A remote sensing-driven land use monitoring system

Abstract

The present application relates to the technical field of remote sensing image processing, in particular to a remote sensing driven land use monitoring system. The present application extracts the ground object boundary contour in the remote sensing image and constructs a morphological parameter set with geometric structure proportion characteristics, establishes a path category data system based on the division of land spatial form based on structure characteristics, introduces normalized processing and Gaussian mixture model to realize the aspect ratio clustering of ground objects, combines the distribution concentration degree to construct a structure difference index set, clearly identifies the path type to track the land block direction or structure change track, calculates the disturbance factor based on the land class number change and the connection direction number of the patch adjacent area in the multi-temporal remote sensing image, and comprehensively discriminates the main direction, boundary contour similarity and spatial disturbance of the land block in the specified period according to the disturbance factor, thereby realizing the dynamic change identification of the land use state, and accurately reflecting the land use form change amplitude and time period characteristics through setting the discrimination benchmark and change state marking mechanism.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of remote sensing image processing, and in particular to a remote sensing driven land use monitoring system. BACKGROUND

[0002] The technical field of remote sensing image processing is a technical system for information extraction and understanding of images collected by remote sensing equipment, and its core matters include discrimination of ground targets, quantization of distribution characteristics, and expression and modeling of spatial attributes based on remote sensing imaging data.

[0003] Among them, the traditional remote sensing driven land use monitoring system refers to a system for analyzing and interpreting the land use form based on remote sensing image data, which is used to identify land use types from remote sensing images and monitor the spatial pattern changes.

[0004] The traditional remote sensing driven land use monitoring system mainly relies on the analysis and interpretation process of ground object types in remote sensing images, and does not refine the modeling of ground object boundary form structure in the identification process, which makes it difficult to accurately distinguish different ground object categories in areas where there are multiple similar forms but significant differences in use, and it is difficult to capture land state changes in time when the spatial pattern has microstructure disturbance due to the lack of dynamic change correlation judgment mechanism of adjacent land blocks. When the multi-period remote sensing image shows slight boundary changes or the evolution trend of land use is not obvious, the system may miss the change due to the lack of continuity parameter comparison mechanism, and it also lacks further classification ability for the identified changed ground objects, making it difficult to accurately reflect the specific type and land evolution trend of the changes. For example, forest and shrub are easily misjudged due to similar forms, or construction land is misidentified due to overlapping spectral characteristics with farmland in the early stage of transformation, resulting in certain limitations in the timeliness and classification accuracy of the land use monitoring results. SUMMARY

[0005] The purpose of the present application is to solve the problems existing in the prior art and to provide a remote sensing driven land use monitoring system.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme: a remote sensing driven land use monitoring system, the system comprises:

[0007] Patch form extraction module: extracting the land surface ground object patch boundary contour from the remote sensing image, constructing a patch form parameter set by proportionally calculating the geometric structure characteristics of the patch boundary contour;

[0008] Path type division module: dividing the land surface form structure according to the patch form parameter set, setting the path type identifier according to the form structure, and forming a path category data set;

[0009] An abutment change identification module: based on the path category dataset, judging the land class change of each patch abutment area through a remote sensing image, and calculating a land class disturbance factor corresponding to the patch;

[0010] A land change monitoring module: according to the path type specified for each patch in the path category dataset, comparing the morphological parameter change of each patch in a specified period with the land class disturbance factor of each patch, judging whether the land use state changes, and obtaining a land use change monitoring result.

[0011] The patch morphological parameter set includes an aspect ratio, a boundary direction, and a spatial distribution position, the path category dataset includes a path type, a morphological zoning category, and a patch path mapping relationship, the land class disturbance factor includes an abutment land class number change, an abutment direction number, and a change frequency statistical value, and the land use change monitoring result includes a change mark, a morphological change amplitude, and a judgment time period.

[0012] The patch morphological extraction module includes:

[0013] A boundary identification submodule: segmenting a land use patch in a remote sensing image, collecting a boundary line pixel position of each patch in the remote sensing image, calculating a coordinate point set of the boundary line in a remote sensing image grid, and generating a boundary line coordinate dataset;

[0014] A structure operation submodule: based on the boundary line coordinate dataset, extracting an outermost side length data of a boundary envelope frame of each patch, judging a direction attribute of both sides of the envelope frame and calculating a length ratio of the corresponding direction, and generating a boundary structure ratio sequence;

[0015] A parameter collection submodule: according to the boundary structure ratio sequence and the patch position index in the remote sensing image coordinate system, integrating the number, position coordinates and geometric ratio of each patch, and obtaining a patch morphological parameter set.

[0016] The path type division module includes:

[0017] A cluster identification submodule: calling the aspect ratio value of each patch in the patch morphological parameter set, using the aspect ratio value after normalization as an input variable, and using a Gaussian mixture model algorithm to cluster all patches to obtain a patch cluster distribution result;

[0018] A region classification submodule: based on the aspect ratio mean value and covariance of each cluster type in the patch cluster distribution result, calculating the numerical deviation degree of each type of patch on the aspect ratio feature, and comparing the distribution concentration degree of each type of patch on the remote sensing image, and generating a patch structure difference index set;

[0019] The path setting sub-module sets a structure recognition path and a direction adjustment path according to the combined characteristics of each type of patch in the aspect ratio and spatial concentration in the patch structure difference index set, pairs the path type with the patch number to establish a corresponding relationship, and obtains a path category data set.

[0020] The abutment change recognition module comprises:

[0021] The abutment extraction sub-module acquires remote sensing images at multiple time points, calls the spatial position of each patch in the path category data set, recognizes adjacent land object patches connected by the boundary, extracts the numbers of the adjacent patches, and obtains an abutment patch index set;

[0022] The land class change determination sub-module collects the land class numbers of adjacent patches at two remote sensing image time points based on the abutment patch index set, compares the number change, summarizes the number of patches with changes, and obtains an abutment land class change count result;

[0023] The disturbance quantity calculation sub-module calls the abutment land class change count result, combines the number of direction connections between the boundary of each patch and the adjacent patch at the current time point, calculates the correlation quantity of the number of adjacent patch land class changes and the number of connection directions, and obtains a land class disturbance factor corresponding to each patch.

[0024] The land change monitoring module comprises:

[0025] The path calling sub-module calls the path type corresponding to each patch in the path category data set, recognizes the direction adjustment path or the structure recognition path, acquires the main direction angle or the boundary contour similarity value of the patch in two remote sensing image periods according to the path type, and obtains a path type parameter set;

[0026] The parameter comparison sub-module sets a morphological parameter change judgment reference matched with the path type based on the path type parameter set and the land class disturbance factor of each patch, calculates the parameter change amplitude of each patch in two periods, and generates a patch change comparison quantity;

[0027] The change judgment sub-module compares the patch change comparison quantity with the set judgment reference, gives the patch a change state mark and a stable state mark according to the comparison result, and obtains a land use change monitoring result.

[0028] The change type discrimination module based on the land use change monitoring result refines the land use change type recognition, and obtains a change classification result.

[0029] The change classification result comprises a cultivated land change type, a forest land change type, a water body change type and a construction land change type.

[0030] The application improves that the change type judging module comprises:

[0031] Patch screening submodule: extract patch numbers marked as changes in the land use change monitoring result, and determine image positioning information in the current period remote sensing image to obtain a change patch index set;

[0032] Feature extraction submodule: based on the change patch index set, obtain principal component feature data and spectral reflection information of corresponding patches in the current period remote sensing image, and collect land class numbers corresponding to patches in the last period to form change patch feature combination data;

[0033] Type judging submodule: according to the class difference between current land classes and historical land classes, spectral value change direction and spatial structure change amplitude in the change patch feature combination data, divide the change patches into multiple categories to obtain a change classification result.

[0034] Compared with the prior art, the application has the advantages and positive effects that:

[0035] In the application, by extracting the feature boundary contour in the remote sensing image and constructing a morphological parameter set with geometric structure proportion characteristics, a path class data system for dividing land spatial morphology based on structure characteristics is established, normalized processing and Gaussian mixture model are introduced to realize object aspect ratio clustering, a structure difference index set is constructed in combination with distribution concentration, path types are clearly identified to track land block direction or structure change track, a disturbance factor is calculated based on land class number change and connection direction number in the patch adjacent area in multi-temporal remote sensing images, and the main direction of the land block in the specified period, boundary contour similarity and spatial disturbance are comprehensively judged based on the disturbance factor, so that dynamic change recognition of land use state is realized, the change amplitude and time period characteristics of land use morphology are accurately reflected by setting a judgment reference and a change state marking mechanism, and after change recognition, fine classification of land use change types is realized based on principal component feature, spectral reflection and historical land class number information, the fine-grained perception and classification judgment ability for changes of cultivated land, forest land, water body and construction land is effectively enhanced, and therefore the ability of the remote sensing monitoring system for structured modeling, continuous evolution perception and change type identification of land use evolution is significantly improved. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 The figure is a system module diagram of the application;

[0037] Figure 2 The figure is a system framework diagram of the application;

[0038] Figure 3 The figure is a schematic diagram of a patch morphology extraction module of the application;

[0039] Figure 4 A schematic diagram of the path type division module of the present application;

[0040] Figure 5 A schematic diagram of the adjacent change identification module of the present application;

[0041] Figure 6 A schematic diagram of the land change monitoring module of the present application;

[0042] Figure 7 A schematic diagram of the change type identification module of the present application. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical scheme and advantages of the present application more clear, the present application is further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0044] In the description of the present application, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, in the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise specifically limited.

[0045] Please refer to Figure 1 The present application provides a technical scheme: a remote sensing driven land use monitoring system, the system comprises:

[0046] Patch shape extraction module: extracting the land surface feature patch boundary contour from the remote sensing image, constructing the patch shape parameter set by proportionally calculating the geometric structure features of the patch boundary contour;

[0047] Path type division module: referring to the patch shape parameter set to divide the land surface shape structure, setting the path type identifier according to the shape structure, and forming the path category data set;

[0048] Adjacent change identification module: based on the path category data set, judging the land class change situation of each patch adjacent area through the remote sensing image, and calculating the land class disturbance factor of the corresponding patch;

[0049] Land change monitoring module: according to the path type specified for each patch in the path category data set, comparing the morphological parameter change situation of each patch in the specified period with the land class disturbance factor of each patch, judging whether the land use state has changed, and obtaining the land use change monitoring result;

[0050] The patch shape parameter set includes an aspect ratio, a boundary direction, and a spatial distribution position, the path category data set includes a path type, a shape zoning category, and a patch path mapping relationship, the land use disturbance factor includes a neighboring land class number change, a neighboring direction number, and a change frequency statistical value, and the land use change monitoring result includes a change mark, a shape change amplitude, and a judgment time period.

[0051] Referring to Figure 2 and Figure 3 , the patch shape extraction module includes:

[0052] a boundary recognition submodule: segmenting a land use patch in a remote sensing image, collecting a boundary line pixel position of each patch in the remote sensing image, calculating a coordinate point set of the boundary line in a remote sensing image grid, and generating a boundary line coordinate data set;

[0053] First, a remote sensing image is acquired, an image segmentation algorithm such as U-Net is used to classify and process the remote sensing image, each type of ground object such as water, building, farmland, and forest in the image is labeled as an independent patch, and then an edge detection algorithm such as a Canny operator or a Sobel operator is used to acquire boundary pixel points of each patch. For example, in a 512x512 pixel remote sensing image, after identifying an independent patch with an area greater than 100 pixels, a Canny edge detection is used to extract the contour points of one of the farmland patches, a total of 300 boundary pixel points are obtained, and the image coordinates (row and column numbers) of these points are converted into actual positions in a geographic coordinate system or a map projection coordinate system. For example, using the WGS84 coordinate system, the geographic coordinates of the boundary point (113.4567, 34.1234) can be obtained according to the affine transformation parameters of the remote sensing image. The complete boundary line coordinate sequence is obtained by repeating the process, and a boundary line coordinate data set is constructed.

[0054] a structure operation submodule: based on the boundary line coordinate data set, extracting the outermost edge length data of the envelope frame of each patch boundary, judging the direction attribute of the two sides of the envelope frame and calculating the length proportion of the corresponding direction, and generating a boundary structure ratio sequence;

[0055] The boundary structure characteristics of each patch are processed based on the boundary line coordinate data set. First, the minimum circumscribed rectangle method is applied to each set of boundary line coordinates to extract the envelope box side length information. This algorithm determines the length and width of the rectangular frame by calculating the maximum and minimum horizontal and vertical coordinates of the boundary points. For example, if the horizontal coordinate range of a patch's boundary line is 113.45-113.55, the latitude and longitude span is 0.10, and the horizontal length is 11.1 km, which is equivalent to 1 degree, and the vertical coordinate range is 34.10-34.16, the vertical length is 6.66 km. The direction attribute is determined by assigning a direction label (e.g., horizontal) based on the long side direction corresponding to the latitude or longitude direction. Then, the envelope box aspect ratio is calculated as 11.1:6.66=1.67, which is the boundary structure ratio. This value is recorded in the sequence for subsequent comparison and classification. If another patch calculates the envelope box ratio as 0.95, it belongs to the near-square category. By traversing all patches and performing this type of ratio extraction, multiple boundary structure ratio sequence sets are formed.

[0056] Parameter collection submodule: based on the boundary structure ratio sequence and the patch position index in the remote sensing image coordinate system, the number, position coordinates, and geometric ratio of each patch are integrated to obtain the patch morphology parameter set;

[0057] Based on the boundary structure ratio sequence and the patch position information in the remote sensing image, the morphology parameter set of each patch is organized. First, the number of each patch in the image is extracted, such as labeling the number 1-N according to the connected region in the segmented image. The boundary structure ratio corresponding to each number is extracted, and the geometric center coordinates of the patch are obtained from the boundary coordinate set. For example, a patch with number 7 has a boundary center coordinate of (113.4789, 34.1432) and a corresponding structure ratio of 1.25, forming a mapping relationship between patch number and position. Further, the patch number, center coordinates, and boundary structure ratio are integrated into a morphology parameter set, such as number 7-(113.4789, 34.1432)-1.25. This operation is performed on all patches to form a batch parameter set, which can be used for feature structure statistics and distribution analysis. Finally, the patch morphology parameter set is obtained.

[0058] Please refer to Figure 2 and Figure 4 , the path type division module includes:

[0059] Clustering identification submodule: call the aspect ratio of each patch in the patch morphology parameter set, and use the Gaussian mixture model algorithm to cluster all patches after normalizing the aspect ratio as input variables to obtain the patch clustering distribution result;

[0060] After the normalization of the plaque aspect ratio is completed, the normalized result is taken as an input to perform clustering analysis using a Gaussian mixture model (GMM). The clustering target is to divide all plaques into three categories according to the normalized aspect ratio. The core model of the GMM is a weighted superposition of multiple Gaussian distributions, and the total probability density function is

[0061]

[0062] According to the maximum probability principle, the value of the total probability density function under each Gaussian distribution category (such as low, medium, and high aspect ratio categories) is compared, and the plaque is classified into the category with the maximum probability value.

[0063] where x is the input variable, i.e., the normalized aspect ratio of the plaque, and the value range is [0, 1]; K is the number of clustering categories, which is 3 in this example, i.e., K=3; : the weight of the kth category, which satisfies , indicating the proportion of samples belonging to this category, and the weight is proportional to the number of sample points in this category, for example, the more sample points in the category, the greater the weight; : the mean of the kth category, indicating the center value of the normalized aspect ratio of this category; : the variance of the kth category, reflecting the dispersion degree of the data in the category; : a normal distribution function with as the mean and as the variance.

[0064] Suppose the normalized aspect ratio values of the six plaques are: 0.15, 0.18, 0.62, 0.68, 0.89, and 0.93.

[0065] In the initial step, the number of clusters K is set to 3, and the parameters of each category are initialized as follows:

[0066] Category 1 (low aspect ratio category): , ,

[0067] Category 2 (medium aspect ratio category): , ,

[0068] Category 3 (high aspect ratio category): , ,

[0069] For an input data x=0.68, the probability of belonging to each category is calculated as follows:

[0070] Calculate the probability term belonging to category 1: ​​​​​

[0071] ;

[0072] (due to the distance is too far, the exponential term tends to 0);

[0073] Calculate the probability term belonging to class 2:

[0074] ;

[0075] Calculate the probability term belonging to class 3:

[0076] .

[0077] The above results are normalized to obtain the posterior probability, and finally x = 0.68 is classified into class 2, indicating that it is a medium aspect ratio class patch.

[0078] The process is repeated for all samples, and the E step is used to update the probability value of each patch belonging to each class, and the M step updates 、 、 If class 2 contains 3 patches in the next round, its weight is updated to , and the mean and variance are recalculated according to the sample. After iteration convergence, each patch will be uniquely classified into a class, and the final clustering distribution label sequence will be formed.

[0079] Specifically, for each patch sample, input its normalized aspect ratio value x, and then use the probability density function of the Gaussian mixture model to calculate the probability value of each Gaussian distribution component, i.e. calculate the probability density of each class (e.g. class 1, class 2, class 3) and multiply the mixing weight of the class These values essentially represent the likelihood of the patch belonging to each class.

[0080] Then, the weighted probabilities of each patch in different categories are normalized so that the sum of the probabilities of all categories is 1, forming the belonging probability of each category. Specifically, for a patch, if its probability in class 1 is 0.1, in class 2 is 0.7, and in class 3 is 0.2, then after normalization, its probability of belonging to class 1 is 0.1 / (0.1+0.7+0.2)=0.1, class 2 is 0.7, and class 3 is 0.2.

[0081] Next, according to these belonging probabilities, the "maximum a posteriori probability principle" is used for category determination, i.e. the patch is classified into the category with the highest belonging probability. In the above example, the patch belongs to class 2 with the highest probability, so it is assigned to class 2.

[0082] The process is repeated for all patches, i.e. the normalized aspect ratio is inputted one by one, the weighted probability density under each Gaussian distribution is calculated, the attribution probability of each class is normalized, and then the class to which it belongs is determined according to the maximum probability. Finally, all patches are assigned to a specific class, forming a complete patch clustering distribution result.

[0083] The regional classification submodule: based on the mean and covariance of the aspect ratio of each cluster type in the patch clustering distribution result, the numerical deviation degree of each class of patch in the aspect ratio feature is calculated, and the distribution concentration degree of each class of patch in the remote sensing image is compared to generate a set of patch structure difference indicators;

[0084] After obtaining the clustering category of each patch, the mean and covariance of the aspect ratio of each category are calculated as the basis for feature analysis. There are three clustering categories, namely low aspect ratio category, medium aspect ratio category and high aspect ratio category. The aspect ratio of the patches contained in each category is counted, and the mean and covariance within the category are calculated. The mean of the low aspect ratio category is 0.92, the covariance is 0.0047, the mean of the medium aspect ratio category is 1.08, the covariance is 0.0012, and the mean of the high aspect ratio category is 1.26, the covariance is 0.0065. Then the difference between the aspect ratio within the category and the mean of the category is used as the evaluation basis for the numerical deviation degree, for example, if the aspect ratio of a patch is 1.00 and the category is medium aspect ratio category, the difference between 1.08 and 1.00 is 0.08, which is lower than 0.10 and is judged as "mild deviation", if the aspect ratio of a patch is 0.85 and belongs to the low aspect ratio category, the difference between 0.92 and 0.85 is 0.07, which is in "moderate deviation", if the aspect ratio of a patch is 1.42 and belongs to the high aspect ratio category, the difference between 1.26 and 1.42 is 0.16, which is more than 0.15 and can be judged as "high deviation". According to the set standard: the difference is less than 0.05 for "close to the center", 0.05-0.10 for "mild deviation", 0.10-0.15 for "moderate deviation", and more than 0.15 for "high deviation", combined with the spatial distribution concentration degree of each class of patch in the remote sensing image, for example, the average distance between the centers of gravity of each class of patch is calculated as the evaluation of the concentration degree, which is less than 300 meters and can be defined as "high concentration", 300-600 meters as "medium concentration", and more than 600 meters as "low concentration". The deviation degree level and the spatial concentration degree level are paired to form a set of structure difference indicators for each class of patch, for example, the high aspect ratio category is "high deviation + low concentration", the medium aspect ratio category is "mild deviation + high concentration", and the low aspect ratio category is "moderate deviation + medium concentration". The generation of the set of patch structure difference indicators is completed.

[0085] The path setting sub-module sets two types of path types, i.e., a structure recognition path and a direction adjustment path, according to the combined characteristics of each type of patch in the aspect ratio and spatial concentration in the patch structure difference index set, pairs the path types with the patch numbers to establish a corresponding relationship, and obtains a path category data set;

[0086] Based on the classification results in the structure difference index set, the aspect ratio deviation level and the spatial concentration level of each type of patch are analyzed, and two types of paths, i.e., a structure recognition path and a direction adjustment path, are divided. The structure recognition path mainly corresponds to the combination of "close to the center + high concentration" and "mild deviation + high concentration", which indicates that the patch has obvious structure characteristics and is concentrated in distribution, and is suitable for direct structure recognition operation. The direction adjustment path corresponds to the combination of "moderate deviation + moderate concentration" and "high deviation + low concentration", which indicates that the patch has obvious fluctuations in shape or is in a dispersed state in space, and needs to be adjusted in direction first and then recognized in structure. For a patch such as No. 18, which belongs to the medium aspect ratio category, the deviation level is "mild deviation" and the concentration is "high concentration", and the path type is set to the structure recognition path. For patch No. 27, which belongs to the high aspect ratio category, the deviation level is "high deviation" and the concentration is "low concentration", and the path type is set to the direction adjustment path. According to the above rules, the number of each patch is matched with the path type to establish a corresponding relationship, and finally a path category data set is obtained, which records the number of each patch and the matched path type label.

[0087] Please refer to Figure 2 and Figure 5 The abutment change recognition module includes:

[0088] The abutment extraction sub-module obtains remote sensing images at multiple time points, calls the spatial position of each patch in the path category data set, identifies adjacent patches connected by the boundary, extracts the numbers of the adjacent patches, and obtains an abutment patch index set;

[0089] First, prepare two time point remote sensing image data, such as 2023 and 2025, and call the spatial coordinate information of each patch in the path category data set, determine the actual position of the patch boundary on the image through coordinate mapping, and then use the boundary detection method to scan the pixel area around the patch boundary. Within a given tolerance range, identify whether there are other patch boundaries in contact with it, for example, when traversing the boundary of patch No. 21 in the image, it is found that the east side of the patch is in contact with patch No. 25 for 5 consecutive pixels, and the west side is in contact with patch No. 19 for 3 pixels. Then determine that patch No. 21, patch No. 25, and patch No. 19 are adjacent patches, and record the adjacency relationship. Repeat this process for all patches to establish the number pairing information of each patch and its adjacent patches, and finally form the adjacency patch index set, for example, patch No. 21 corresponds to adjacent patches 19 and 25, patch No. 19 corresponds to adjacent patches 21 and 22, and patch No. 25 corresponds to adjacent patches 21 and 27. The entire adjacency index set is organized and stored according to patch number.

[0090] Land use change determination submodule: based on the adjacency patch index set, collect the land use numbers of adjacent patches at two remote sensing image time points, compare the number changes, and summarize the number of patches that have changed to obtain the adjacent land use change count result;

[0091] After obtaining the adjacency patch index set, read the land use numbers of all patches from two remote sensing image time points, and compare the land use numbers of adjacent patches at two time points one by one. For example, patch No. 21 has the same land use number "farmland" in 2023 and 2025, so it is determined that there is no change. If its adjacent patch No. 19 has "forest" in 2023 and "building" in 2025, it is determined that the land use has changed, and the change is counted as 1 change event. Continue to check patch No. 25, which is "forest" to "forest", and do not count the change. Only patch No. 19 of patch No. 21 has changed, and the change number is 1. Compare the adjacent land use of all patches, and accumulate the number of land use changes in the adjacent area of each patch. If a certain number patch has 3 land use changes in its 4 adjacent patches, record its change count as 3. Summarize the adjacent land use change numbers of all patches to obtain the adjacent land use change count result.

[0092] Disturbance calculation submodule: call the adjacent land use change count result, combine the direction connection number between the boundary of each patch at the current time point and the adjacent patch, calculate the correlation between the number of adjacent patch land use changes and the number of connection directions, and obtain the land use disturbance factor corresponding to each patch;

[0093] After obtaining the adjacent class change count result, the class disturbance factor of each patch is calculated by combining the spatial connection relationship between each patch and its adjacent patch at the current time point. The factor reflects the strength of the influence of the adjacent class change on the spatial structure of the patch. The specific calculation method is as follows: the number of adjacent patches in which the class change occurs is compared with the actual number of connection directions, and the adjacent direction strength parameter is introduced to optimize the discrimination degree of the disturbance factor in the structure. The unified calculation formula of the disturbance factor is as follows:

[0094] ;

[0095] : the class disturbance factor of patch i (unitless); : the number of adjacent patches in which the class change occurs (unit: pieces); : the actual number of adjacent directions of patch i (unit: pieces), which is usually identified as up, down, left, right, and diagonal directions, and the maximum is 8; : the direction connection strength coefficient, which represents the connection strength of the adjacent patch in which the class change occurs in the direction, and the value range is 0.5-2.0, the longer the boundary contact length, the larger the value, for example, the main boundary contact is more than 10 pixels, which is set to 2.0, 5-10 pixels is 1.5, less than 5 pixels is 1.0, and point contact is 0.5.

[0096] Suppose patch No. 34 has 5 adjacent directions (i.e. ) in the current remote sensing image, and among the adjacent patches, patch No. 31 and patch No. 36 have class changes in the two periods of images, i.e. . The boundary contact pixels between patch No. 31 and patch No. 34 are 8, so the direction strength coefficient is 1.5, and the contact pixels of patch No. 36 are 3, so the coefficient is 1.0, then the average direction strength coefficient can be calculated as: .

[0097] Finally, the disturbance factor of patch No. 34 is obtained by substituting the disturbance factor formula: ; it is indicated that the patch is moderately affected by the adjacent class change at the current time point. The calculation process is performed on all patches to form a complete class disturbance factor sequence.

[0098] Please refer to Figure 2 and Figure 6 , the land change monitoring module comprises:

[0099] Path calling sub-module: call the path type corresponding to each patch in the path category data set, identify as direction adjustment path or structure identification path, obtain the main direction angle or boundary contour similarity value of the patch in the two remote sensing image periods according to the path type, and obtain the path type parameter set;​

[0100] According to the path type data set of the patch belongs to the path type of its morphological parameters classification extraction, if the path type is structure recognition path, the boundary contour information of the patch needs to be extracted in two remote sensing image periods, and the contour similarity value is calculated. The contour similarity is calculated by using the weighted normalized structure distance model, and the expression is:

[0101] ;

[0102] Wherein, S is the contour similarity of the target patch at two time points, the value range is between 0 and 1, and the closer to 1 indicates that the contour is closer; m is the number of comparable boundary points in two time points, d is the Euclidean distance between the corresponding boundary points in two time points, the unit is meter, and the normalized processing is , The average distance of all the maximum boundary points in the region is meter; 、 The envelope area of the target patch boundary contour at two time points is respectively, the unit is square meter, and the normalized expression is , The maximum envelope area in the study area; The boundary shape complexity weight is calculated according to the complexity index , p is the boundary length, and A is the patch envelope area: when the index is less than 20, it is set to 0.6; between 20 and 30, it is set to 0.75; more than 30, it is set to 0.9; The area change sensitivity weight is set according to the patch area: less than 2000 square meters, set to 0.4; between 2000 and 5000, set to 0.25; more than 5000, set to 0.1; the weight reflects the smaller the area, the more sensitive the change, and the two weights meet: .

[0103] Taking the patch No. 52 as an example, the comparable boundary points extracted in two image periods are 10, the Euclidean distance is 12, 8, 9, 7, 11, 10, 9, 8, 6, 10 meters, the average value of points is 9 meters, the maximum boundary distance normalization factor is 20 meters, the normalized average distance is 0.45, the boundary perimeter of the patch is 260 meters, the envelope area is 2300 square meters, the complexity index is , set , the area of two time points is 2300 and 2100 square meters respectively, the area difference is 200 square meters, the maximum area is 6400 square meters, the normalized area change is 0.03125, the patch area is in the interval of 2000 to 5000 square meters, set , after substituting into the formula, the calculation is:

[0104] ;

[0105] The contour similarity of the patch numbered 52 between the two time points is 0.65, indicating that the overall change of the contour is small but there is a certain deformation trend, and this value will be written into the path type parameter set. The basis for similarity judgment is the degree of geometric change of the contour between the two time points, including the spatial position difference of the boundary shape and the change amplitude of the envelope area. When the contour similarity is close to 1, the boundary shapes of the two time points are highly consistent, the spatial positions are basically coincided, and the envelope area changes slightly; when the similarity is much less than 1, the boundary points are obviously shifted, the contour shape is significantly changed, or the envelope area difference is large. In actual application, the similarity value is compared with the set threshold value, for example, greater than 0.85 can be judged as “highly similar”, 0.65 to 0.85 as “moderately similar”, and less than 0.65 as “low similarity”, which is used to distinguish the stability and change trend of the patch shape over time.

[0106] The parameter comparison submodule: based on the path type parameter set, combined with the land class disturbance factor of each patch, the morphological parameter change judgment reference value matching the path type is set, the parameter change amplitude of each patch in two periods is calculated, and the patch change comparison quantity is generated;

[0107] After obtaining the path type parameter set and the land class disturbance factor of each patch, different morphological parameter change judgment reference values are set according to the path type. If the path type is a structure recognition path, the contour similarity value is used for evaluation, and the judgment reference value is set in combination with the disturbance factor, for example, when the disturbance factor is less than 0.3, the similarity reference is set to 0.70, when the disturbance factor is 0.3 to 0.6, it is set to 0.75, and when it is higher than 0.6, it is set to 0.80, in order to enhance the change sensitivity; if the path type is a direction adjustment path, the main direction angle value is extracted for comparison, and the judgment reference is segmented as 10 degrees, 8 degrees and 5 degrees, the larger the disturbance, the smaller the tolerance, and the parameter change amplitude is calculated by taking the absolute value of the difference value of the data of the two time points, for example, the path type of the patch numbered 36 is a structure recognition path, the contour similarity is 0.62, the disturbance factor is 0.45, and the judgment reference is 0.75, then the change comparison quantity is 0.13, which is the difference value of 0.75 and 0.62. The path type of the patch numbered 41 is a direction adjustment path, the main direction angle at the two time points is 85 degrees and 96 degrees respectively, the disturbance factor is 0.28, and the tolerance angle is 10 degrees, then the angle change is 11 degrees, which exceeds the reference value. Finally, the change amplitude value of each patch is recorded to form the patch change comparison quantity sequence.

[0108] The change judging submodule compares the patch change contrast quantity with the set judging reference, and according to the comparison result, assigns the patch with a change state mark and a stable state mark to obtain a land use change monitoring result.

[0109] The patch change contrast quantity is compared with the corresponding judging reference value, if the change amplitude does not exceed the reference value, the patch is assigned with a "stable" state mark, if the change amplitude is equal to or exceeds the reference value, a "change" state mark is assigned, for example, the contour similarity difference of No. 36 is 0.13, the reference value is 0.75, the difference value causes the actual similarity to be lower than the reference, so it is marked as "change"; the directional angle difference of No. 41 is 11 degrees, which is higher than the judging reference 10 degrees, and is also marked as "change"; the directional angle difference of No. 55 is 6 degrees, which is lower than the corresponding reference 8 degrees, so it is assigned with a "stable" state, after the process is completed for all patches one by one, a patch change state data set is formed, including the number, path type, disturbance factor, parameter change quantity and state mark of each patch, which is the output result of the land use change monitoring system.

[0110] Please refer to Figure 2 and Figure 7 The change type discrimination module based on the land use change monitoring result refines the land use change type identification to obtain a change classification result.

[0111] The change classification result includes cultivated land change type, forest land change type, water body change type and construction land change type.

[0112] The change type discrimination module includes:

[0113] The patch screening submodule extracts the patch number with a change state mark in the land use change monitoring result, and determines the image positioning information in the current period remote sensing image to obtain a change patch index set.

[0114] The land use change monitoring result is filtered to extract all patch numbers with a "change" state mark, and further search the spatial positioning information of these patches in the current period remote sensing image, the positioning information includes image row and column number, image coordinates, map projection coordinates and the like, which are used to clearly define the accurate range of these patches in the image, for example, the boundary starting point of the patch No. 58 in the image is the row and column coordinates (120, 210), and the end point is (145, 235), the range can be used for subsequent feature extraction of image data, the positioning process is repeated for all patches judged as "change", and finally a change patch index set is generated, which is used to identify all regions that need to be further identified and analyzed.

[0115] Feature extraction submodule: Based on the change patch index set, it obtains the principal component feature data and spectral reflectance information of the corresponding patch in the current period remote sensing image, and collects the land use code corresponding to the patch in the previous period to form change patch feature combination data;

[0116] Using the set of changed patch indexes as input, image feature data corresponding to each changed patch in the current period's remote sensing image is extracted. This mainly includes the first few principal component values ​​obtained from principal component transformation and the original spectral reflectance information. For example, for patch number 58, the first principal component value extracted for the current period is 26.5, and the reflectance is 0.34 in the green band and 0.52 in the near-infrared band. At the same time, the land use code for this patch in the previous period is retrieved as "forest". The image features of the current period are combined with the land use information of the previous period to form a complete set of changed patch feature combination data. This process is performed on all changed patches to obtain a multidimensional data set containing image features and historical land use labels.

[0117] Type determination submodule: Based on the differences between the current land use type and the historical land use type, the direction of spectral value change, and the magnitude of spatial structure change in the feature combination data of the changed patches, the changed patches are divided into multiple categories to obtain the change classification results;

[0118] By combining feature data of changed patches, the differences in land use characteristics between the current period and the historical land use of the previous period, the direction of change in spectral reflectance, and the magnitude of change in spatial structure of each patch are analyzed. A comprehensive judgment is made to classify the change category. If a patch was "forest land" in the previous period, and its principal component has significantly decreased, near-infrared reflectance has decreased, boundary contour has simplified, and spatial structure similarity is less than 0.55 in the current period, then the patch is classified as "forest land converted to construction land." If a patch was historically "cultivated land," and its principal component has increased, red-edge reflectance has strengthened, and boundary structure remains stable, then it is classified as "cultivated land restored to cultivation." If the current principal component of a patch does not show significant change, but the reflectance in the red band of the spectrum decreases while the blue band increases, combined with a large perturbation factor and a moderate contour change value, then it is classified as a "potentially degraded area." After completing this type of judgment for all changed patches, a change classification result set is formed. This result set can support the statistical and dynamic analysis of regional land use transformation trends.

[0119] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A remote sensing driven land use monitoring system, characterized by: The system comprises a patch morphology extraction module, a path type division module, an adjacency change identification module, and a land change monitoring module. The patch morphology extraction module extracts patch boundary contours of land surface features from remote sensing images, and constructs a patch morphology parameter set by proportionally calculating geometric structural features of the patch boundary contours. The path type division module comprises: A cluster identification submodule calls the aspect ratio value of each patch in the patch morphology parameter set, normalizes the aspect ratio value as an input variable, and uses a Gaussian mixture model algorithm to cluster all patches to obtain patch cluster distribution results. A region classification submodule calculates the numerical deviation of each type of patch in the aspect ratio feature based on the mean and covariance of the aspect ratio of each cluster type in the patch cluster distribution results, and compares the distribution concentration of each type of patch in the remote sensing image to generate a patch structure difference index set. A path setting submodule sets two types of path types, structure recognition paths and direction adjustment paths, according to the combined features of the numerical deviation of each type of patch in the aspect ratio and the spatial distribution concentration in the patch structure difference index set, and then pairs the path types with patch numbers to establish a corresponding relationship to obtain a path category data set; wherein the patches belonging to the path type of the structure recognition path represent patches with obvious structural features and concentrated distribution, and the patches belonging to the path type of the direction adjustment path represent patches with obvious fluctuations in morphology or in a dispersed state in space. The adjacency change identification module judges the land class change of each patch based on the path category data set, and calculates the land class disturbance factor of the corresponding patch; the land class disturbance factor includes adjacent land class number change, adjacent direction number, and change frequency statistics. The land change monitoring module comprises: A path calling submodule calls the path type corresponding to each patch in the path category data set, obtains the main direction angle of the patch in two remote sensing image periods when the path type is identified as a direction adjustment path, and obtains the boundary contour similarity value of the patch in two remote sensing image periods when the path type is identified as a structure recognition path, to further obtain a path type parameter set. A parameter comparison submodule sets a morphology parameter change judgment benchmark matched with the path type based on the path type parameter set and the land class disturbance factor of each patch, calculates the morphology parameter change amplitude of each patch in two remote sensing image periods, and generates a patch change comparison quantity. A change judgment submodule compares the patch change comparison quantity with the set judgment benchmark, assigns a change state mark and a stable state mark to the patch according to the comparison result, and obtains a land use change monitoring result.

2. The remote sensing driven land use monitoring system according to claim 1, characterized in that: The patch morphology parameter set comprises an aspect ratio, a boundary direction, and a spatial distribution position; the path category data set comprises a path type, a morphology zoning category, and a patch path mapping relationship; and the land use change monitoring result comprises a change mark, a morphology change amplitude, and a judgment time period.

3. The remote sensing driven land use monitoring system of claim 1, wherein: The patch morphology extraction module comprises: Boundary recognition submodule: Segment land use feature patches in remote sensing images, collect the boundary line pixel position of each patch in the remote sensing image, calculate the set of coordinate points of the boundary line in the remote sensing image raster, and generate a boundary line coordinate dataset; The structure operation submodule extracts the outermost side length data of the boundary envelope of each patch based on the boundary line coordinate dataset, determines the directional attributes of the two sides of the envelope and calculates the length ratio of the corresponding directions, and generates a boundary structure ratio sequence. The parameter collection submodule integrates the number, location coordinates, and geometric ratio of each patch based on the boundary structure ratio sequence and the patch location index in the remote sensing image coordinate system to obtain a patch morphology parameter set.

4. The remote sensing driven land use monitoring system of claim 1, wherein: Cluster all patches using the following formula: ； Computing the total probability density function of the Gaussian distributions According to the maximum probability principle, the total probability density function is compared in each Gaussian distribution category, and the plaque is classified into the category with the maximum probability value. where K is the number of cluster classes set, is the weight of the kth class, representing the proportion of samples belonging to this class, and the weight is set in proportion to the number of sample points in this class, is the mean of the kth class, also representing the center value of the normalized aspect ratio of the kth class, is the variance of the kth class, reflecting the dispersion of data in the kth class, is a normal distribution function with as the mean and as the variance, and x is the input variable, i.e., the normalized aspect ratio of the patch, with a value range of [0, 1].

5. The remote sensing driven land use monitoring system of claim 1, wherein: The adjacency change identification module includes: The adjacency extraction submodule acquires remote sensing images from multiple time points, calls the spatial location of each patch in the path category dataset, identifies adjacent land cover patches connected by the boundary, and extracts the numbers of adjacent patches to obtain an adjacency patch index set. Land use change determination submodule: Based on the adjacent patch index set, collect the land use numbers of adjacent patches at two remote sensing image time points, compare the changes in the numbers, summarize the number of patches with changes, and obtain the adjacent land use change count result; Disturbance Calculation Submodule: Calls the neighboring land type change count results, combines them with the number of connection directions between each patch boundary and neighboring patches at the current time point, calculates the correlation between the number of land type changes in neighboring patches and the number of connection directions, and obtains the land type disturbance factor corresponding to each patch.

6. The remote-sensing driven land use monitoring system of claim 1, wherein: The formula is used to determine the similarity of the boundary contours of patches across two remote sensing image cycles: ； Where S represents the contour similarity of the target patch at two time points, with a value between 0 and 1. The closer to 1, the closer the contours are. m represents the number of comparable boundary points at the two time points, and d represents the Euclidean distance between the corresponding boundary points at the two time points. , These represent the envelope area of ​​the target patch boundary contour at two different time points. To determine the maximum envelope area within the study region, This represents the average distance between the maximum boundary points of all patches within the region. Weights for boundary shape complexity. As the area change sensitivity weight, .

7. The remote sensing driven land use monitoring system of claim 1, wherein: It also includes a change type identification module: based on the land use change monitoring results, it refines the identification of land use change types and obtains change classification results; The change classification results include cultivated land change types, forest land change types, water body change types, and construction land change types.

8. The remote-sensing driven land use monitoring system of claim 7, wherein: The change type discrimination module includes: Patch filtering submodule: Extracts the patch numbers marked as changed from the land use change monitoring results, and determines the image location information in the current period's remote sensing image to obtain a set of changed patch indexes; Feature extraction submodule: Based on the changed patch index set, obtain the principal component feature data and spectral reflectance information of the corresponding patch in the current period remote sensing image, and collect the land use number corresponding to the patch in the previous period to form changed patch feature combination data; Type determination submodule: Based on the differences between the current land type and the historical land type, the direction of spectral value change, and the magnitude of spatial structure change in the combined feature data of the changed patches, the changed patches are divided into multiple categories to obtain the change classification results.

Images