Intelligent situation recognition method based on multi-source data fusion

Abstract

The application discloses an intelligent situation recognition method based on multi-source data fusion, and relates to the technical field of situation recognition, and comprises the following steps: in the case that the existence consistency of a target region exists and the boundary closure inconsistency is inconsistent, path extension tracking and fracture propagation analysis are performed on a closed evolution sequence to form a closed damage sequence, and a closed attenuation parameter and an expansion gradient parameter are calculated in combination with an existence discrimination sequence to represent the boundary closure difference degree; the closed attenuation parameter and the expansion gradient parameter are coupled and operated, a consistent offset parameter is generated by introducing an existence consistency strength, and a target region integrity evaluation coefficient is constructed through nonlinear mapping. The application solves the problem of integrity misjudgment caused by the existence consistency of the target region but the boundary closure inconsistency in the multi-source data fusion process, realizes dynamic determination and regulation based on the boundary closure difference, and thus improves the situation recognition accuracy.

Description

Technical Field

[0001] This invention relates to the field of situational awareness technology, and more specifically to an intelligent situational awareness method based on multi-source data fusion. Background Technology

[0002] Intelligent situational awareness based on multi-source data fusion refers to an intelligent information processing method that, in complex application environments, integrates heterogeneous data (such as sensor data, video surveillance information, network logs, environmental monitoring data, and geographic information) from different sources, structures, and spatiotemporal scales to construct a unified data representation and semantic model. It then leverages intelligent algorithms such as machine learning, deep learning, and knowledge graphs to comprehensively analyze and dynamically reason about system operating status, event evolution processes, and potential risks. Existing technologies typically first collect and preprocess multi-source data, including data cleaning, format conversion, and time synchronization. Then, feature extraction and standardization are used to achieve comparability between different data sources. Finally, spatiotemporal alignment methods are employed to map various data types to a unified model. Within the time axis and spatial coordinate system, multimodal fusion strategies are used to fuse and model data features from different sources, forming a unified situational feature representation. Subsequently, the fused data is analyzed using classification and recognition models, time series analysis models, or graph structure reasoning models to identify target states, uncover relationships between events, and predict development trends. Ultimately, visualized or structured situational awareness results are output to support decision analysis. Overall, this type of technology generally includes multiple stages such as a data acquisition layer, a data preprocessing layer, a feature extraction and fusion layer, an intelligent analysis and reasoning layer, and a result output and application layer. These stages are closely connected through data flow and model calculation, thereby realizing the transformation from dispersed perceived information to global situational awareness.

[0003] The existing technology has the following shortcomings: In the process of intelligent situational awareness based on multi-source data fusion, when determining the integrity of a target area, visible light data presents a closed boundary due to its clear structural edges. However, infrared data is affected by thermal diffusion and imaging blurring, and its corresponding regional boundaries are prone to breakage but still maintain an overall response, resulting in a situation where the region's existence is consistent but the boundary closure is inconsistent. Due to the differences in boundary representation capabilities of different modalities, the same target exhibits inconsistent structural integrity characteristics after fusion. Existing intelligent situational awareness technologies based on multi-source data fusion cannot adjust the target area integrity determination in situational awareness based on the degree of boundary closure difference in the case of consistent target area existence but inconsistent boundary closure during multi-source data fusion. They still mainly rely on the region's existence for judgment, which can cause broken boundary areas to be misjudged as complete targets, leading to errors in object integrity in the situational awareness results and affecting the accuracy of subsequent behavior analysis and state judgment.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent situational awareness method based on multi-source data fusion to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent situational awareness method based on multi-source data fusion, specifically comprising the following steps: S1. Locate the target area in the visible light data and infrared data, extract the region existence sequence and boundary change trajectory, construct the existence discrimination sequence and the closure evolution sequence, and generate the closure trend indicator based on the boundary change direction to determine whether there is a situation where the existence of the target area is consistent but the boundary closure is inconsistent during the multi-source data fusion process. S2. In the case of consistent existence of target regions but inconsistent boundary closure, perform path extension tracking and fracture propagation analysis on the closed evolution sequence to form a closed failure sequence. Combine the existence discrimination sequence to calculate the closure attenuation parameter and the expansion gradient parameter to characterize the degree of difference in boundary closure. S3. Couple the closed attenuation parameter with the extended gradient parameter to generate a consistent offset parameter by introducing a consistent intensity, and construct the target region integrity evaluation coefficient through nonlinear mapping. S4. Perform hierarchical mapping based on the integrity evaluation coefficient of the target area to form complete intervals, transition intervals and damaged intervals, and classify and determine the integrity of the target area in combination with the degree of difference in boundary closure. S5. Track the temporal evolution sequence of boundary closure differences, continuously update the target region integrity evaluation coefficient, and simultaneously adjust the hierarchical mapping interval to complete the dynamic control of target region integrity determination.

[0007] Preferably, S1 specifically includes the following steps: S101. The target area in the visible light data and infrared data is located accordingly. By performing spatial coordinate mapping and time series alignment, and combining the regional response consistency of pixel grayscale distribution and temperature distribution, the corresponding position of the target area in the visible light data and infrared data is determined. The pixel set is extracted at the corresponding position, and the pixel response change is calculated along the time series to form the regional existence sequence. At the same time, the gradient change path is extracted along the boundary of the pixel set to form the boundary change trajectory. S102. Utilize the region existence sequence to perform continuous consistency encoding to generate an existence discrimination sequence, and perform path connection and direction continuity analysis on the boundary change trajectory to form a closed evolution sequence. The closed evolution sequence is used to describe the evolution relationship of the boundary connection state over time. S103. Perform vectorized statistics on the boundary change direction in the closed evolution sequence to form a direction distribution result, and generate a closure trend indicator based on the direction distribution result; jointly determine the existence of the discrimination sequence and the closure trend indicator. When the existence of the discrimination sequence is consistent and the closure trend indicator corresponds to a non-closed trend, it is determined that there is a situation in the multi-source data fusion process where the existence of the target region is consistent but the boundary closure is inconsistent.

[0008] Preferably, S102 specifically is: The region response value in each time frame is differentially calculated using the region existence sequence, and a response change matrix is ​​constructed in chronological order. Threshold segmentation encoding is then performed on the response change matrix to map the continuous change state into a consistent state label sequence. By accumulating the consistent state label sequence frame by frame, an existence discrimination sequence is formed. Neighborhood connectivity calculations are performed on the boundary pixel paths in the boundary change trajectory. By constraining and matching the spatial distance and gradient changes between adjacent boundary pixels, a set of boundary connectivity paths is constructed. The set of boundary connectivity paths is then spliced ​​together to form a boundary connectivity sequence. Based on the boundary connection sequence, the directional changes of each connection path are calculated point by point, the directional continuity changes are statistically analyzed in chronological order, a directional continuity sequence is constructed, and the directional continuity sequence is jointly mapped with the boundary connection sequence to form a closed evolution sequence, which is used to describe the evolutionary relationship of the boundary connection state over time.

[0009] Preferably, S2 specifically includes the following steps: S201. When there is a consistent existence of target regions but inconsistent boundary closure, path extension tracing and fracture propagation analysis are performed on the closed evolution sequence. Path extension tracing extends the connecting paths in the closed evolution sequence frame by frame in chronological order and matches the positions of the path endpoints in adjacent time frames to record the path continuity status. Fracture propagation analysis identifies the interruption positions in the connecting paths and continuously marks the expansion range of the interruption positions in subsequent time frames to form path extension results and fracture propagation records. S202. Based on the path extension results and fracture propagation records, the connection paths in the closed evolution sequence are segmented, and the continuously extended path segments are distinguished from the interrupted path segments. The connection states of each path segment are integrated in chronological order to form a closed failure sequence, which is used to represent the failure changes of the boundary connection state. S203. For the path continuity changes in the closed failure sequence, combined with the regional existence state in the existence discrimination sequence, the length of the continuous path segment and the fracture propagation range are statistically calculated to obtain the closure attenuation parameter and the expansion gradient parameter, and the degree of boundary closure difference is characterized according to the combination relationship between the closure attenuation parameter and the expansion gradient parameter.

[0010] Preferably, S203 is as follows: To address the changes in path continuity in a closed broken sequence, the length of continuous path segments within each time frame is statistically analyzed, and a path length change sequence is constructed in chronological order. Simultaneously, the expansion range of the path interruption location in adjacent time frames is extracted to form a fracture propagation range sequence. By combining the regional existence state in the existence discrimination sequence, joint constraint statistics are performed on the path length change sequence and the fracture propagation range sequence. Within the time interval where the regional existence state remains consistent, the change amplitude of the continuous path length is cumulatively calculated, and the gradient of the expansion amplitude of the fracture propagation range is statistically analyzed to generate closure attenuation parameters and expansion gradient parameters. Based on the combination relationship between the closure decay parameter and the extended gradient parameter, the closure decay parameter is classified into levels according to its numerical range, and the extended gradient parameter is segmented according to its changing trend. Within the same time interval, the classification results of the closure decay parameter and the segmentation marks of the extended gradient parameter are matched accordingly. A multi-level mapping table is constructed according to the matching combination, and different difference levels are assigned to different matching combinations according to the mapping table to form the boundary closure difference characterization results.

[0011] Preferably, S3 specifically includes the following steps: S301. Couple the closed-loop decay parameter and the extended gradient parameter by arranging the closed-loop decay parameter and the extended gradient parameter in chronological order, normalizing the values ​​of the closed-loop decay parameter and the extended gradient parameter within the same time interval, and weighting the normalized closed-loop decay parameter and the extended gradient parameter according to the preset weight to form a sequence of coupled operation results. S302. Based on the coupling operation result sequence, an existence consistency strength is introduced. The existence consistency strength is generated by statistically calculating the proportion of consistent states in continuous time intervals in the existence discrimination sequence. The existence consistency strength is then applied to the coupling operation result sequence according to the time correspondence, and the coupling operation result sequence is offset and adjusted to generate a consistency offset parameter. S303. Construct a nonlinear mapping relationship for the uniform offset parameter. By segmenting the uniform offset parameter according to the numerical range, perform mapping calculations in each segment according to the nonlinear combination relationship, and integrate the mapping results in chronological order to form the target area integrity evaluation coefficient.

[0012] Preferably, S302 is as follows: Based on the result sequence of the coupling operation, the existence discrimination sequence is divided into time intervals, the label sequence of consistent states in continuous time intervals is extracted, and the number of frames with consistent states in each time interval and the total number of frames are statistically calculated to form a consistent state proportion sequence to generate the existence consistency strength. The existence consistency intensity is aligned frame by frame with the coupling operation result sequence according to the time correspondence. The coupling operation result and the existence consistency intensity at the corresponding time point in each time frame are combined. By applying a proportional adjustment coefficient to the coupling operation result, an offset processing sequence is formed. The offset processing sequence is continuously integrated in chronological order. The offset-adjusted values ​​in each time frame are serialized and arranged to generate a consistent offset parameter. The consistent offset parameter is used to characterize the offset change of the coupling operation result under the influence of consistent strength.

[0013] Preferably, S4 is as follows: Based on the integrity evaluation coefficient of the target area, a hierarchical mapping is performed. The integrity evaluation coefficient of the target area is extracted in chronological order and divided into segments according to a preset numerical interval. The integrity evaluation coefficient of the target area is then mapped to the corresponding interval position to form a hierarchical mapping result. Based on the results of the hierarchical mapping, the integrity evaluation coefficient of the target area is classified according to the numerical range of each interval. Intervals with high values ​​are marked as complete intervals, intervals with medium values ​​are marked as transitional intervals, and intervals with low values ​​are marked as damaged intervals, thus forming complete intervals, transitional intervals, and damaged intervals. After forming complete, transitional, and damaged intervals, the degree of boundary closure difference is matched frame by frame with the interval classification results according to the time correspondence. The degree of boundary closure difference within the same time interval is classified into levels, and the classified difference levels are combined with the corresponding interval categories. Based on the correspondence between the interval category and the difference level in the combined tags, the target area is classified and determined. When the interval category is a complete interval and the difference level is low, it is determined to be in a complete state. When the interval category is a transitional interval or the difference level is medium, it is determined to be in a transitional state. When the interval category is a damaged interval or the difference level is high, it is determined to be in a damaged state, thus completing the classification and determination of the integrity of the target area.

[0014] Preferably, S5 is as follows: The temporal evolution sequence of boundary closure differences is tracked and processed. The degree of boundary closure difference in each time frame is extracted in chronological order, and the difference changes between adjacent time frames are continuously calculated to form a difference change sequence. The target area integrity evaluation coefficient is continuously updated based on the difference change sequence. The target area integrity evaluation coefficient in each time frame is numerically corrected by introducing the difference change amount at the corresponding time point to form a continuously updated target area integrity evaluation coefficient. Based on the continuous updating of the target area integrity evaluation coefficient, the hierarchical mapping interval is adjusted synchronously. By matching the difference change sequence with the hierarchical mapping interval according to the time correspondence, the numerical boundaries of each interval are dynamically adjusted. Based on the adjusted hierarchical mapping interval, the continuously updated target area integrity evaluation coefficient is remapped to complete the dynamic control of the target area integrity judgment.

[0015] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention introduces a boundary closure difference modeling mechanism during multi-source data fusion, extending the traditional situational awareness method based solely on regional existence to a comprehensive judgment system that considers the integrity of boundary structures. This allows for refined characterization and quantitative analysis of boundary changes even when target regions exhibit consistent existence but inconsistent boundary closure. By constructing closure evolution sequences, closure failure sequences, closure attenuation parameters, and extended gradient parameters, it achieves a dynamic description of boundary fracturing behavior and its propagation process. Furthermore, through coupling operations and nonlinear mapping, the difference information is transformed into target region integrity evaluation coefficients. This enables the situational awareness results to directly reflect the impact of boundary structure changes on target integrity, avoiding misjudging regions with fractured boundaries but overall existence as intact targets, thus improving the accuracy and robustness of situational awareness in complex scenarios.

[0016] 2. This invention achieves dynamic control of target area integrity assessment by hierarchically mapping the target area integrity evaluation coefficient and classifying it based on the degree of boundary closure difference. Simultaneously, it introduces a time evolution sequence of boundary closure differences to continuously update the evaluation coefficients and synchronously adjusts the hierarchical mapping interval. This enables the situational awareness process to adaptively adjust to changes in boundary structure. This technical solution not only accurately identifies the integrity, transitional, and damaged states of the target area at different time stages but also continuously tracks the evolution trend of boundary closure capability. This eliminates structural misjudgment caused by differences in imaging modalities during multi-source data fusion for intelligent situational awareness, improving the reliability and continuity of subsequent behavior analysis and state judgment. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0018] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0019] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0020] This invention provides, for example Figure 1 The intelligent situational awareness method based on multi-source data fusion shown includes the following steps: S1. Locate the target area in the visible light data and infrared data, extract the region existence sequence and boundary change trajectory, construct the existence discrimination sequence and the closure evolution sequence, and generate the closure trend indicator based on the boundary change direction to determine whether there is a situation where the existence of the target area is consistent but the boundary closure is inconsistent during the multi-source data fusion process. In this embodiment, S1 specifically includes the following steps: S101. The target area in the visible light data and infrared data is located accordingly. By performing spatial coordinate mapping and time series alignment, and combining the regional response consistency of pixel grayscale distribution and temperature distribution, the corresponding position of the target area in the visible light data and infrared data is determined. The pixel set is extracted at the corresponding position, and the pixel response change is calculated along the time series to form the regional existence sequence. At the same time, the gradient change path is extracted along the boundary of the pixel set to form the boundary change trajectory. Visible light and infrared data are spatially aligned using a unified reference coordinate system. By calibrating the imaging parameters of both types of sensors, the pixel positions in the infrared image are transformed into the visible light image coordinate system, achieving spatial coordinate mapping. Simultaneously, the two types of data are time-series aligned based on acquisition time information, ensuring a one-to-one correspondence between image frames at the same moment. After alignment, target regions are determined using image segmentation or region extraction methods. For example, region outlines are determined based on texture changes or brightness distribution in the visible light image, and candidate regions are extracted from the infrared image based on temperature distribution differences. Matching is then performed by comparing the consistency between pixel grayscale change trends and temperature change trends within the two regions; regions with similar trends are identified as the same target region. After determining the corresponding location, a pixel set is extracted within the region, and the numerical changes of each pixel are recorded chronologically, forming a region existence sequence reflecting changes in the region's state. Simultaneously, the difference changes between neighboring pixels at the boundary of the pixel set are calculated, extracting the boundary direction change path. The paths from each time frame are then connected sequentially to form a boundary change trajectory describing the boundary change process.

[0021] Visible light data represents image data reflecting the surface texture and brightness information of an object, while infrared data represents image data reflecting the temperature distribution information of an object. The target region represents the area corresponding to the same actual object in both types of data. Spatial coordinate mapping represents the process of transforming the two types of data into a unified coordinate system. Time series alignment represents the synchronous arrangement of data collected at different times. The regional response consistency of pixel grayscale distribution and temperature distribution represents the degree of consistency between the two regions in terms of brightness and temperature change trends. The corresponding position of the target region in visible light and infrared data represents the spatial mapping position of the same object in both types of data. The pixel set represents all the pixels constituting the target region. Pixel response change represents the process of pixel values ​​changing over time. The region existence sequence represents the change in the existence state of the target region over continuous time. The pixel set boundary represents the edge range of the target region. The gradient change path represents the directional path formed by the pixel difference change at the boundary. The boundary change trajectory represents the set of paths formed by the boundary change over time.

[0022] S102. Utilize the region existence sequence to perform continuous consistency encoding to generate an existence discrimination sequence, and perform path connection and direction continuity analysis on the boundary change trajectory to form a closed evolution sequence. The closed evolution sequence is used to describe the evolution relationship of the boundary connection state over time. S103. Perform vectorized statistics on the boundary change direction in the closed evolution sequence to form a direction distribution result, and generate a closure trend indicator based on the direction distribution result; jointly determine the existence of the discrimination sequence and the closure trend indicator. When the existence of the discrimination sequence is consistent and the closure trend indicator corresponds to a non-closed trend, it is determined that there is a situation in the multi-source data fusion process where the existence of the target region is consistent but the boundary closure is inconsistent.

[0023] The closed evolution sequence records the connection state and direction change information of the boundary over a continuous time period. Vectorized statistics can be achieved by extracting the direction vectors of each connection path in each time frame and performing statistical analysis on all direction vectors. For example, the direction vectors can be divided into multiple direction categories according to their angle range, and the frequency of each category can be counted to form a direction distribution result that reflects the overall trend of boundary change. When a certain direction has a significantly higher proportion, it indicates that the boundary change is consistent, while when multiple directions are distributed dispersedly, it indicates that the boundary change is more discrete. Based on the direction distribution result, a closure trend identifier can be further generated. For example, when the direction distribution is concentrated and the change is continuous, it is marked as a closure trend. When the direction distribution is dispersed or there is a significant reverse change, it is marked as a non-closure trend. Subsequently, the closure trend identifier is jointly analyzed with the existence discrimination sequence. For example, if the discrimination sequence is in a stable state within a continuous time period and the closure trend identifier is consistently a non-closure trend, it can be determined that the target area exists during the multi-source data fusion process but the boundary has not formed a closure, thus identifying the situation where the existence is consistent but the boundary closure is inconsistent.

[0024] The boundary change direction in the closed evolution sequence represents the motion direction information of the boundary in each time frame. Vectorized statistics represent the conversion of this directional information into a unified vector form and the classification and statistical processing. The directional distribution result represents the frequency and distribution of different directions. The closure trend indicator represents a judgment result based on the directional distribution on whether the boundary tends to close. The consistency of the existence discrimination sequence indicates that the target region is in a stable state in continuous time. The non-closure trend corresponding to the closure trend indicator indicates that the boundary change direction is not concentrated or is obviously dispersed, reflecting that the boundary has not formed a continuous closure. The non-closure trend indicates that the boundary is broken or the directional change is discontinuous. The consistent existence of the target region indicates that the region is identified as existing in different data sources. The inconsistent boundary closure indicates that the boundary structure shows differences between closed and non-closed in different data. These information are combined to determine whether there is inconsistency in the structural state of the target region in multi-source data fusion.

[0025] In this embodiment, S102 specifically refers to: The region response value in each time frame is differentially calculated using the region existence sequence, and a response change matrix is ​​constructed in chronological order. Threshold segmentation encoding is then performed on the response change matrix to map the continuous change state into a consistent state label sequence. By accumulating the consistent state label sequence frame by frame, an existence discrimination sequence is formed. A region existence sequence is used to describe the response changes of a target region over continuous time. The region response value in each time frame can be obtained by statistically analyzing the grayscale or temperature values ​​of all pixels within the target region, for example, by taking the average or weighted average to represent the overall response level of the region in the current time frame. After obtaining the region response values ​​of consecutive time frames, the response values ​​between adjacent time frames are differentially calculated, that is, the magnitude and direction of change between consecutive frames are calculated, and these change results are arranged in chronological order to form a response change matrix reflecting the region response change process. Based on this, the response change matrix is ​​subjected to threshold segmentation encoding, that is, according to a preset change range... The response changes are divided into multiple intervals. For example, smaller changes correspond to a stable state, while larger changes correspond to a changing state. Continuous changing states are mapped to discrete, consistent state label sequences. Subsequently, the consistent state label sequences are accumulated frame by frame. That is, the labeling results of consecutive frames are accumulated or statistically analyzed in chronological order to obtain an existence discrimination sequence that reflects whether the region is stable or changing over a period of time. For example, if multiple consecutive frames are in a stable state, the region is determined to exist continuously, while the appearance of a sudden change label indicates that the region's state has changed. Through this processing, the original continuous change information can be transformed into a discrete sequence that can be used to determine the existence of the region, improving the stability and consistency of subsequent analysis.

[0026] Neighborhood connectivity calculations are performed on the boundary pixel paths in the boundary change trajectory. By constraining and matching the spatial distance and gradient changes between adjacent boundary pixels, a set of boundary connectivity paths is constructed. The set of boundary connectivity paths is then spliced ​​together to form a boundary connectivity sequence. The boundary pixel paths in the boundary change trajectory represent the positional changes of the target region boundary over continuous time. Each path consists of a series of boundary pixels arranged in chronological order. Neighborhood connectivity calculation involves setting a fixed-range neighborhood window around each boundary pixel, searching for boundary pixels with similar positions in adjacent time frames within this window, and calculating the spatial distance between adjacent boundary pixels as well as the magnitude and direction of gradient changes at their corresponding positions. Spatial distance measures the proximity of pixel positions, while gradient change reflects the consistency of boundary orientation. Constraint matching involves simultaneously setting spatial distance and gradient change thresholds. When two boundary pixels are sufficiently close in spatial position and their gradient change directions are consistent, the matching is performed. When points are continuous, they are identified as connectable points. For example, if the boundary shifts to the right and the gradient direction is consistent in two consecutive frames, the corresponding pixels are connected. After point-by-point matching, all connection relationships that meet the constraints are summarized to form a set of boundary connection paths. Each path represents a continuous boundary evolution trajectory. Then, continuous path splicing is performed on these paths in chronological order, that is, path segments that are temporally adjacent and have the same connection direction are merged to construct a longer continuous boundary trajectory, which finally forms a boundary connection sequence to describe the continuous changes of the boundary in space and time. This transforms the originally discrete boundary change points into continuous boundary evolution relationships, providing a stable basis for subsequent analysis of boundary closure.

[0027] Based on the boundary connection sequence, the directional changes of each connection path are calculated point by point, the directional continuity changes are statistically analyzed in chronological order, a directional continuity sequence is constructed, and the directional continuity sequence is jointly mapped with the boundary connection sequence to form a closed evolution sequence, which is used to describe the evolutionary relationship of the boundary connection state over time.

[0028] The boundary connection sequence consists of multiple connection paths arranged in chronological order. Each connection path contains continuous boundary pixel positions, so the directional changes of each connection path can be characterized by the positional changes between adjacent pixels. Point-by-point vector calculation represents the coordinate difference between two adjacent pixels in the path as a direction vector. For example, if the path moves from left to right, the direction vector points horizontally. By calculating point by point along the entire path, a sequence of direction vectors can be obtained. The change in directional continuity is achieved by comparing the changes in direction vectors between adjacent time frames. Small directional changes indicate strong boundary continuity, while abrupt directional changes indicate a break or inflection point in the boundary. These changes are recorded in chronological order. A directional continuity sequence is formed; the boundary connection sequence provides spatial connectivity, while the directional continuity sequence provides directional change information. The joint mapping associates and encodes the path connection state in each time frame with the corresponding directional continuity, for example, combining the connection state and the degree of directional stability into a unified identifier, thereby constructing a closed evolution sequence. The closed evolution sequence is used to describe whether the boundary remains continuously closed. For example, when a path is directionally stable and has complete connectivity in continuous time, it indicates that the boundary tends to be closed, while when there is a sudden change in direction or a break in the connection in the path, it indicates that the boundary is broken. Through this processing, spatial connectivity information and directional change information can be fused to form a continuous description that can reflect the boundary closure change process.

[0029] S2. In the case of consistent existence of target regions but inconsistent boundary closure, perform path extension tracking and fracture propagation analysis on the closed evolution sequence to form a closed failure sequence. Combine the existence discrimination sequence to calculate the closure attenuation parameter and the expansion gradient parameter to characterize the degree of difference in boundary closure. In this embodiment, S2 specifically includes the following steps: S201. When there is a consistent existence of target regions but inconsistent boundary closure, path extension tracing and fracture propagation analysis are performed on the closed evolution sequence. Path extension tracing extends the connecting paths in the closed evolution sequence frame by frame in chronological order and matches the positions of the path endpoints in adjacent time frames to record the path continuity status. Fracture propagation analysis identifies the interruption positions in the connecting paths and continuously marks the expansion range of the interruption positions in subsequent time frames to form path extension results and fracture propagation records. Closed evolution sequences are used to describe the changes in boundary connectivity over time, where each connection path represents the spatial continuity of the boundary in consecutive time frames. Path extension tracking is achieved by matching the connection paths frame by frame in each time frame. That is, the path endpoint position is obtained in the current time frame, and the boundary point with the same spatial position and direction is searched in the next time frame as the continuation point. For example, in a scenario where the target boundary moves slowly, the path endpoint will undergo a small displacement in adjacent time frames. Matching can be completed by setting the spatial neighborhood range and direction consistency conditions, thereby recording the path continuity state. Break propagation analysis is achieved by detecting discontinuous positions in the connection paths. When the path cannot be continued in a certain time frame, the position is marked as an interruption position, and the new boundary point and its expansion range are continuously tracked in subsequent time frames. For example, after the boundary breaks due to occlusion or blurring, the broken area gradually expands or moves in subsequent frames. By continuously marking these changes, a break propagation record is formed. Through path extension results and break propagation records, the evolution process of the boundary from continuity to breakage can be completely described, thus providing a basis for subsequent difference analysis.

[0030] The closed evolution sequence represents the set of changes in the boundary connection state over time; the connection path represents the connection relationship of boundary pixels in consecutive time frames; path extension tracking represents the process of continuously tracking the connection path along the time direction; frame-by-frame extension represents the continuation matching of the path in each time frame; the path endpoint represents the termination position of the connection path in the current time frame; the path continuation state represents whether the path can continue to extend in adjacent time frames; the interruption position in the connection path represents the discontinuity point of the path in a certain time frame; the expansion range of the interruption position in subsequent time frames represents the temporal expansion of the discontinuous region; the continuous marker represents the continuous recording of the interruption position and its expansion process; the path extension result represents the set of continuations of all paths over time; the break propagation record represents the set of records of the change of the interruption position over time. All of this information is used to characterize the process of the boundary changing from continuous to broken.

[0031] S202. Based on the path extension results and fracture propagation records, the connection paths in the closed evolution sequence are segmented, and the continuously extended path segments are distinguished from the interrupted path segments. The connection states of each path segment are integrated in chronological order to form a closed failure sequence, which is used to represent the failure changes of the boundary connection state. The path extension results record the continuity of each connected path in consecutive time frames, while the break propagation record reflects the expansion process of the break location in the time dimension. Based on these two types of information, the connected paths can be segmented. That is, the path is checked frame by frame along the time sequence to see if it remains continuous. When the path endpoints can be matched and continued in adjacent time frames, the corresponding segment is divided into a continuously extended path segment. When the path cannot be matched in a certain time frame, that position is used as the dividing point to divide it into a path segment where an interruption occurs. For example, if a path exists continuously in the first three frames and breaks in the fourth frame, then the first three frames constitute a continuously extended path segment, and the fourth frame and its subsequent extended area constitute an interrupted path segment. After the segmentation is completed, the path segments are integrated in chronological order. The connection status of each path segment in each time frame is recorded uniformly, for example, marked as continuous or interrupted, and arranged in time to form a sequence, thereby constructing a closed break sequence to reflect the change process of the boundary from continuous to broken.

[0032] Segmentation refers to the process of dividing the connecting path according to changes in path continuity. Continuously extended path segments represent path fragments that maintain connectivity across multiple consecutive time frames, while interrupted path segments represent path fragments whose connectivity is broken in a specific time frame. Differentiation refers to classifying and labeling different types of path segments. The connectivity status of each path segment indicates the connectivity or interruption status of the path in the corresponding time frame. The closure breakdown sequence represents a set of sequences formed by integrating all path segments in chronological order, containing information on changes in the continuity and interruption status of the paths. This sequence can describe the breakdown evolution of the boundary connection structure in the time dimension, providing a basis for subsequent analysis of boundary closure differences.

[0033] S203. For the path continuity changes in the closed failure sequence, combined with the regional existence state in the existence discrimination sequence, the length of the continuous path segment and the fracture propagation range are statistically calculated to obtain the closure attenuation parameter and the expansion gradient parameter, and the degree of boundary closure difference is characterized according to the combination relationship between the closure attenuation parameter and the expansion gradient parameter.

[0034] In this embodiment, S203 specifically refers to: To address the changes in path continuity in a closed broken sequence, the length of continuous path segments within each time frame is statistically analyzed, and a path length change sequence is constructed in chronological order. Simultaneously, the expansion range of the path interruption location in adjacent time frames is extracted to form a fracture propagation range sequence. The closed-loop sequence records the continuity and interruption information of the lower boundary connection state in each time frame. By traversing the path segments in each time frame, the number of pixels or path length of continuous path segments can be counted to achieve length statistics. For example, for a continuous path, the number of boundary pixels or cumulative spatial distance can be counted, and the statistical results of each frame can be arranged in chronological order to construct a path length change sequence, thus reflecting the shrinking or extending trend of the path over time. Simultaneously, for path interruption locations, a search is conducted in adjacent time frames to see if a new interruption point or extended region appears nearby. By calculating the spatial expansion range of the interrupted region, such as recording the number of pixels covered by the interrupted region or the change in the expansion direction in the next frame, these changes are arranged in chronological order. A fracture propagation range sequence is formed. For example, if a fracture point gradually expands in subsequent frames, the corresponding expansion range value will continue to increase. By simultaneously analyzing the path length change sequence and the fracture propagation range sequence, the process of weakening boundary continuity and fracture spread can be characterized, providing a quantitative basis for subsequent closure difference analysis. Among them, the path continuity change in the closure failure sequence represents the transition between continuity and interruption of the path in the time dimension. The length statistics represent the quantification of the scale of continuous path segments. The path length change sequence represents the process of path length changing over time. The expansion range of the path interruption position in adjacent time frames represents the spatial expansion of the fracture area in the time dimension. The fracture propagation range sequence represents the set of changes formed by this expansion in chronological order.

[0035] By combining the regional existence state in the existence discrimination sequence, joint constraint statistics are performed on the path length change sequence and the fracture propagation range sequence. Within the time interval where the regional existence state remains consistent, the change amplitude of the continuous path length is cumulatively calculated, and the gradient of the expansion amplitude of the fracture propagation range is statistically analyzed to generate closure attenuation parameters and expansion gradient parameters. The path length change sequence reflects the changes in the continuous path at the boundary in the time dimension, while the fracture propagation range sequence reflects the expansion of the boundary interruption region in the time dimension. Combining the region existence status in the existence discrimination sequence, time intervals where the region existence status remains consistent can be selected first. Within these time intervals, the path length change sequence can be cumulatively analyzed. For example, the magnitude of the continuous path length change across multiple frames can be statistically analyzed, and the length change in each frame can be accumulated chronologically to form the magnitude of the continuous path length change. Simultaneously, the expansion range sequence can be analyzed, for example, by recording the expansion range change of the fracture region in consecutive frames and calculating the expansion difference between adjacent time frames. These differences can be statistically analyzed frame by frame to form the expansion magnitude change result. Then, based on the change trend, gradient statistics can be performed on the expansion magnitude, for example, to distinguish between accelerated or decelerated expansion. By simultaneously considering the changes in continuous path length and fracture expansion, two interrelated quantities can be obtained. A continuously decreasing path length and a continuously expanding fracture range indicate a weakening boundary closure capability, which can be used to generate closure attenuation parameters and expansion gradient parameters, providing a quantitative basis for subsequent closure difference determination.

[0036] The existence discrimination sequence indicates whether the target region remains stable within a continuous time frame; the path length change sequence indicates the variation of continuous path length in different time frames; the fracture propagation range sequence indicates the expansion of the interrupted region in the time dimension; joint constraint statistics indicate that, under the premise of maintaining a consistent existence state, the path length change and fracture expansion are statistically analyzed simultaneously; the time interval in which the existence state remains consistent indicates the time range within which the existence state of the region does not change in multiple consecutive frames; the magnitude of the change in continuous path length indicates the degree of increase or decrease in path length in the time series; cumulative calculation indicates the superposition and statistical analysis of the magnitude of change in multiple time frames; the expansion magnitude of the fracture propagation range indicates the degree of spatial expansion of the fracture region; gradient statistics indicate the analysis of the trend of the expansion magnitude in time; the closure attenuation parameter indicates the degree of weakening of path continuity; the expansion gradient parameter indicates the trend of fracture region expansion; these two parameters are used together to characterize the change in boundary closure capability.

[0037] Based on the combination relationship between the closure decay parameter and the extended gradient parameter, the closure decay parameter is classified into levels according to its numerical range, and the extended gradient parameter is segmented according to its changing trend. Within the same time interval, the classification results of the closure decay parameter and the segmentation marks of the extended gradient parameter are matched accordingly. A multi-level mapping table is constructed according to the matching combination, and different difference levels are assigned to different matching combinations according to the mapping table to form the boundary closure difference characterization results.

[0038] The closure attenuation parameter reflects the degree of weakening of path continuity, while the expansion gradient parameter reflects the trend of fracture expansion. Combining the two can characterize the overall change state of the boundary from closure to fracture. In practical implementation, the closure attenuation parameter can first be classified according to its value range. For example, cases with small changes in continuity can be classified as low attenuation intervals, and cases with significant changes as high attenuation intervals. Simultaneously, the expansion gradient parameter can be segmented according to its trend, for example, slow expansion, stable expansion, and accelerated expansion can be labeled as different categories. Within the same time interval, the classification results of the closure attenuation parameter and the segmentation labels of the expansion gradient parameter are combined accordingly. For example, if the path length is significantly shortened and fracture expansion accelerates within a certain time period, a specific combination relationship is formed. A multi-level mapping table is established for all possible combinations, mapping different combinations to different levels of difference. For example, weak continuity and accelerated expansion correspond to a high level of difference, while stable continuity and slow expansion correspond to a low level of difference, thus forming a characterization result of the degree of boundary closure difference. This process can unify changes in two different dimensions into comparable levels of difference, facilitating subsequent judgment.

[0039] The combination relationship between the closure decay parameter and the extended gradient parameter represents the correspondence between changes in path continuity and changes in fracture extension. The numerical range represents the value range for dividing the closure decay parameter. The hierarchical division represents dividing the parameter into multiple levels according to the range. The change trend represents the direction and rate of change of the extended gradient parameter in the time dimension. The segmentation label represents the classification of parameters according to the change trend. The same time interval represents the corresponding analysis of the two types of parameters within the same time range. The hierarchical results of the closure decay parameter and the segmentation label of the extended gradient parameter represent the results after processing the two types of parameters separately. The correspondence matching represents the one-to-one combination of the two types of results according to time. The multi-level mapping table represents the corresponding set between different combination relationships and difference levels. The difference level represents the division of the strength of boundary closure difference. The boundary closure difference degree characterization result represents the final difference evaluation result obtained through combination mapping.

[0040] S3. Couple the closed attenuation parameter with the extended gradient parameter to generate a consistent offset parameter by introducing a consistent intensity, and construct the target region integrity evaluation coefficient through nonlinear mapping. In this embodiment, S3 specifically includes the following steps: S301. Couple the closed-loop decay parameter and the extended gradient parameter by arranging the closed-loop decay parameter and the extended gradient parameter in chronological order, normalizing the values ​​of the closed-loop decay parameter and the extended gradient parameter within the same time interval, and weighting the normalized closed-loop decay parameter and the extended gradient parameter according to the preset weight to form a sequence of coupled operation results. The closure attenuation parameter reflects the degree of weakening of boundary continuity, while the expansion gradient parameter reflects the trend of fracture propagation. Both need to be processed in a corresponding manner over time to achieve joint analysis. Specifically, the closure attenuation parameter and the expansion gradient parameter are first arranged frame-by-frame in chronological order, forming a set of corresponding parameter values ​​at each time point. Then, within the same time interval, the two types of parameters are normalized, that is, parameters with different dimensions or value ranges are uniformly converted to the same scale. For example, by scaling the maximum and minimum ranges, the parameters fall into a unified interval, thus ensuring the comparability of subsequent calculations. After normalization, preset weights are introduced to weightedly combine the two types of parameters. For example, when the boundary fracture has a more significant impact, the expansion gradient parameter is given a higher weight; when the continuity has a more significant impact, the closure attenuation parameter is given a higher weight. The two types of parameters at each time point are superimposed according to their weights to form a new value. Arranging these values ​​in chronological order yields the coupled calculation result sequence. This sequence simultaneously reflects the combined changes in continuity attenuation and fracture propagation. For example, if the closure attenuation is large and the expansion gradient increases within a certain time period, the coupled result value increases significantly, reflecting a significant decrease in boundary closure capability.

[0041] The coupling operation represents the process of jointly calculating the closure decay parameter and the expansion gradient parameter; the corresponding arrangement according to time order means that the two types of parameters are matched one-to-one in the time dimension, so that each time point has a matching relationship; the numerical normalization process means that the parameters with different value ranges are converted to a uniform scale for comparison; the preset weight means that the proportional coefficients are set to determine the degree of influence of different parameters in the combined operation; the weighted combination means that the normalized parameters are superimposed according to the weights; the coupling operation result sequence represents the set of combined calculation results arranged in time order, which comprehensively reflects the joint state of boundary continuity change and fracture expansion change.

[0042] S302. Based on the coupling operation result sequence, an existence consistency strength is introduced. The existence consistency strength is generated by statistically calculating the proportion of consistent states in continuous time intervals in the existence discrimination sequence. The existence consistency strength is then applied to the coupling operation result sequence according to the time correspondence, and the coupling operation result sequence is offset and adjusted to generate a consistency offset parameter. S303. Construct a nonlinear mapping relationship for the uniform offset parameter. By segmenting the uniform offset parameter according to the numerical range, perform mapping calculations in each segment according to the nonlinear combination relationship, and integrate the mapping results in chronological order to form the target area integrity evaluation coefficient.

[0043] The uniform offset parameter reflects the overall offset change of the coupling operation result under the influence of uniform strength. This parameter can be transformed into an evaluation quantity that can be used for judgment by constructing a nonlinear mapping relationship. In specific implementation, the uniform offset parameter is first segmented according to its numerical range. For example, the part with small offset change is divided into a low interval, and the part with significant change is divided into a high interval. Then, different nonlinear combination relationships are used for mapping calculation in each interval. For example, a gradual mapping is used in the interval with small offset change to keep the numerical change stable, and an enhanced mapping is used in the interval with large offset change to further amplify the numerical difference, thereby distinguishing different degrees of offset state. After obtaining the mapping results of each time frame, they are integrated in chronological order. For example, the mapping results of consecutive time frames are smoothly connected or trend accumulated to form a continuously changing evaluation sequence. This sequence is the target area integrity evaluation coefficient. For example, when the uniform offset parameter continues to increase in a certain period of time, the mapping result increases accordingly, indicating that the integrity of the target area is gradually decreasing. Through this processing, the complex offset change can be transformed into a continuous evaluation index that can be used for judgment.

[0044] Nonlinear mapping represents a non-linear correspondence between input parameters and output results; segmented processing according to numerical intervals means dividing the data into different ranges based on the value range of the consistent offset parameter and processing the data in different ranges separately; nonlinear combination relationship means using different combination methods to transform parameters within each interval; mapping calculation means converting the consistent offset parameter into a new value according to the set relationship; integrating the mapping results according to time order means arranging the calculation results of each time frame in time and processing them continuously; the target area integrity evaluation coefficient represents the final value used to characterize the structural integrity of the target area, which comprehensively reflects the impact of offset changes on area integrity.

[0045] In this embodiment, S302 specifically refers to: Based on the result sequence of the coupling operation, the existence discrimination sequence is divided into time intervals, the label sequence of consistent states in continuous time intervals is extracted, and the number of frames with consistent states in each time interval and the total number of frames are statistically calculated to form a consistent state proportion sequence to generate the existence consistency strength. The existence discrimination sequence reflects the existence state of a target region in the time dimension. Continuous state segments can be identified by dividing this sequence into time intervals. Specifically, the existence discrimination sequence can be traversed chronologically. When the state labels of multiple consecutive time frames are consistent, these frames are divided into the same time interval. For example, several consecutive frames showing an existence state constitute a continuous time interval. Within each time interval, the corresponding consistent state label sequence is extracted, and the number of frames with consistent states within that interval is statistically calculated against the total number of frames. For example, the number of frames marked as existence within a continuous time interval is counted and compared with the total number of frames in that interval to obtain the proportion of consistent states. The proportions of consistent states in each time interval are arranged chronologically to form the consistent state proportion sequence. This sequence is used to reflect the stability of the state of the target region in different time intervals. For example, a proportion close to the whole indicates that the region exists continuously in the interval, while a low proportion indicates that the state is unstable. This process can transform discrete state information into a continuous stability measure, thereby generating the consistency strength of existence, which provides a weight basis for subsequent coupling operations. The time interval division represents the segmentation of the time series. The label sequence of consistent states in continuous time intervals represents the set of labels that maintain the same state in consecutive frames. The number of frames with consistent states in each time interval and the total number of frames represent the statistical objects. The consistency state proportion sequence represents the result of arranging the proportion of each interval by time. The consistency strength of existence represents the stability of the target region in the time dimension.

[0046] The existence consistency intensity is aligned frame by frame with the coupling operation result sequence according to the time correspondence. The coupling operation result and the existence consistency intensity at the corresponding time point in each time frame are combined. By applying a proportional adjustment coefficient to the coupling operation result, an offset processing sequence is formed. Aligning the existence consistency intensity with the coupling operation result sequence frame by frame according to the time correspondence can be achieved through a unified time index. That is, based on the time order, the existence consistency intensity sequence and the coupling operation result sequence are mapped one-to-one on the same time axis, ensuring that each time frame has a matching set of values. After frame-by-frame alignment, the coupling operation results and the existence consistency intensity at the corresponding time point within each time frame are combined. For example, the existence consistency intensity can be used as an adjustment factor on the coupling operation results, applying a proportional adjustment coefficient to the coupling operation results to shift the values ​​according to the stability of the existence state. For instance, during periods of high existence consistency intensity, the coupling operation results are kept stable. If the change amplitude is small, but the consistency strength is low in a time period, the change of the coupling operation result is amplified to form an offset processing sequence. This sequence is used to reflect the dynamic adjustment of the coupling operation result under the existence state constraint. The time correspondence represents the consistency matching relationship between the two sequences in the time dimension, the frame-by-frame alignment represents the one-to-one matching processing according to the time order, the consistency strength between the coupling operation result in each time frame and the corresponding time point represents the two numerical sources involved in the combination, the combination processing represents the joint calculation of the two, the proportional adjustment coefficient represents the weight factor used to adjust the change amplitude of the coupling operation result, and the offset processing sequence represents the set of numerical values ​​arranged in time order after adjustment.

[0047] The offset processing sequence is continuously integrated in chronological order. The offset-adjusted values ​​in each time frame are serialized and arranged to generate a consistent offset parameter. The consistent offset parameter is used to characterize the offset change of the coupling operation result under the influence of consistent strength.

[0048] The offset processing sequence reflects the adjustment of the coupled operation results within each time frame under a consistent intensity of influence. A unified descriptive result can be formed by continuously integrating this sequence in chronological order. Specifically, using the time index as a reference, the offset-adjusted values ​​within each time frame are extracted sequentially and arranged in chronological order to form a continuous data sequence. During the integration process, the numerical changes between adjacent time frames can be smoothed or accumulated. For example, values ​​with consistent trends within consecutive time frames can be weighted and superimposed, while abrupt changes can be averaged using a neighborhood average, thus ensuring the continuity and stability of the sequence. After integration, this sequence is treated as a single... The offset parameters are output. For example, if the offset value continues to rise within a certain time period, it indicates that the coupling operation result is affected by a continuous enhancement of consistent strength within that time period. In this way, the offset information in discrete time frames can be transformed into a continuously changing parameter sequence to characterize the overall offset change trend. Continuous integration according to time order means continuous combination of data from each time frame in the time dimension. The offset-adjusted values ​​within each time frame represent the coupling result after proportional adjustment. Serialization means organizing these values ​​into a continuous sequence in time order. Consistent offset parameters represent the final set of parameters used to describe the overall offset change.

[0049] S4. Perform hierarchical mapping based on the integrity evaluation coefficient of the target area to form complete intervals, transition intervals and damaged intervals, and classify and determine the integrity of the target area in combination with the degree of difference in boundary closure. In this embodiment, S4 specifically refers to: Based on the integrity evaluation coefficient of the target area, a hierarchical mapping is performed. The integrity evaluation coefficient of the target area is extracted in chronological order and divided into segments according to a preset numerical interval. The integrity evaluation coefficient of the target area is then mapped to the corresponding interval position to form a hierarchical mapping result. The target area integrity evaluation coefficient reflects the change in the structural integrity of the target area over time. A continuous data sequence can be constructed by extracting values ​​from this coefficient in chronological order. Specifically, each time frame is used as a unit; the corresponding evaluation coefficient value is read from each time frame and arranged chronologically to form a time series. Based on this, the sequence is segmented according to a pre-defined numerical range. For example, the range with higher evaluation coefficients is divided into one type of interval, the middle range into another, and the lower range into yet another. Then, the evaluation coefficient corresponding to each time frame is mapped to its corresponding interval, thus forming a hierarchical mapping result. For instance, if the evaluation coefficient remains consistently high within a certain time period, that time period is continuously mapped to the same interval. This method transforms continuously changing values ​​into discrete interval identifiers, making the subsequent judgment process clearer and more stable, while avoiding the fluctuations caused by directly using the original values.

[0050] The target region integrity evaluation coefficient represents a numerical value used to measure the structural integrity of the target region; hierarchical mapping represents the process of dividing continuous values ​​into multiple hierarchical intervals according to certain rules; numerical extraction represents obtaining evaluation coefficient values ​​frame by frame from the time series; preset numerical intervals represent multiple numerical ranges divided according to experience or set standards; segmentation represents the allocation of continuous data to different intervals based on the numerical range; corresponding interval position represents the interval category to which each value belongs after division; the hierarchical mapping result represents the set of intervals corresponding to all time frames, used to describe the hierarchical changes in the target region integrity in the time dimension.

[0051] Based on the results of the hierarchical mapping, the integrity evaluation coefficient of the target area is classified according to the numerical range of each interval. Intervals with high values ​​are marked as complete intervals, intervals with medium values ​​are marked as transitional intervals, and intervals with low values ​​are marked as damaged intervals, thus forming complete intervals, transitional intervals, and damaged intervals. After the hierarchical mapping results are formed, the integrity evaluation coefficient of the target area can be classified according to the numerical range corresponding to each interval. In specific implementation, the numerical range boundary corresponding to each interval is first clarified. For example, the part of the evaluation coefficient in the higher value range is identified as the high range interval, the part in the middle value range is identified as the middle range interval, and the part in the lower value range is identified as the low range interval. Then, the corresponding interval position in each time frame is categorized. For example, when the evaluation coefficient of a certain time frame is mapped to the high range interval, the time frame is categorized as the complete interval. When mapped to the middle range interval, it is categorized as the transition interval. When mapped to the low range interval, it is categorized as the damaged interval. By arranging the interval categorization results of continuous time frames, a time series of complete intervals, transition intervals, and damaged intervals can be formed. For example, a target is initially in the complete interval, then gradually enters the transition interval, and then enters the damaged interval, thus reflecting the trend of the target area integrity changing over time. This method can transform continuous numerical changes into clear category divisions, which is convenient for subsequent classification and judgment.

[0052] The numerical range of each interval represents the boundary of the value interval corresponding to each layer in the hierarchical mapping; interval classification represents the process of classifying the values ​​according to their range; high-range intervals represent intervals with higher evaluation coefficients, corresponding to a more complete structure; complete intervals represent categories formed by high-range intervals; intermediate-range intervals represent intervals with evaluation coefficients in the middle range; transitional intervals represent categories formed by intermediate-range intervals; low-range intervals represent intervals with lower evaluation coefficients, corresponding to a damaged structure; and damaged intervals represent categories formed by low-range intervals. These intervals together constitute a hierarchical description of the integrity status of the target area.

[0053] After forming complete, transitional, and damaged intervals, the degree of boundary closure difference is matched frame by frame with the interval classification results according to the time correspondence. The degree of boundary closure difference within the same time interval is classified into levels, and the classified difference levels are combined with the corresponding interval categories. Based on the correspondence between the interval category and the difference level in the combined tags, the target area is classified and determined. When the interval category is a complete interval and the difference level is low, it is determined to be in a complete state. When the interval category is a transitional interval or the difference level is medium, it is determined to be in a transitional state. When the interval category is a damaged interval or the difference level is high, it is determined to be in a damaged state, thus completing the classification and determination of the integrity of the target area.

[0054] After establishing complete, transitional, and broken intervals, a one-to-one correspondence between the degree of boundary closure difference and the interval classification results can be established based on the time index. Frame-by-frame matching enables synchronous alignment of the two types of data on the same time dimension. After alignment, the degree of boundary closure difference within each time interval is graded, for example, into low, medium, and high levels based on the range of difference values. The difference level of each time frame is then combined with the corresponding interval category. For example, if a time frame contains an interval category of complete and a difference level of low, a combined label is formed. Subsequently, based on the combined label... The correspondence between interval categories and difference levels is used to perform classification judgment. For example, when the interval category is a complete interval and the difference level is low, the target area is judged as complete. When the interval category is a transitional interval or the difference level is medium, the target area is judged as transitional. When the interval category is a damaged interval or the difference level is high, the target area is judged as damaged. In this way, structural integrity evaluation and boundary difference information can be integrated for judgment, thereby improving the accuracy of classification results. For example, if the evaluation coefficient is in the middle range but the difference level is high in a certain period of time, it can be directly judged as damaged.

[0055] The temporal correspondence represents the consistency between the degree of boundary closure difference and the interval classification result in the time dimension; the interval classification result represents the interval category corresponding to the integrity evaluation coefficient of the target area; frame-by-frame matching represents the one-to-one correspondence processing of the two types of data according to the time sequence; the degree of boundary closure difference within the same time interval represents the corresponding difference value within the same time range; the level division represents the level processing according to the size of the difference value; the divided difference level represents the classification result of the degree of difference; the corresponding interval category represents the interval category where the evaluation coefficient is located; the combined label represents the joint identification of the difference level and the interval category; the correspondence between the interval category and the difference level in the combined label represents the combination rule used for judgment; the target area represents the object that needs to be analyzed for integrity; the interval category is a complete interval and the difference level is at a low level, corresponding to a complete state; the interval category is a transitional interval or the difference level is at a medium level, corresponding to a transitional state; the interval category is a damaged interval or the difference level is at a high level, corresponding to a damaged state. These relationships together constitute the basis for the classification and judgment of the integrity of the target area.

[0056] S5. Track the temporal evolution sequence of boundary closure differences, continuously update the target region integrity evaluation coefficient, and simultaneously adjust the hierarchical mapping interval to complete the dynamic control of target region integrity determination.

[0057] In this embodiment, S5 specifically refers to: The temporal evolution sequence of boundary closure differences is tracked and processed. The degree of boundary closure difference in each time frame is extracted in chronological order, and the difference changes between adjacent time frames are continuously calculated to form a difference change sequence. The target area integrity evaluation coefficient is continuously updated based on the difference change sequence. The target area integrity evaluation coefficient in each time frame is numerically corrected by introducing the difference change amount at the corresponding time point to form a continuously updated target area integrity evaluation coefficient. The temporal evolution sequence of boundary closure differences reflects the change process of boundary closure differences over continuous time. This time series can be constructed by extracting the difference values ​​in each time frame. Specifically, the degree of boundary closure difference is read frame by frame in chronological order, and the difference values ​​between adjacent time frames are continuously compared. For example, the change amplitude between the current frame and the previous frame is calculated to form a difference change sequence. After obtaining the difference change sequence, it is time-aligned with the target region integrity evaluation coefficient. The corresponding difference change amount is introduced into each time frame to numerically correct the evaluation coefficient. For example, when the difference change in a certain time frame increases significantly, the corresponding evaluation coefficient is reduced. When the difference change tends to stabilize, the evaluation coefficient is smoothly adjusted, thus forming a continuously updated target region integrity evaluation coefficient. This processing method allows the evaluation coefficient to be adjusted in real time according to the changes in boundary closure differences. For example, as the boundary gradually breaks down, the evaluation coefficient continuously decreases, reflecting the dynamic changes in structural integrity.

[0058] The temporal evolution sequence of boundary closure differences represents the continuous set of changes in boundary closure differences over time. Extracting the degree of boundary closure differences within each time frame in chronological order means collecting the difference values ​​for each frame. Difference changes represent the changes in difference values ​​between adjacent time frames. Continuous calculation means processing the difference changes frame by frame and forming a sequence. The difference change sequence represents the set of difference changes arranged by time. The target region integrity evaluation coefficient represents a value used to measure the structural integrity of the target region. Continuous updating means correcting this value frame by frame over time. Introducing the difference change amount at the corresponding time point means using the difference change as an adjustment basis for the numerical update. Numerical correction means adjusting the original evaluation coefficient. The continuously updated target region integrity evaluation coefficient represents the time series result formed after dynamic correction.

[0059] Based on the continuous updating of the target area integrity evaluation coefficient, the hierarchical mapping interval is adjusted synchronously. By matching the difference change sequence with the hierarchical mapping interval according to the time correspondence, the numerical boundaries of each interval are dynamically adjusted. Based on the adjusted hierarchical mapping interval, the continuously updated target area integrity evaluation coefficient is remapped to complete the dynamic control of the target area integrity judgment.

[0060] After the target region integrity evaluation coefficients are continuously updated, the hierarchical mapping intervals can be synchronously adjusted by introducing a difference change sequence. Specifically, the difference change sequence is first mapped one-to-one with the hierarchical mapping intervals according to the time sequence, so that each time frame is associated with the corresponding interval boundary. Then, the numerical boundaries of each interval are dynamically corrected according to the magnitude of the difference change. For example, when the difference change continues to increase within a certain time period, the original interval boundary is shrunk to make it easier for the evaluation coefficients to enter the lower interval. When the difference change tends to stabilize, the interval boundary is expanded to make the distribution of evaluation coefficients more gradual. After the interval boundary adjustment is completed, the continuously updated target region integrity evaluation coefficients are remapped to the adjusted intervals. For example, the evaluation coefficients that were originally in the transition interval may be reclassified into the damaged interval after the interval shrinkage, thereby realizing the dynamic change of the judgment result. This process can adjust the classification result in real time with the boundary change trend, improving the sensitivity and adaptability of the judgment.

[0061] The continuous updating of the target region integrity evaluation coefficient represents the dynamic change process of the evaluation value in the time dimension; the synchronous adjustment of the hierarchical mapping interval represents the synchronous update of the interval division while the evaluation coefficient changes; the difference change sequence represents the set of values ​​whose boundary closure differences change over time; the time correspondence represents the one-to-one matching relationship of different data in the time dimension; the hierarchical mapping interval represents multiple numerical ranges used to divide the evaluation coefficient; the interval numerical boundary represents the upper and lower limits of each interval; dynamic adjustment represents the change processing of the interval boundary according to the difference change; interval remapping represents the re-division of the updated evaluation coefficient to a new interval range; the dynamic control of the target region integrity judgment represents the continuous updating of the classification results of the target region according to the interval changes.

[0062] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions according to the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means (e.g., infrared, wireless, microwave, etc.). A computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0063] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0064] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0065] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0066] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0067] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0068] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An intelligent situational awareness method based on multi-source data fusion, characterized in that, Specifically, the following steps are included: S1. Locate the target area in the visible light data and infrared data, extract the region existence sequence and boundary change trajectory, construct the existence discrimination sequence and the closure evolution sequence, and generate the closure trend indicator based on the boundary change direction to determine whether there is a situation where the existence of the target area is consistent but the boundary closure is inconsistent during the multi-source data fusion process. S2. In the case of consistent existence of target regions but inconsistent boundary closure, perform path extension tracking and fracture propagation analysis on the closed evolution sequence to form a closed failure sequence. Combine the existence discrimination sequence to calculate the closure attenuation parameter and the expansion gradient parameter to characterize the degree of difference in boundary closure. S3. Couple the closed attenuation parameter with the extended gradient parameter to generate a consistent offset parameter by introducing a consistent intensity, and construct the target region integrity evaluation coefficient through nonlinear mapping. S4. Perform hierarchical mapping based on the integrity evaluation coefficient of the target area to form complete intervals, transition intervals and damaged intervals, and classify and determine the integrity of the target area in combination with the degree of difference in boundary closure. S5. Track the temporal evolution sequence of boundary closure differences, continuously update the target region integrity evaluation coefficient, and simultaneously adjust the hierarchical mapping interval to complete the dynamic control of target region integrity determination.

2. The intelligent situational awareness method based on multi-source data fusion according to claim 1, characterized in that, S1 specifically includes the following steps: S101. The target area in the visible light data and infrared data is located accordingly. By performing spatial coordinate mapping and time series alignment, and combining the regional response consistency of pixel grayscale distribution and temperature distribution, the corresponding position of the target area in the visible light data and infrared data is determined. The pixel set is extracted at the corresponding position, and the pixel response change is calculated along the time series to form the regional existence sequence. At the same time, the gradient change path is extracted along the boundary of the pixel set to form the boundary change trajectory. S102. Utilize the region existence sequence to perform continuous consistency encoding to generate an existence discrimination sequence, and perform path connection and direction continuity analysis on the boundary change trajectory to form a closed evolution sequence. The closed evolution sequence is used to describe the evolution relationship of the boundary connection state over time. S103. Perform vectorized statistics on the boundary change direction in the closed evolution sequence to form a direction distribution result, and generate a closure trend indicator based on the direction distribution result; jointly determine the existence of the discrimination sequence and the closure trend indicator. When the existence of the discrimination sequence is consistent and the closure trend indicator corresponds to a non-closed trend, it is determined that there is a situation in the multi-source data fusion process where the existence of the target region is consistent but the boundary closure is inconsistent.

3. The intelligent situational awareness method based on multi-source data fusion according to claim 2, characterized in that, S102 specifically refers to: The region response value in each time frame is differentially calculated using the region existence sequence, and a response change matrix is ​​constructed in chronological order. Threshold segmentation encoding is then performed on the response change matrix to map the continuous change state into a consistent state label sequence. By accumulating the consistent state label sequence frame by frame, an existence discrimination sequence is formed. Neighborhood connectivity calculations are performed on the boundary pixel paths in the boundary change trajectory. By constraining and matching the spatial distance and gradient changes between adjacent boundary pixels, a set of boundary connectivity paths is constructed. The set of boundary connectivity paths is then spliced ​​together to form a boundary connectivity sequence. Based on the boundary connection sequence, the directional changes of each connection path are calculated point by point, the directional continuity changes are statistically analyzed in chronological order, a directional continuity sequence is constructed, and the directional continuity sequence is jointly mapped with the boundary connection sequence to form a closed evolution sequence, which is used to describe the evolutionary relationship of the boundary connection state over time.

4. The intelligent situational awareness method based on multi-source data fusion according to claim 1, characterized in that, S2 specifically includes the following steps: S201. When there is a consistent existence of target regions but inconsistent boundary closure, path extension tracing and fracture propagation analysis are performed on the closed evolution sequence. Path extension tracing extends the connecting paths in the closed evolution sequence frame by frame in chronological order and matches the positions of the path endpoints in adjacent time frames to record the path continuity status. Fracture propagation analysis identifies the interruption positions in the connecting paths and continuously marks the expansion range of the interruption positions in subsequent time frames to form path extension results and fracture propagation records. S202. Based on the path extension results and fracture propagation records, the connection paths in the closed evolution sequence are segmented, and the continuously extended path segments are distinguished from the interrupted path segments. The connection states of each path segment are integrated in chronological order to form a closed failure sequence, which is used to represent the failure changes of the boundary connection state. S203. For the path continuity changes in the closed failure sequence, combined with the regional existence state in the existence discrimination sequence, the length of the continuous path segment and the fracture propagation range are statistically calculated to obtain the closure attenuation parameter and the expansion gradient parameter, and the degree of boundary closure difference is characterized according to the combination relationship between the closure attenuation parameter and the expansion gradient parameter.

5. The intelligent situational awareness method based on multi-source data fusion according to claim 4, characterized in that, S203 specifically refers to: To address the changes in path continuity in a closed broken sequence, the length of continuous path segments within each time frame is statistically analyzed, and a path length change sequence is constructed in chronological order. Simultaneously, the expansion range of the path interruption location in adjacent time frames is extracted to form a fracture propagation range sequence. By combining the regional existence state in the existence discrimination sequence, joint constraint statistics are performed on the path length change sequence and the fracture propagation range sequence. Within the time interval where the regional existence state remains consistent, the change amplitude of the continuous path length is cumulatively calculated, and the gradient of the expansion amplitude of the fracture propagation range is statistically analyzed to generate closure attenuation parameters and expansion gradient parameters. Based on the combination relationship between the closure decay parameter and the extended gradient parameter, the closure decay parameter is classified into levels according to its numerical range, and the extended gradient parameter is segmented according to its changing trend. Within the same time interval, the classification results of the closure decay parameter and the segmentation marks of the extended gradient parameter are matched accordingly. A multi-level mapping table is constructed according to the matching combination, and different difference levels are assigned to different matching combinations according to the mapping table to form the boundary closure difference characterization results.

6. The intelligent situational awareness method based on multi-source data fusion according to claim 1, characterized in that, S3 specifically includes the following steps: S301. Couple the closed-loop decay parameter and the extended gradient parameter by arranging the closed-loop decay parameter and the extended gradient parameter in chronological order, normalizing the values ​​of the closed-loop decay parameter and the extended gradient parameter within the same time interval, and weighting the normalized closed-loop decay parameter and the extended gradient parameter according to the preset weight to form a sequence of coupled operation results. S302. Based on the coupling operation result sequence, an existence consistency strength is introduced. The existence consistency strength is generated by statistically calculating the proportion of consistent states in continuous time intervals in the existence discrimination sequence. The existence consistency strength is then applied to the coupling operation result sequence according to the time correspondence, and the coupling operation result sequence is offset and adjusted to generate a consistency offset parameter. S303. Construct a nonlinear mapping relationship for the uniform offset parameter. By segmenting the uniform offset parameter according to the numerical range, perform mapping calculations in each segment according to the nonlinear combination relationship, and integrate the mapping results in chronological order to form the target area integrity evaluation coefficient.

7. The intelligent situational awareness method based on multi-source data fusion according to claim 6, characterized in that, S302 specifically refers to: Based on the result sequence of the coupling operation, the existence discrimination sequence is divided into time intervals, the label sequence of consistent states in continuous time intervals is extracted, and the number of frames with consistent states in each time interval and the total number of frames are statistically calculated to form a consistent state proportion sequence to generate the existence consistency strength. The existence consistency intensity is aligned frame by frame with the coupling operation result sequence according to the time correspondence. The coupling operation result and the existence consistency intensity at the corresponding time point in each time frame are combined. By applying a proportional adjustment coefficient to the coupling operation result, an offset processing sequence is formed. The offset processing sequence is continuously integrated in chronological order. The offset-adjusted values ​​in each time frame are serialized and arranged to generate a consistent offset parameter. The consistent offset parameter is used to characterize the offset change of the coupling operation result under the influence of consistent strength.

8. The intelligent situational awareness method based on multi-source data fusion according to claim 1, characterized in that, S4 specifically refers to: Based on the integrity evaluation coefficient of the target area, a hierarchical mapping is performed. The integrity evaluation coefficient of the target area is extracted in chronological order and divided into segments according to a preset numerical interval. The integrity evaluation coefficient of the target area is then mapped to the corresponding interval position to form a hierarchical mapping result. Based on the results of the hierarchical mapping, the integrity evaluation coefficient of the target area is classified according to the numerical range of each interval. Intervals with high values ​​are marked as complete intervals, intervals with medium values ​​are marked as transitional intervals, and intervals with low values ​​are marked as damaged intervals, thus forming complete intervals, transitional intervals, and damaged intervals. After forming complete, transitional, and damaged intervals, the degree of boundary closure difference is matched frame by frame with the interval classification results according to the time correspondence. The degree of boundary closure difference within the same time interval is classified into levels, and the classified difference levels are combined with the corresponding interval categories. Based on the correspondence between the interval category and the difference level in the combined tags, the target area is classified and determined. When the interval category is a complete interval and the difference level is low, it is determined to be in a complete state. When the interval category is a transitional interval or the difference level is medium, it is determined to be in a transitional state. When the interval category is a damaged interval or the difference level is high, it is determined to be in a damaged state, thus completing the classification and determination of the integrity of the target area.

9. The intelligent situational awareness method based on multi-source data fusion according to claim 1, characterized in that, S5 specifically refers to: The temporal evolution sequence of boundary closure differences is tracked and processed. The degree of boundary closure difference in each time frame is extracted in chronological order, and the difference changes between adjacent time frames are continuously calculated to form a difference change sequence. The target area integrity evaluation coefficient is continuously updated based on the difference change sequence. The target area integrity evaluation coefficient in each time frame is numerically corrected by introducing the difference change amount at the corresponding time point to form a continuously updated target area integrity evaluation coefficient. Based on the continuous updating of the target area integrity evaluation coefficient, the hierarchical mapping interval is adjusted synchronously. By matching the difference change sequence with the hierarchical mapping interval according to the time correspondence, the numerical boundaries of each interval are dynamically adjusted. Based on the adjusted hierarchical mapping interval, the continuously updated target area integrity evaluation coefficient is remapped to complete the dynamic control of the target area integrity judgment.

Images