherited patches is 0.9 + 0.75 = 1.65 km².

[0127] Obtain the patch area of ​​the effective inherited patch from the previous time phase. Specifically, read the spatial area information of P2 in 2022 and use it as the area benchmark for the previous time phase. For example: P2 area = 1.8 km². This area serves as a reference value for determining whether multiple inherited patches in the subsequent time phase originated from the same fragmentation process.

[0128] Determine whether there is a continuous continuity between the total area of ​​multiple subsequent valid inherited patches and the area of ​​valid inherited patches in the previous time phase. Specifically, compare: patch area in the previous time phase: Sprevious = 1.8 km2; total area of ​​inherited patches in the subsequent time phase: Slater = 1.65 km2. Calculate the area continuity ratio: Slater / Sprevious = 1.65 / 1.8 = 91.7%.

[0129] Since the total area of ​​multiple inherited patches in the later time phase still maintains a high proportion of coverage over the area of ​​patches in the previous time phase, and the area loss is within the preset natural variation range, it is believed that multiple subsequent effective inherited patches maintain a continuous succession relationship with patches in the previous time phase.

[0130] Spatial relationships can be used as a supplementary factor in the judgment. For example, although P4 and P5 were spatially separated in 2023, both were located within the coverage area of ​​P2 in 2022, and their combined area could cover the original main area of ​​P2. Therefore, it can be determined that P4 and P5 are not two independent newly added patches, but rather inherited branches formed after the fragmentation of P2.

[0131] When the total area corresponding to multiple subsequent valid inherited patches maintains a continuous succession relationship, it is determined that multiple subsequent valid inherited patches originated from the same fragmentation process. Specifically, it is considered that: 2022: a single fragmentation response patch P2 → 2023: P2 split to form P4 and P5. This change satisfies: continuous spatial location between the two phases; a succession relationship in area; and an increase in the number of inherited patches. Therefore, it is determined that P4 and P5 originated from the same ecological patch fragmentation process.

[0132] Based on the same fragmentation process, multiple subsequent effective inherited patches are determined to be in a fragmentation response state. Specifically, since P4 and P5 are both formed by the splitting of the internal structure of P2, and the inheritance process exhibits the following characteristics: the number of original patches decreases while the corresponding unit expands; the number of inherited patches increases; the area of ​​a single patch decreases; and the spatial structure changes from continuous to discrete. Therefore, P4 and P5 are both marked as fragmentation response states, forming a continuous fragmentation response relationship:

[0133] Phase Patch number area Change Relationship state 2022 P2 1.8km² Single fragmentation response patch Fragmentation initiation state 2023 P4 0.9km² Formed by P2 inheritance Fragmentation response continuation 2023 P5 0.75km² Formed by P2 inheritance split Fragmentation response continuation

[0134] Preferably, determining that multiple subsequent valid inherited patches are all in a crushing response state based on the same crushing process specifically means:

[0135] Obtain the common preceding inherited patch corresponding to multiple subsequent valid inherited patches;

[0136] Extract the boundary continuation segment along the boundary contour of the common preceding inherited patch;

[0137] Obtain the boundary inheritance segments corresponding to multiple subsequent valid inherited patches;

[0138] Determine whether each boundary inheritance segment can continuously inherit from the boundary inheritance segment;

[0139] When each boundary inheritance segment continuously inherits another boundary inheritance segment, multiple subsequent valid inheritance patches are identified as corresponding to the same fragmentation process.

[0140] Multiple subsequent valid inherited patches corresponding to the same fracturing process are identified as fracturing response patches.

[0141] In one embodiment, the patch inheritance correspondence is read, and the common source of multiple valid inherited patches in subsequent time phases is retrieved in reverse. For example: 2023: P4; P5; both valid inherited patches can be traced back to the same patch P2 in 2022 through spatial overlap. Therefore, it is determined that: common preceding inherited patch = P2. Wherein: P2 is a continuous patch before the fragmentation; P4 and P5 are multiple inheritance branches formed after the fragmentation.

[0142] Extract the outer boundary contour of P2 in 2022 and traverse it according to the boundary cell order to obtain the boundary nodes that can reflect the changes in the outer edge structure of the patch. For example, the boundary of P2 consists of multiple consecutive boundary points: B1, B2, B3...B100.

[0143] Further analysis of the location where P2 broke apart revealed boundary regions related to the subsequent splitting direction. For example, the eastern and southern boundary regions of P2 formed P4 and P5 in 2023, respectively. Therefore, the following boundary segments were extracted: eastern boundary segment B30-B55; southern boundary segment B70-B85; as boundary transition segments.

[0144] Subsequently, the boundary inheritance segments corresponding to multiple subsequent valid inherited patches are obtained. Specifically, the boundary contours of P4 and P5 are extracted, and their inheritance source is determined based on their positional correspondence with P2. For example, P4 inherits the eastern region of P2, so the boundary segment corresponding to B30-B55 in the P4 boundary is extracted; P5 inherits the southern region of P2, so the boundary segment corresponding to B70-B85 in the P5 boundary is extracted. Thus: P4 boundary inheritance segment: C30-C55; P5 boundary inheritance segment: D70-D85.

[0145] Map the boundary inheritance segments of P4 and P5 back to the original boundary coordinate system of P2, and determine whether the mapped boundary positions are continuous with the corresponding boundary segments of P2. For example, mapping the boundary inheritance segment C30-C55 of P4 to the P2 boundary: C30-C55 and B30-B55 are spatially continuous; there are no obvious breaks or jumps. Simultaneously, mapping the boundary inheritance segment D70-D85 of P5 to the P2 boundary: D70-D85 and B70-B85 are continuously corresponding. Therefore, it is determined that both P4 and P5 can inherit the original boundary structure of P2.

[0146] When each boundary inheritance segment inherits from another boundary inheritance segment, it is determined that multiple subsequent valid inheritance patches correspond to the same fragmentation process. Specifically: if multiple subsequent inheritance patches correspond to different boundary regions of a common preceding patch, and these boundary regions can completely continue the original patch boundary structure, it indicates that the multiple subsequent patches are not generated independently, but are formed by the same spatial fragmentation process.

[0147] For example: In 2022, P2 was a complete and continuous patch; in 2023, degradation and fracturing occurred within P2, forming P4 and P5. P4 and P5 inherited different boundary regions from P2, and the boundary changes corresponded continuously. Therefore, it is determined that P4 and P5 originated from the same fragmentation process.

[0148] Multiple subsequent valid inherited patches corresponding to the same fracturing process are identified as fracturing response patches. Specifically, P4 and P5 are uniformly labeled as fracturing response patches generated from the fracturing evolution of P2, forming the following inheritance relationship:

[0149] Phase Patch number Boundary relations determination 2022 P2 Continuous boundary connection sections B30-B55 and B70-B85 Common Pre-inherited Plaques 2023 P4 Inherited from the P2 boundary region B30-B55 Fragmented response patches 2023 P5 Inherited from the P2 boundary region B70-B85 Fragmented response patches

[0150] Of particular importance, step S2 includes:

[0151] Based on the coverage type of the ecological index image set, each pixel is divided into vegetation pixels and bare land pixels.

[0152] Extract the vegetation pixel set and the bare land pixel set separately to obtain the vegetation distribution map and the bare land distribution map, and then remove the bare land distribution map;

[0153] Based on the adjacency relationship between pixels in the vegetation distribution map, connected components are constructed from the pixels in the vegetation distribution map to obtain multiple vegetation connected components.

[0154] Each vegetation connectivity domain is independently identified, and corresponding vegetation patch identification information is generated.

[0155] Based on the vegetation patch identification information, the boundary contours and spatial locations of each vegetation patch in the vegetation distribution map are extracted to establish the spatial distribution relationship of vegetation patches.

[0156] An ecological patch spatial distribution map is generated by analyzing the spatial distribution relationship, boundary outline, and vegetation patch identification information of vegetation patches.

[0157] In one embodiment, a threshold is used to determine the ecological index value of each pixel in the image. For example, when NDVI ≥ 0.5, it is determined to be a vegetation pixel, and when NDVI < 0.5, it is determined to be a bare land pixel. This forms two types of pixel sets: vegetation pixel set: covering continuous vegetation areas, such as riparian forest belts and wetland meadows; bare land pixel set: covering bare soil or water body edge areas.

[0158] Only the spatial raster structure corresponding to the vegetation pixel set is retained to form a single vegetation distribution map for subsequent ecological patch analysis; the bare land distribution map is removed or temporarily stored in this step because it does not participate in the ecological patch connectivity analysis.

[0159] An 8-neighborhood connectivity rule is used to cluster spatially adjacent raster cells that are all vegetation cells. For example, in a riverbank region, multiple adjacent vegetation cells are continuously connected to form a connected component, while vegetation areas spatially separated by water bodies or bare land form multiple independent connected domains. For example, in this wetland region, the following might be formed: connected domain A (riverbank vegetation zone); connected domain B (central wetland meadow); and connected domain C (patched vegetation area). Subsequently, each vegetation connected domain is independently identified to generate corresponding vegetation patch identification information.

[0160] Specifically, a unique ID is assigned to each connected component, for example: A→P1; B→P2; C→P3; where each P represents a vegetation patch identifier, which is used for subsequent cross-temporal tracking and evolution analysis.

[0161] For each connected region, boundary extraction is performed. For example, edge tracing algorithms are used to extract the outer contour of pixels, forming a polygonal boundary structure, and its spatial coordinate range, area, and relative positional relationship are calculated. For example: P1 is located on the north bank of the river; P2 is located in the central area of ​​the wetland; P3 is located in the southern transition zone; based on this, the spatial adjacency relationship and distribution topology between patches are established.

[0162] All vegetation patches are organized in a graph structure, where nodes represent patch identifiers, edges represent spatial adjacency or proximity relationships, and corresponding boundary contour information is overlaid to form a unified ecological patch spatial distribution map data structure. For example, the final output structure can be represented as:

[0163] Patch ID area Spatial location Adjacency P1 3.2km² North Shore Adjacent to P2 P2 2.8km² central wetland Adjacent to P1 and P3 P3 1.5km² South transition zone Adjacent to P2

[0164] Of particular importance is that, before independently identifying each vegetation connectivity domain, the following steps are also included:

[0165] Spatial topology pre-inheritance parsing is performed on the vegetation connected domain, and a topological connected skeleton inside the connected domain is generated based on the cell adjacency relationship;

[0166] Based on the topological connectivity skeleton, the boundary separable response sequences of each vegetation connectivity domain are extracted, where the boundary separable response sequences are used to characterize the continuity change trend of the connectivity domain boundaries.

[0167] Breakpoint consistency screening is performed on the boundary separable response sequence to determine the potential structural fracture location and generate fracture candidate boundary segments;

[0168] Based on the candidate boundary segments of the fracture, a structural splitting pre-mapping relationship is constructed for the connected domain, where the structural splitting pre-mapping relationship is used to characterize the possible splitting directions of the same connected domain in space;

[0169] Each vegetation connectivity domain is independently identified based on the pre-mapping relationship of structural splitting.

[0170] In one embodiment, a single vegetation connectivity domain is used as the analysis object. All vegetation cells within it are modeled as a graph structure according to 8-neighborhood relationships. Cells are regarded as nodes and adjacency relationships are regarded as edges, thereby generating the skeleton structure of the connectivity domain. For example, in a wetland vegetation area with a strip-shaped distribution, a topological connectivity skeleton similar to "trunk + branches" can be extracted, where the trunk corresponds to a continuous high-density vegetation area and the branches correspond to sparse edge areas.

[0171] The curvature and density changes of the boundary are calculated pixel by pixel along the outer contour of the connected domain. For example, in the riverbank vegetation area, some boundary segments may show abrupt curvature changes or density decreases. These changes are arranged in spatial order to form a boundary separable response sequence, which is used to describe whether the boundary "tends to break".

[0172] In the sequence, abrupt changes in continuous variation are detected. For example, if the boundary curvature suddenly changes from a gentle state to a highly volatile state, or if the vegetation density drops sharply over a short distance, the location is identified as a candidate breakpoint. Combining multiple adjacent breakpoints forms a candidate fault boundary segment. For example, if a continuous weak vegetation zone appears in a certain area on the southern edge of a wetland, it is marked as a potential fault zone.

[0173] Analyzing the spatial directionality of candidate fault boundary segments, for example, if a connected domain shows a faulting trend on both its eastern and southern sides, pre-mapping relationships of "eastward splitting direction" and "southward splitting direction" are generated respectively to describe the spatial paths that may lead to future splitting into multiple sub-patches. For example: candidate fault on the eastern side of connected domain A (wetland vegetation patch) → pre-mapped as potential splitting region A-1; candidate fault on the southern side → pre-mapped as potential splitting region A-2, thus forming a structural splitting pre-mapping relationship map.

[0174] Instead of directly assigning a single identifier to the entire connected domain, a pre-mapping structure is pre-divided based on pre-mapping relationships. For example, for a connected domain A with a bidirectional breaking tendency, its internal structure may be pre-identified as potentially split into two sub-structural units, A1 and A2, and each is assigned an independent identifier during the identification stage. This avoids cross-temporal mismatches caused by premature breaking in subsequent temporal analysis. The final processing logic is as follows:

[0175] Connected Components Topological features Fault candidate boundary Premapped splitting direction Identification results A Main trunk + lateral branches East / South A1 / A2 A1 and A2 are independent identifiers. B Single stable structure No obvious fracture none B. Maintain overall branding.

[0176] This invention provides a system for identifying ecologically vulnerable areas based on remote sensing imagery, used to perform the aforementioned method for identifying ecologically vulnerable areas based on remote sensing imagery. The system includes:

[0177] The image acquisition module is used to acquire multi-temporal remote sensing images of the target area and perform radiometric correction, geometric registration and vegetation and bare land index inversion to generate a standardized ecological index image set.

[0178] The patch analysis module is used to classify land cover based on the ecological index image set, extract vegetation patches and bare land patches, construct pixel-level connected domains, and form a spatial distribution map of ecological patches.

[0179] The growth confirmation module is used to confirm the continuous growth index of patch fragmentation based on the spatial distribution map of ecological patches in adjacent time phases.

[0180] The vulnerable area identification module is used to compare the continuous growth index of patch fragmentation with the preset patch fragmentation growth threshold and output the vulnerable area distribution identification result of the target area based on the comparison result.

[0181] The present invention also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for identifying ecologically vulnerable areas based on remote sensing images.

[0182] Therefore, the embodiments should be considered as 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.

[0183] 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 identifying ecologically vulnerable areas based on remote sensing imagery, characterized in that, Includes the following steps: Step S1: Acquire multi-temporal remote sensing images of the target area and perform radiometric correction, geometric registration, and vegetation and bare land index inversion to generate a standardized ecological index image set; Step S2: Based on the ecological index image set, perform land cover classification, extract vegetation patches and bare land patches, and construct pixel-level connected domains to form a spatial distribution map of ecological patches; Step S3: Confirm the continuous growth index of patch fragmentation based on the spatial distribution map of ecological patches in adjacent time phases; Step S4: Compare the continuous growth index of patch fragmentation with the preset patch fragmentation growth threshold and output the vulnerability distribution identification result of the target area based on the comparison result.

2. The method for identifying ecologically vulnerable areas based on remote sensing imagery according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: Align the ecological patch spatial distribution maps of adjacent time phases with the same scale grid to obtain the patch set unit of the corresponding time phase; Step S32: Identify newly added patch units and reduced patch units in the ecological patch spatial distribution map of adjacent time phases based on patch set units, and mark newly added patch units that meet the preset independent patch conditions as fragmented generated patches. Step S33: Using the fragmented patch as the core node, trace its corresponding inherited patch in adjacent time phases along the spatial adjacency relationship, construct the patch inheritance comparison chain, and record the changes in the number of patches and the area splitting ratio within the chain; Step S34: Confirm the valid fragmentation and growth events of inherited patches based on the patch inheritance control chain; Step S35: Accumulate and count the effective fragmentation growth events, and combine them with the duration of inherited patches in consecutive time phases to generate a patch fragmentation continuous growth index.

3. The method for identifying ecologically vulnerable areas based on remote sensing imagery according to claim 2, characterized in that, In step S32, the newly added patch units that meet the preset independent patch conditions are marked as fragmented patches. The continuous temporal phases based on newly added patch units are divided into newly added patch units in the previous temporal phase and newly added patch units in the subsequent temporal phase; A single continuous patch region is identified based on newly added patch units in the previous time phase and newly added patch units in the subsequent time phase; The first judgment result is obtained by determining whether a single continuous patch region is replaced by at least two spatially independent new patch units in the current time phase. Based on the first judgment result, determine whether the newly added patch units that are independent of each other in space do not satisfy the connectivity constraint, and thus obtain the second judgment result; Newly added patch units are selected based on the second judgment result and marked as fragmented patches.

4. The method for identifying ecologically vulnerable areas based on remote sensing imagery according to claim 2, characterized in that, Step S34 includes: Traverse each inherited patch in the patch inheritance comparison chain, obtain the number of patches and the area of ​​patches in the corresponding time phase of each inheritance node, and establish a record of changes in inherited patches; Based on the inherited patch change records, calculate the change in the number of patches and the change in the average area of ​​a single patch relative to the previous time phase. When the change in the number of patches is greater than zero and the change in the average area of ​​a single patch is less than zero, mark the corresponding inherited patch as a fragmentation response patch. Inheritance relationships are traced for fragmentation response patches in continuous time phases to determine whether there is a continuous fragmentation response state in adjacent time phases; if a continuous fragmentation response state exists, the corresponding patch inheritance reference chain is determined as a continuous fragmentation chain. The number and duration of the breakage response patches in the continuous breakage chain are counted, and the breakage growth weight is determined according to the proportion of the breakage response patches to the total number of corresponding inherited patches. Based on the fragmentation growth weight and duration, a patch fragmentation growth unit value is generated, and inherited patches that meet the preset growth conditions are identified as valid fragmentation growth events.

5. The method for identifying ecologically vulnerable areas based on remote sensing imagery according to claim 4, characterized in that, To trace the inheritance relationship of fracture response patches in consecutive time phases and determine whether there is a continuous fracture response state in adjacent time phases, the specific steps are as follows: Using the fragmentation response patch of the current time phase as the reference patch, the inherited patch is matched in the adjacent next time phase patch set based on the spatial overlap relationship to form a patch inheritance correspondence relationship; Based on the patch inheritance correspondence, the inherited patches are checked for consistency. When the overlapping area between the inherited patch and the reference patch satisfies the connectivity coverage relationship, the inherited patch is determined to be a valid inherited patch. Based on the positional correspondence of effective inherited patches, it is determined whether inherited patches in adjacent time phases simultaneously meet the breakage response state. When they do, it is determined that a continuous breakage response state pair is formed. Chain the continuous crushing response state pairs to construct a continuous crushing response inheritance chain; A continuous fracture response state is determined to exist when the continuous fracture response state pairs in the continuous fracture response inheritance chain are consecutively true in time.

6. The method for identifying ecologically vulnerable areas based on remote sensing imagery according to claim 5, characterized in that, Determining whether inherited patches in adjacent time phases simultaneously satisfy the fragmentation response state based on the location correspondence of effective inherited patches includes: Obtain the valid inherited patches at both ends of the location correspondence; Determine whether there are multiple subsequent valid inherited patches corresponding to the valid inherited patch in the previous time phase; When multiple subsequent valid inherited patches correspond to the same valid inherited patch, determine whether all the multiple subsequent valid inherited patches are in a fragmented response state. When multiple subsequent valid inherited patches are all in the fragmentation response state, it is determined that the fragmentation state of the valid inherited patches in the previous time phase is continued in the subsequent time phase. The positional correspondence of the continued fracture state is determined as a continuous fracture response state pair.

7. The method for identifying ecologically vulnerable areas based on remote sensing imagery according to claim 6, characterized in that, When multiple subsequent valid inherited patches correspond to the same valid inherited patch, determining whether all multiple subsequent valid inherited patches are in a fragmentation response state also includes: Calculate the total area corresponding to multiple subsequent valid inherited patches; Obtain the patch area of ​​the effective inherited patch from the previous time phase; Determine whether the total area corresponding to multiple subsequent valid inherited patches maintains a continuous succession relationship with the area of ​​the valid inherited patch in the previous time phase; When the total area corresponding to multiple subsequent valid inherited patches maintains a continuous inheritance relationship, it is determined that multiple subsequent valid inherited patches originated from the same fragmentation process. Based on the same fracturing process, multiple subsequent valid inherited patches were determined to be in a fracturing response state.

8. The method for identifying ecologically vulnerable areas based on remote sensing imagery according to claim 7, characterized in that, Based on the same fracturing process, it is determined that multiple subsequent valid inherited patches are all in the fracturing response state, specifically: Obtain the common preceding inherited patch corresponding to multiple subsequent valid inherited patches; Extract the boundary continuation segment along the boundary contour of the common preceding inherited patch; Obtain the boundary inheritance segments corresponding to multiple subsequent valid inherited patches; Determine whether each boundary inheritance segment can continuously inherit from the boundary inheritance segment; When each boundary inheritance segment continuously inherits another boundary inheritance segment, multiple subsequent valid inheritance patches are identified as corresponding to the same fragmentation process. Multiple subsequent valid inherited patches corresponding to the same fracturing process are identified as fracturing response patches.

9. A system for identifying ecologically vulnerable areas based on remote sensing imagery, characterized in that, For performing the method for identifying ecologically vulnerable areas based on remote sensing imagery as described in claim 1, the system for identifying ecologically vulnerable areas based on remote sensing imagery includes: The image acquisition module is used to acquire multi-temporal remote sensing images of the target area and perform radiometric correction, geometric registration and vegetation and bare land index inversion to generate a standardized ecological index image set. The patch analysis module is used to classify land cover based on the ecological index image set, extract vegetation patches and bare land patches, construct pixel-level connected domains, and form a spatial distribution map of ecological patches. The growth confirmation module is used to confirm the continuous growth index of patch fragmentation based on the spatial distribution map of ecological patches in adjacent time phases. The vulnerable area identification module is used to compare the continuous growth index of patch fragmentation with the preset patch fragmentation growth threshold and output the vulnerable area distribution identification result of the target area based on the comparison result.

10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for identifying ecologically vulnerable areas based on remote sensing imagery as described in any one of claims 1-8.