Breeding pond dam remote sensing extraction method based on decision rule

Abstract

The invention discloses a culture pond dam remote sensing extraction method based on a decision rule, and the method comprises the steps: obtaining remote sensing image data of a target region, extracting classification features to construct a classification feature data set, and constructing a local culture pond dam data set; constructing a potential decision rule according to the classification feature data set, iteratively optimizing each classification feature threshold in the potential decision rule through a genetic algorithm, evaluating the accuracy, terminating iteration until the accuracy is no longer improved, obtaining the optimized decision rule and the corresponding accuracy, and further obtaining an optimal decision rule and a core decision rule; analyzing response conditions of the optimal decision rule and the core decision rule to different land cover types, and determining a decision rule application range; and according to the core decision rule, extracting a breeding pond dam remote sensing image of the target area. According to the invention, the remote sensing extraction efficiency and accuracy of the breeding pond dam can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of remote sensing image recognition, and relates to but is not limited to a remote sensing extraction method for aquaculture pond embankment based on decision rules. BACKGROUND

[0002] An aquaculture pond is composed of an embankment and a water body surrounded by the embankment and other land cover types. As an important part of artificial construction, the embankment has high spatial stability and is not easily affected by factors such as tides, water level changes or aquaculture stage adjustments. Compared with the embankment, the area and morphology of the water body are easily affected by multiple factors such as the type of aquaculture objects, management methods and climate change, and can easily fluctuate greatly. Therefore, the embankment, as a target with spatial stability and structural recognizability, is more suitable as a key feature for identifying the boundary of the aquaculture pond in remote sensing images. The embankment of the aquaculture pond bears the functions of delimiting the regional range, regulating water exchange and resisting external interference, and through its spatial layout and structural characteristics, it can reflect the zoning of land use and the ownership of property. The construction materials of the aquaculture pond embankment, such as stones and earthworks, often have strong reflective properties, especially in medium spatial resolution remote sensing images, which form a sharp contrast with the low reflectivity of the water body. This contrast effect makes the high reflectivity of the embankment still significantly raise the brightness value of the remote sensing pixel even if the embankment is narrow and only occupies a small part of the remote sensing pixel, thereby becoming a recognizable remote sensing feature, providing a stable foundation for remote sensing extraction of the aquaculture pond embankment.

[0003] In the prior art, embankment recognition techniques are mostly based on machine learning and deep learning algorithms. Such methods rely on a large number of training samples and use complex models for feature learning and pattern recognition. Although they can achieve high recognition accuracy in certain areas, the internal mechanisms of the models are highly complex and lack interpretability, making it difficult to intuitively understand the classification basis. Moreover, when faced with different regions or different remote sensing data sources, these methods often require retraining of the models.

[0004] Therefore, there is an urgent need for a more accurate remote sensing extraction method for aquaculture pond embankments to address the problems of strong dependence on training samples, weak classification ability and high recognition complexity in existing extraction methods, and to significantly improve the classification ability, interpretability and rule transfer ability of embankment remote sensing extraction. Ultimately, it aims to accurately monitor the spatial pattern of aquaculture ponds and provide reliable support for aquaculture resource management and ecological assessment. SUMMARY

[0005] The embodiments of the present application provide a remote sensing extraction method for aquaculture pond embankment based on decision rules.

[0006] The technical solution of the embodiments of the present application is implemented as follows: In a first aspect, the embodiments of the present application provide a breeding pond dam remote sensing extraction method based on decision rules, which comprises: acquiring remote sensing image data of a target area, and extracting classification features in the remote sensing image data, constructing a classification feature dataset, and constructing a local breeding pond dam dataset; constructing potential decision rules according to the classification feature dataset, iteratively optimizing each classification feature threshold in the potential decision rules based on the local breeding pond dam dataset through a genetic algorithm, and evaluating the accuracy until the iteration is terminated when the accuracy no longer improves, obtaining an optimized decision rule and a corresponding accuracy; obtaining an optimal decision rule and a core decision rule according to the optimized decision rule; analyzing the response of the optimal decision rule and the core decision rule to different land cover types to determine the application scope of the decision rule; and extracting the breeding pond dam remote sensing image of the target area according to the core decision rule.

[0007] The technical scheme provided in the application obtains remote sensing image data of a target region, screens remote sensing images with clear imaging, no cloud layer or snow and ice interference, ensures the reliability of the data, guarantees the data quality, reduces the influence of noise on feature extraction, extracts classification features in the remote sensing image data, fuses various classification features to construct a classification feature data set, comprehensively captures the high reflection characteristics and morphological characteristics of the dam, enhances the feature distinguishing ability, constructs a local aquaculture pond dam data set, and facilitates subsequent rule construction; potential decision rules are constructed according to the classification feature data set, each classification feature threshold in the potential decision rules is iteratively optimized based on the local aquaculture pond dam data set through a genetic algorithm, and the accuracy is evaluated until the iteration is terminated when the accuracy no longer improves, so that the rule accuracy is maximized, and finally the optimized decision rules and the corresponding accuracy are obtained, the decision rules are automatically generated throughout the process, and the dependence on artificial experience is avoided; the optimal decision rules and the core decision rules are obtained according to the optimized decision rules, the core part in the optimal decision rules is determined by analyzing the pixel response state of each component in the high-accuracy rules, the internal logic is mined, and the interpretability of the rules is strengthened, the rule simplification is completed by eliminating the redundant components in the rules and retaining only the core decision rules, the accuracy is maintained while the operation efficiency is improved, and in addition, the core decision rules focus on the universal characteristics and are not dependent on specific numerical combinations, which facilitates cross-regional adaptation and is the basis for the transferability of the rules; the response of the optimal decision rules and the core decision rules to different land cover types is analyzed, the application range of the decision rules is determined, the controllability of the misclassification is ensured, the generalization ability of the core decision rules is improved, the overfitting risk is reduced, and the rule reliability is improved; according to the core decision rules, the aquaculture pond dam remote sensing image of the target region is extracted, the extraction process of the remote sensing image is simplified while the high accuracy of extraction is maintained, the efficient and accurate extraction of the aquaculture pond dam remote sensing image is realized, the remote sensing image extraction depends on lightweight rules rather than a calculation model, large-area images can be quickly processed, the large-scale applicability of the core decision rules is improved, and in addition, the rules have strong transferability and can be adapted to different regional requirements by adjusting the threshold value only, so that the long-term dynamic monitoring demand can be met.

[0008] Optionally, the remote sensing image data includes clear imaging synthetic aperture radar data and multispectral satellite data, and the extraction of the classification features in the remote sensing image data and the construction of the classification feature data set include: extracting multispectral band features, band ratios and common remote sensing indexes according to the multispectral satellite data in the remote sensing image data; extracting a backscattering coefficient according to the synthetic aperture radar data in the remote sensing image data; and constructing the classification feature data set according to the multispectral band features, the band ratios, the common remote sensing indexes and the backscattering coefficient.

[0009] Optionally, the local aquaculture pond dam data set comprises an aquaculture pond dam distribution grid file of a local area, 1 in the aquaculture pond dam distribution grid file represents an aquaculture pond dam area, and 0 in the aquaculture pond dam distribution grid file represents a non-aquaculture pond dam area.

[0010] Optionally, the constructing the potential decision rule according to the classification feature data set comprises: generating a potential decision rule set by randomly combining classification features and randomly setting a logical relationship between the classification features and a threshold value according to the classification feature data set, the logical relationship comprising the classification feature being greater than the threshold value and the classification feature being less than the threshold value; screening rules with a component number of 3 to 7 in the potential decision rule set and deleting repeated rules to obtain the potential decision rule.

[0011] Optionally, the obtaining the optimal decision rule and the core decision rule according to the optimized decision rule comprises: arranging the optimized decision rule in a descending order of correctness to obtain the optimal decision rule; drawing a pixel response condition of each component of the optimal decision rule, analyzing the role of each component, and obtaining a core part of the optimal decision rule, the core part of the optimal decision rule being the core decision rule.

[0012] Optionally, the analyzing the response conditions of the optimal decision rule and the core decision rule to different land cover types and determining a decision rule application range comprises: analyzing pixel response conditions of the optimal decision rule and the core decision rule to nine typical land covers of aquaculture pond dams, tidal flats, salt marsh vegetation, mangroves, impervious surfaces, dry lands, paddy fields, forest lands, and grasslands, explicitly classifying errors of the decision rule on different land cover types, and determining the decision rule application range.

[0013] Optionally, the extracting the aquaculture pond dam remote sensing image of the target area according to the core decision rule comprises: applying the core decision rule to classify the remote sensing image of the target area into 0 and 1, extracting a 1 part in the remote sensing image, and obtaining the aquaculture pond dam remote sensing image.

[0014] In a second aspect, an embodiment of the present application provides an electronic device, comprising a memory and a processor, the memory storing a computer program capable of running on the processor, and the processor implements the steps in the above-mentioned aquaculture pond dam remote sensing extraction method based on a decision rule when executing the program.

[0015] In a third aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the computer program implements the steps in the above-mentioned aquaculture pond dam remote sensing extraction method based on a decision rule when executed by a processor.

[0016] The technical scheme provided by the embodiments of the present application brings at least the following beneficial effects: The application provides a breeding pond dam remote sensing extraction method based on decision rules, obtains remote sensing image data of a target area, screens remote sensing images with clear imaging, no cloud layer or ice and snow interference, ensures data reliability, guarantees data quality, reduces the influence of noise on feature extraction, and extracts classification features in the remote sensing image data, fuses various classification features to construct a classification feature data set, comprehensively captures the high reflection characteristics and morphological features of the dam, enhances the feature differentiation ability, constructs a local breeding pond dam data set, and facilitates subsequent rule construction; potential decision rules are constructed according to the classification feature data set, based on the local breeding pond dam data set, each classification feature threshold in the potential decision rules is iteratively optimized through a genetic algorithm, and the accuracy is evaluated until the iteration is terminated when the accuracy no longer improves, ensuring that the rule accuracy is maximized, and finally obtaining the optimized decision rules and the corresponding accuracy, the decision rules are automatically generated throughout the process, avoiding dependence on artificial experience; the optimal decision rule and the core decision rule are obtained according to the optimized decision rule, the pixel response state of each component in the high-accuracy rule is analyzed to determine the core part of the optimal decision rule, the internal logic is excavated, and the rule interpretability is strengthened, the redundant components in the rule are removed, only the core decision rule is retained, the rule simplification is completed, the accuracy is maintained while the operation efficiency is improved, and moreover, the core decision rule focuses on the universal characteristics and does not depend on specific numerical combinations, facilitating cross-regional adaptation and serving as the basis for rule transferability; the response of the optimal decision rule and the core decision rule to different land cover types is analyzed, the application range of the decision rule is determined, the controllability of misclassification is ensured, the generalization ability of the core decision rule is improved, the overfitting risk is reduced, and the rule reliability is improved; according to the core decision rule, the remote sensing image of the breeding pond dam in the target area is extracted, the extraction process of the remote sensing image is simplified and the high precision of the extraction is maintained, the efficient and accurate extraction of the remote sensing image of the breeding pond dam is realized, the remote sensing image extraction depends on lightweight rules rather than calculation models, large-area images can be quickly processed, the large-scale applicability of the core decision rule is improved, and moreover, the rule transferability is strong, and only the threshold needs to be adjusted to adapt to different regional requirements, which can meet the long-term dynamic monitoring demand.The technical scheme provided in the application can finally realize the transparency and interpretability of the aquaculture pond dam identification process, realize efficient and accurate extraction of the aquaculture pond dam, overcome the problems of complex model structure, unclear classification basis and related uninterpretability of traditional deep learning and machine learning methods, and meanwhile, the technical scheme provided in the application fully considers the response characteristics of different land covers in the rule mining process, so that the core decision rule has strong generalization ability and good migration ability, can be flexibly applied to the aquaculture pond dam identification requirements in different regions, different scales and different remote sensing image data sources, has lower dependence on the number of samples compared with the traditional method which depends on a large number of labeled samples, has higher extraction efficiency, can more stably identify the dam structure with a relatively narrow width but obvious reflection characteristics, is particularly suitable for dam change monitoring in complex aquaculture regions with large scales and multiple time phases, in addition, the advantages of the stable form and high reflection characteristics of the aquaculture pond dam in remote sensing images can improve the robustness and applicability of dam identification, have good practical value and promotion prospect, and provide reliable data support and technical support for regional aquaculture resource management and ecological environment evaluation. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor. Figure 1 A flowchart of a decision rule-based aquaculture pond dam remote sensing extraction method provided by the embodiments of the present application; Figure 2 A schematic diagram of the pixel response state of the optimal decision rule and each component of the aquaculture pond dam remote sensing identification provided by the embodiments of the present application; Figure 3 A schematic diagram of the pixel response state of the optimal decision rule and the core decision rule of the aquaculture pond dam remote sensing identification to different land cover types provided by the embodiments of the present application; Figure 4 A schematic diagram of the aquaculture pond dam remote sensing image of the target region based on the core decision rule provided by the embodiments of the present application; Figure 5 A hardware entity schematic diagram of an electronic device provided by the embodiments of the present application. DETAILED DESCRIPTION

[0018] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the following will be combined with the drawings in the embodiments of the present application to make a clear and complete description of the technical solutions in the embodiments of the present application. Obviously, the described embodiments are some of the embodiments of the present application but not all the embodiments of the present application. The following embodiments are used to illustrate the present application but not to limit the scope of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of the present application.

[0019] In the following description, “some embodiments” are related to a subset of all possible embodiments, but it can be understood that “some embodiments” can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0020] It should be noted that the terms “first\second\third” involved in the embodiments of the present application are only to distinguish similar objects and do not represent the specific order of the objects. It can be understood that “first\second\third” can be interchanged with specific order or sequence as allowed, so that the embodiments of the present application described here can be implemented in an order other than that illustrated or described here.

[0021] Those skilled in the art can understand that, unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as generally understood by those of ordinary skill in the art to which the embodiments of the present application belong. It should also be understood that terms such as those defined in general dictionaries should be understood as having meanings consistent with those in the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as such.

[0022] The embodiments of the present application will be further described below in combination with the drawings.

[0023] In view of the problems existing in the research on remote sensing extraction of aquaculture pond embankment in the field of remote sensing image recognition technology, the embodiments of the present application provide a remote sensing extraction method of aquaculture pond embankment based on decision rules.

[0024] The technical solutions of the present application will be introduced below, and first the method embodiments of the present application will be introduced.

[0025] Please refer to Figure 1 which shows a flowchart of a remote sensing extraction method of aquaculture pond embankment based on decision rules provided by the embodiments of the present application, as shown in Figure 1 The method at least includes the following steps S110 to S150.

[0026] Step S110, obtain remote sensing image data of the target region, and extract classification features in the remote sensing image data to construct a classification feature data set and a local aquaculture pond dam data set.

[0027] In the embodiment of the present application, remote sensing image data of the target region is obtained, which is obtained through Sentinel-1 synthetic aperture radar data and Sentinel-2 multispectral satellite data. Specifically, satellite image data imaged during the period from April 2023 to November 2023 is selected, and the satellite image data in this period is arranged in descending order according to the cloud cover, so as to obtain remote sensing image data with the best imaging quality in the target region. The remote sensing image data includes Sentinel-1 synthetic aperture radar data and Sentinel-2 multispectral satellite data, and the remote sensing image is clear and is not affected by cloud and shadow, ice and snow cover, etc.

[0028] In the embodiment of the present application, the classification features in the remote sensing image data are extracted to construct a classification feature data set. Specifically, the classification features in the remote sensing image data include multispectral band features, band ratios, common remote sensing indices, and backscatter coefficients. The multispectral band features, band ratios, and common remote sensing indices are extracted from the Sentinel-2 multispectral satellite data in the remote sensing image data, and the backscatter coefficients are extracted from the Sentinel-1 synthetic aperture radar data in the remote sensing image data. Finally, the classification feature data set is constructed according to the multispectral band features, band ratios, common remote sensing indices, and backscatter coefficients.

[0029] In the embodiment of the present application, a local aquaculture pond dam data set is constructed. The selected local region meets the condition that the land cover type is diverse and the coverage area is not less than 10 km 2 , which can ensure the reliability of the decision rule. The local aquaculture pond dam data set includes a local region aquaculture pond dam distribution raster file, wherein the value 1 in the aquaculture pond dam distribution raster file represents an aquaculture pond dam region, and the value 0 in the aquaculture pond dam distribution raster file represents a non-aquaculture pond dam region. It should be particularly noted that the above local region coverage area is an example number obtained through better experimental effect in the embodiment, and can be adjusted according to actual conditions during use. The present application does not limit this.

[0030] Step S120, according to the classification feature data set, a potential decision rule is constructed, based on the local aquaculture pond dam data set, each classification feature threshold in the potential decision rule is iteratively optimized through a genetic algorithm, and the accuracy is evaluated until the accuracy no longer improves, the iteration is terminated, and the optimized decision rule and the corresponding accuracy are obtained.

[0031] In the embodiment of the present application, the potential decision rules are constructed according to the classification feature data set. Specifically, according to the classification feature data set, the classification features are randomly combined by a genetic algorithm, and the logical relationship between the classification features and the threshold values is randomly set, the logical relationship being a size relationship between the classification features and the threshold values, including two relationships of the classification features being greater than the threshold values and the classification features being less than the threshold values, and finally a series of decision rules are generated to constitute the potential decision rule set. Further, the potential decision rule set is traversed, and the components of each rule are counted, the rules with the component number being 3 to 7 in the potential decision rule set are screened, and the repeated rules are deleted, so that the potential decision rules are obtained.

[0032] In the embodiment of the present application, the threshold value of each classification feature in the potential decision rule is determined by the representation of the aquaculture pond dam distribution grid file in the local area. Specifically, a new genetic algorithm, which can be a gene algorithm, is applied, and based on the potential decision rule and the aquaculture pond dam distribution grid file in the local area, the threshold value of each classification feature in the potential decision rule is adjusted by continuously iterating and optimizing the gene algorithm. The accuracy is calculated and evaluated by comparing the rule extraction result with the aquaculture pond dam distribution grid file in the local area, and the iteration is terminated when the accuracy no longer improves, so that the optimized decision rule and the corresponding accuracy are obtained. At this time, the optimized decision rule and the corresponding accuracy are recorded for subsequent analysis. In a specific embodiment, the number of decision rules is 2000. It should be particularly noted that the number of decision rules above is an example number obtained by better experimental effect in the embodiment, and the number can be adjusted according to the actual situation in specific use, which is not limited in the present application.

[0033] In the embodiment of the present application, the threshold value of each classification feature in the potential decision rule is determined by the representation of the aquaculture pond dam distribution grid file in the local area. Specifically, a new genetic algorithm, which can be a gene algorithm, is applied, and based on the potential decision rule and the aquaculture pond dam distribution grid file in the local area, the threshold value of each classification feature in the potential decision rule is adjusted by continuously iterating and optimizing the gene algorithm. The accuracy is calculated and evaluated by comparing the rule extraction result with the aquaculture pond dam distribution grid file in the local area, and the iteration is terminated when the accuracy no longer improves, so that the optimized decision rule and the corresponding accuracy are obtained. At this time, the optimized decision rule and the corresponding accuracy are recorded for subsequent analysis. In a specific embodiment, the number of decision rules is 2000. It should be particularly noted that the number of decision rules above is an example number obtained by better experimental effect in the embodiment, and the number can be adjusted according to the actual situation in specific use, which is not limited in the present application.

[0034] In the embodiment of the present application, the threshold value of each classification feature in the potential decision rule is determined by the representation of the aquaculture pond dam distribution grid file in the local area. Specifically, a new genetic algorithm, which can be a gene algorithm, is applied, and based on the potential decision rule and the aquaculture pond dam distribution grid file in the local area, the threshold value of each classification feature in the potential decision rule is adjusted by continuously iterating and optimizing the gene algorithm. The accuracy is calculated and evaluated by comparing the rule extraction result with the aquaculture pond dam distribution grid file in the local area, and the iteration is terminated when the accuracy no longer improves, so that the optimized decision rule and the corresponding accuracy are obtained. At this time, the optimized decision rule and the corresponding accuracy are recorded for subsequent analysis. In a specific embodiment, the number of decision rules is 2000. It should be particularly noted that the number of decision rules above is an example number obtained by better experimental effect in the embodiment, and the number can be adjusted according to the actual situation in specific use, which is not limited in the present application. re2), Normalized Difference Snow Index (NDSI), Vertically Transmitted, Horizontally Received (VH), Normalized Difference Water Index (NDWI), Band 5 / Band 4 reflectance ratio (B5 / B4), and Normalized Difference Vegetation Index using Red Edge band 4 (NDVI re4 ), and the optimal decision rule is NDVI re2 ≥ 0.027, NDSI ≥ -0.235, -23.719 ≤ VH < -15.140, NDWI ≥ -0.365, B5 / B4 < 3.124, and NDVI re4 < 0.325, drawing the pixel response conditions of each component of the optimal decision rule, analyzing the roles of each component, further analyzing the components with core roles according to the analysis result, and mining the internal logic of the components. For examples, please refer to Figure 2 , which shows a schematic diagram of an optimal decision rule of aquaculture pond dam remote sensing recognition and pixel response conditions of each component of the optimal decision rule provided by an embodiment of the present application. Part (a) of the diagram is the pixel response condition corresponding to NDVI re2 ≥ 0.027 in the optimal decision rule, part (e) of the diagram is the pixel response condition corresponding to NDSI ≥ -0.235 in the optimal decision rule, part (f) of the diagram is the pixel response condition corresponding to VH < -15.140 in the optimal decision rule, part (g) of the diagram is the pixel response condition corresponding to -23.719 ≤ VH in the optimal decision rule, part (h) of the diagram is the pixel response condition corresponding to NDWI ≥ -0.365 in the optimal decision rule, part (i) of the diagram is the pixel response condition corresponding to B5 / B4 < 3.124 in the optimal decision rule, and part (j) of the diagram is the pixel response condition corresponding to NDVI re4 < 0.325 in the optimal decision rule, part (b) of the diagram is the pixel response condition corresponding to NDVI re2 ≥ 0.027 and NDSI ≥ -0.235 in the optimal decision rule, part (c) of the diagram is the pixel response condition corresponding to NDVI re2 ≥ 0.027, NDSI ≥ -0.235, and VH < -15.140 in the optimal decision rule, and part (d) of the diagram is the pixel response condition corresponding to NDVI re2≥0.027, NDSI≥-0.235, VH<-15.140, -23.719≤VH, the pixel response condition corresponding to the part (k) of the figure is the pixel response condition corresponding to the optimal decision rule, and the pixel response condition corresponding to the part (d) of the figure is similar in performance, and the following conclusions can be drawn: NDVI re2 ≥0.027 can effectively identify low water content land objects and dikes, and NDSI≥-0.235 is obviously reacted to water bodies and adjacent dikes, and -23.719≤VH<-15.140 can exclude internal interference of water bodies, and only through the above three conditions, the identification accuracy similar to the complete rule can be realized, and NDWI≥-0.365 has a similar effect with VH, and the utility of the remaining two components is insufficient. In summary, the core part of the optimal decision rule for remote sensing extraction of aquaculture pond dikes (i.e., the core decision rule) is NDVI re2 ≥0.027, NDSI≥-0.235. It should be particularly pointed out that the core decision rule described in the present application is not limited to a specific numerical combination, and any rule expression method with similar classification meaning and discrimination ability should be considered as an equivalent technical solution of the present application, and also within the protection scope of the present application.

[0035] Step S140, analyzing the response of the optimal decision rule and the core decision rule to different land cover types to determine the application range of the decision rule.

[0036] In the embodiments of the present application, the response of the optimal decision rule and the core decision rule to different land cover types is analyzed to determine the application range of the decision rule. Specifically, the pixel response conditions of the optimal decision rule and the core decision rule in nine typical land cover types are analyzed, the land cover types include aquaculture pond dikes, tidal flats, salt marsh vegetation, mangroves, impervious surfaces, dry land, paddy fields, forest land and grassland, a total of 9 kinds, the misclassification of the decision rule on different land cover types is determined, and the application range of the decision rule is determined. For example, please refer to Figure 3Fig. 1 shows a schematic diagram of pixel response conditions of an optimal decision rule and its core decision rule for different land cover types, wherein (a) is a pixel response condition of the optimal decision rule and its core decision rule for a fish pond dam, (b) is a pixel response condition of the optimal decision rule and its core decision rule for a tidal flat, (c) is a pixel response condition of the optimal decision rule and its core decision rule for a salt marsh vegetation, (d) is a pixel response condition of the optimal decision rule and its core decision rule for a mangrove forest, (e) is a pixel response condition of the optimal decision rule and its core decision rule for an impervious surface, (f) is a pixel response condition of the optimal decision rule and its core decision rule for a dry land, (g) is a pixel response condition of the optimal decision rule and its core decision rule for a paddy field, (h) is a pixel response condition of the optimal decision rule and its core decision rule for a forest land, and (i) is a pixel response condition of the optimal decision rule and its core decision rule for a grassland. The upper left corner of each of (a) to (i) is an original remote sensing image of each land cover type, the lower left corner is a pixel response condition of the optimal decision rule for each land cover type, the upper right corner is a pixel response condition of the core part (i.e., the core decision rule) of the optimal decision rule for each land cover type, and the lower right corner is a pixel response condition of a random forest model for each land cover type. The core decision rule has high sensitivity to the dam and can achieve an identification accuracy similar to that of the random forest model, and can effectively identify representative pixels in each target land class. However, when dealing with mixed pixels formed at the junction of the dam and the water body, the random forest model has a small amount of misjudgment, and some non-dam pixels close to the water body are mistakenly included. In contrast, the random forest model trained by a series of classification features still cannot completely solve the interference caused by the mixed pixels.

[0037] In step S150, the fish pond dam remote sensing image of the target area is extracted according to the core decision rule.

[0038] In the embodiments of the present application, according to the finally selected core decision rule, the fish pond dam remote sensing image in the target area is efficiently extracted. Specifically, the core decision rule is applied to classify the remote sensing image of the target area into 0 value and 1 value, and the 1 value part of the remote sensing image is extracted, so that the fish pond dam remote sensing image of the target area is obtained. For example, refer to Figure 4 Fig. 2 shows a schematic diagram of a fish pond dam remote sensing image of a target area based on a core decision rule, wherein the left part is a remote sensing image of the target area, and the right part is a fish pond dam remote sensing image of the target area based on the core decision rule.

[0039] In summary, the embodiment of the present application provides a kind of based on decision rule's aquaculture pond dam remote sensing extraction method, obtains the remote sensing image data of target area, filters the remote sensing image of clear imaging, no cloud layer or ice and snow interference, ensure the reliability of data, guarantee data quality, reduce the influence of noise on feature extraction, and extract the classification feature in the remote sensing image data, fuse multiple classification features to construct classification feature dataset, fully capture the high reflection characteristics and morphological characteristics of dam, enhance the feature distinguishing ability, construct local aquaculture pond dam dataset, facilitate subsequent rule construction;According to the classification feature dataset, potential decision rule is constructed, based on the local aquaculture pond dam dataset, each classification feature threshold in the potential decision rule is iteratively optimized by genetic algorithm, and the accuracy is evaluated until the iteration is terminated when the accuracy is no longer improved, to ensure that the rule accuracy is maximized, and finally the optimized decision rule and the corresponding accuracy are obtained, the decision rule is automatically generated throughout, avoids the dependence on artificial experience;According to the optimized decision rule, the optimal decision rule and the core decision rule are obtained, by analyzing the pixel response state of each component in the high accuracy rule, the core part in the optimal decision rule is determined, the internal logic is mined, and the rule interpretability is strengthened, by eliminating the redundant components in the rule, only the core decision rule is retained, the rule simplification is completed, the accuracy is maintained while the operation efficiency is improved, and, the core decision rule focuses on universal characteristics, does not depend on specific numerical combination, is convenient for cross-regional adaptation, and is the basis for rule transferability;The response of the optimal decision rule and the core decision rule to different land cover types is analyzed, the application range of decision rule is determined, the controllability of misclassification is guaranteed, the generalization ability of the core decision rule is improved, the overfitting risk is reduced, and the rule reliability is improved;According to the core decision rule, the aquaculture pond dam remote sensing image of the target area is extracted, the extraction process of remote sensing image is simplified and the high precision of extraction is maintained, the efficient and accurate extraction of aquaculture pond dam remote sensing image is realized, the remote sensing image extraction depends on lightweight rule rather than calculation model, can quickly process large area image, improve the large-scale applicability of the core decision rule, and the transferability of the rule is stronger, only threshold needs to be adjusted to adapt to different regional requirements, can meet long-term dynamic monitoring demand.The technical scheme provided in the application can finally improve the transparency and interpretability of the aquaculture pond dam identification process, realizes efficient and accurate extraction of the aquaculture pond dam, overcomes the problems of complex model structure, unclear classification basis and related uninterpretability of traditional deep learning and machine learning methods, and meanwhile, the technical scheme provided in the application fully considers the response characteristics of different land covers in the rule mining process, so that the core decision rule has strong generalization ability and good migration ability, can be flexibly applied to the aquaculture pond dam identification requirements in different regions, different scales and different remote sensing image data sources, has lower dependence on the number of samples compared with the traditional method which depends on a large number of labeled samples, has higher extraction efficiency, can more stably identify the dam structure with narrow width but obvious reflection characteristics, is particularly suitable for dam change monitoring in complex aquaculture regions with large scale and multiple time phases, in addition, the aquaculture pond dam has stable form and high reflection characteristics in remote sensing images, so that the robustness and applicability of dam identification can be improved, and the technical scheme has good practical value and promotion prospect, and provides reliable data support and technical support for regional aquaculture resource management and ecological environment evaluation.

[0040] It should be noted that, in the embodiments of the present application, if the above-mentioned aquaculture pond dam remote sensing extraction method based on decision rules is realized in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical scheme of the embodiments of the present application or the part that contributes to the related art can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing an electronic device to execute all or part of the method described in the embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a magnetic disk or an optical disk, and various media that can store program codes. Thus, the embodiments of the present application are not limited to any specific hardware and software combination.

[0041] Correspondingly, the embodiments of the present application provide a computer-readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the steps of the aquaculture pond dam remote sensing extraction method based on decision rules in any of the above-mentioned embodiments. Correspondingly, the embodiments of the present application also provide a computer program product for implementing the steps of the aquaculture pond dam remote sensing extraction method based on decision rules in any of the above-mentioned embodiments when the computer program product is executed by a processor of an electronic device.

[0042] Based on the same technical concept, the embodiments of the present application provide an electronic device for implementing the aquaculture pond dam remote sensing extraction method based on decision rules described in the above-mentioned method embodiments. Figure 5A hardware entity schematic diagram of an electronic device provided in the embodiments of the present application is shown in Figure 5 The electronic device 500 includes a memory 510 and a processor 520, the memory 510 stores a computer program executable on the processor 520, and the processor 520 implements the steps in any of the decision rule based aquaculture pond dam remote sensing extraction methods according to the embodiments of the present application when executing the program.

[0043] The memory 510 is configured to store instructions and applications executable by the processor 520, and can also cache data to be processed by the processor 520 and modules in the electronic device (for example, image data, audio data, voice communication data and video communication data), which can be implemented by FLASH or Random Access Memory (RAM).

[0044] The processor 520 implements the steps in any of the decision rule based aquaculture pond dam remote sensing extraction methods when executing the program. The processor 520 generally controls the overall operation of the electronic device 500.

[0045] The above processor can be at least one of an Application Specific Integrated Circuit (ASIC), a Digital Signal Processor (DSP), a Digital Signal Processing Device (DSPD), a Programmable Logic Device (PLD), a Field Programmable Gate Array (FPGA), a Central Processing Unit (CPU), a controller, a microcontroller, and a microprocessor. It can be understood that the electronic device realizing the functions of the above processor can also be other devices, and the embodiments of the present application are not limited specifically.

[0046] The computer storage medium / memory can be a Read Only Memory (ROM), a Programmable Read-Only Memory (PROM), an Erasable Programmable Read-Only Memory (EPROM), an Electrically Erasable Programmable Read-Only Memory (EEPROM), a Ferromagnetic Random Access Memory (FRAM), a Flash Memory, a magnetic surface storage, an optical disc, a Compact Disc Read-Only Memory (CD-ROM), or the like memory; or can be various electronic devices including one or any combination of the above memories, such as a mobile phone, a computer, a tablet device, a personal digital assistant, and the like.

[0047] It should be noted that the above description of the storage medium and device embodiments is similar to the description of the above method embodiments, and has similar beneficial effects as the method embodiments. For technical details not disclosed in the storage medium and device embodiments of the present application, please refer to the description of the method embodiments of the present application for understanding.

[0048] It should be understood that the "one embodiment" or "an embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that the size of the sequence number of each process in various embodiments of the present application does not mean the execution order, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application. The sequence number of the above embodiments of the present application is only for description, not representing the advantages and disadvantages of the embodiments.

[0049] It should be noted that, in the present document, the terms "comprising", "containing", or any other similar term are intended to encompass non-exclusive inclusions, such that a process, method, article, or apparatus that comprises a list of elements does not necessarily include those elements only, but can include other elements not expressly listed, or can include elements inherent in such process, method, article, or apparatus. Without more limitations, an element defined by the phrase "comprising a" does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0050] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative, for example, the division of the units is only a logical functional division, and actual implementation can have another division manner, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed components can be through some interface, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0051] The units described above as separate components can or can not be physically separated, and the components shown as units can or can not be physical units; they can be located in one place or distributed on multiple network units; and part or all of the units can be selected according to actual needs to achieve the purpose of the embodiments of the present application.

[0052] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be realized in the form of hardware or in the form of hardware plus software functional unit.

[0053] Alternatively, the integrated unit of the present application, if implemented in the form of a software function module and sold or used as an independent product, can also be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the embodiments of the present application can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for causing an apparatus to perform all or part of the methods described in the embodiments of the present application. The foregoing storage medium includes: mobile storage devices, ROM, magnetic or optical disks, and various other media that can store program codes.

[0054] The methods disclosed in the several method embodiments provided by the present application can be combined arbitrarily without conflict to obtain new method embodiments.

[0055] The features disclosed in the several method or device embodiments provided by the present application can be combined arbitrarily without conflict to obtain new method embodiments or device embodiments.

[0056] The above merely provides the implementation manners of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A remote sensing extraction method for aquaculture pond embankments based on decision rules, characterized in that, The method includes: Acquire remote sensing image data of the target area, extract classification features from the remote sensing image data, construct a classification feature dataset, and construct a local aquaculture pond dam dataset; Based on the classification feature dataset, potential decision rules are constructed. Based on the local aquaculture pond dam dataset, the thresholds of each classification feature in the potential decision rules are iteratively optimized using a gene algorithm, and the accuracy is evaluated. The iteration is terminated when the accuracy no longer improves, resulting in the optimized decision rules and the corresponding accuracy. The optimal decision rule and the core decision rule are obtained based on the optimized decision rule. Analyze the responses of the optimal decision rule and the core decision rule to different land cover types to determine the scope of application of the decision rules; Based on the core decision-making rules, remote sensing images of aquaculture pond embankments in the target area are extracted.

2. The method according to claim 1, characterized in that, The remote sensing image data includes clearly imaged synthetic aperture radar data and multispectral satellite data. The extraction of classification features from the remote sensing image data and the construction of a classification feature dataset include: Multispectral band features, band ratios, and commonly used remote sensing indices are extracted from the multispectral satellite data in the remote sensing image data. The backscattering coefficients are extracted from the synthetic aperture radar data in the remote sensing image data; The classification feature dataset is constructed based on the multispectral band features, the band ratios, the commonly used remote sensing indices, and the backscattering coefficients.

3. The method according to claim 1, characterized in that, The local aquaculture pond dam dataset includes a raster file showing the distribution of aquaculture pond dams in a local area. A value of 1 in the raster file represents an aquaculture pond dam area, and a value of 0 in the raster file represents a non-aquaculture pond dam area.

4. The method according to claim 1, characterized in that, The step of constructing potential decision rules based on the classification feature dataset includes: Based on the classification feature dataset, a genetic algorithm is used to randomly combine classification features, and a logical relationship between the classification features and a threshold is randomly set to generate a set of potential decision rules. The logical relationship includes classification features greater than the threshold and classification features less than the threshold. The potential decision rules are obtained by filtering out rules with 3 to 7 components from the set of potential decision rules and removing duplicate rules.

5. The method according to claim 1, characterized in that, The process of obtaining the optimal decision rule and the core decision rule based on the optimized decision rule includes: The optimized decision rules are arranged in descending order of accuracy to obtain the optimal decision rules. The pixel response status of each component of the optimal decision rule is plotted, the role of each component is analyzed, and the core part of the optimal decision rule is obtained. The core part of the optimal decision rule is the core decision rule.

6. The method according to claim 1, characterized in that, The analysis of the optimal decision rule and the core decision rule in response to different land cover types determines the scope of application of the decision rules, including: The pixel response of the optimal decision rule and the core decision rule is analyzed for nine typical land cover types: aquaculture pond embankment, tidal flat, salt marsh vegetation, mangrove, impermeable surface, dry land, paddy field, woodland and grassland. The misclassification of the decision rule on different land cover types is clarified, and the scope of application of the decision rule is determined.

7. The method according to claim 1, characterized in that, The step of extracting remote sensing images of aquaculture pond embankments in the target area according to the core decision rule includes: By applying the core decision rule, the remote sensing images of the target area are classified into 0 and 1 values, and the 1 value portion of the remote sensing images is extracted to obtain the remote sensing images of the aquaculture pond dam.

8. An electronic device comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 7.

Images