Gutinel No.2 image photovoltaic panel extraction method and system

Abstract

The invention discloses a sentinel No.2 image photovoltaic panel extraction method and system, and relates to the technical field of remote sensing image processing, and the method comprises the steps: obtaining Sentinel-2L2A-level median synthesis image data of a training region and a test region based on GEE; obtaining photovoltaic panel and non-photovoltaic panel samples of the training area based on GEE, and obtaining spectral features and index features of the photovoltaic panel and non-photovoltaic panel samples; carrying out photovoltaic panel and non-photovoltaic panel pixel labeling on the training and testing area through ArcMap; performing feature separability analysis on the spectral features and the index features, and finally determining high separability features; and carrying out weighted random combination on the high-separability features, and carrying out optimal threshold search by depending on an optimization algorithm to determine a final optimal feature combination. According to the method, the optimal threshold value is searched in the high-separability feature combination based on the optimization algorithm so as to obtain the optimal classification feature combination of the photovoltaic panel. According to the method, the extraction efficiency of the photovoltaic panel is effectively improved, and the method is suitable for large-scale photovoltaic panel monitoring application.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing image processing, and particularly to a method and system for extracting photovoltaic panels from Sentinel-2 images. Background Art

[0002] The accurate extraction of photovoltaic panels is of great significance in remote sensing image analysis. Especially in the fields of environmental monitoring, energy management, and urban planning, the precise identification of photovoltaic panels can provide strong data support for the distribution and maintenance of clean energy facilities. Due to the obvious differences in the spectral reflection characteristics of photovoltaic panels and surrounding ground objects in different bands, their spectral characteristics are manifested as unique spectral curves in remote sensing images. Therefore, the automatic extraction of photovoltaic panels through remote sensing images has high feasibility and accuracy.

[0003] In the application of Sentinel-2 images, the extraction of photovoltaic panels mainly relies on the analysis of the characteristics of each band in the images. Through feature separability analysis, the performance of different features in distinguishing photovoltaic panels from background ground objects can be effectively evaluated, so as to select the most favorable feature combination for photovoltaic panel identification. Subsequently, the genetic algorithm is used to optimize these features to further improve the accuracy and efficiency of the extraction process. The optimization of spectral indices and feature combinations can not only effectively capture the reflection differences of photovoltaic panels in multiple bands, but also improve the stability and reliability of photovoltaic panel extraction in complex environments, ensuring the accurate identification of photovoltaic panel areas under different lighting conditions and ground object interferences. Summary of the Invention

[0004] In view of the problems existing in the existing methods and systems for extracting photovoltaic panels from Sentinel-2 images, the present invention is proposed.

[0005] To solve the above technical problems, the present invention provides the following technical solutions:

[0006] In a first aspect, an embodiment of the present invention provides a method for extracting photovoltaic panels from Sentinel-2 images, which includes the following steps:

[0007] Obtain the median composite image data of Sentinel-2 L2A level for the training area and the test area based on GEE;

[0008] Obtain the photovoltaic panel and non-photovoltaic panel samples in the training area based on GEE, and obtain their spectral characteristics and index characteristics;

[0009] Label the photovoltaic panel and non-photovoltaic panel pixels in the training and test areas through ArcMap;

[0010] Conduct feature separability analysis on the spectral characteristics and index characteristics, and finally determine the highly separable features;

[0011] Perform weighted random combination on highly separable features and rely on an optimization algorithm to search for the optimal threshold to determine the final optimal feature combination.

[0012] As a preferred solution of the method for extracting photovoltaic panels from Sentinel-2 images according to the present invention, wherein: the spectral features and index features include the blue band (B2), green band (B3), red band (B4), red edge bands (B5, B6, B7), near-infrared bands (B8, B8A), short-wave infrared bands (B11, B12).

[0013] As a preferred solution of the method for extracting photovoltaic panels from Sentinel-2 images according to the present invention, wherein: the steps of performing feature separability analysis on spectral features and index features include,

[0014] First, obtain the maximum and minimum values of each feature;

[0015] By analyzing the maximum and minimum values of the features, determine the threshold interval for searching photovoltaic panel sample points;

[0016] The moving step size of the lower limit of the threshold interval should be synchronized with the moving step size of the upper limit;

[0017] After the threshold interval is determined, calculate the ratio of the number of photovoltaic panel samples to the number of non-photovoltaic panel samples within this interval.

[0018] As a preferred solution of the method for extracting photovoltaic panels from Sentinel-2 images according to the present invention, wherein: the calculation formula for the ratio of the number of photovoltaic panel samples to the number of non-photovoltaic panel samples is,

[0019]

[0020] In the formula, Sample_Panel is the number of photovoltaic panel samples within a specific threshold interval of a certain feature, Sample-_Non_Panel is the number of non-photovoltaic panel samples within the specific threshold interval of this feature, and N is the purity of this feature.

[0021] As a preferred solution of the method for extracting photovoltaic panels from Sentinel-2 images according to the present invention, wherein: the inputs of the constructed optimization algorithm include,

[0022] Feature combination, feature threshold interval, and training area image;

[0023] The feature threshold interval will accelerate the convergence speed of the optimization algorithm;

[0024] The input of the training area image will be restricted by different thresholds of the feature combination;

[0025] Finally, an early stopping strategy will be set.

[0026] As a preferred solution of the method for extracting photovoltaic panels from Sentinel-2 images according to the present invention, among them: in the parameters of the optimization algorithm, the fitness function is the IOU function, which is specifically expressed as:

[0027]

[0028] In the formula, A represents the classification result generated based on the feature combination, and B represents the annotation result obtained in step two.

[0029] As a preferred solution of the method for extracting photovoltaic panels from Sentinel-2 images according to the present invention, among them: the feature separability analysis is carried out in the Python environment;

[0030] Moreover, the output result of the feature separability analysis includes the threshold interval and purity value of a single feature.

[0031] In the second aspect, an embodiment of the present invention provides a system for extracting photovoltaic panels from Sentinel-2 images, which includes a data acquisition module, a feature separability analysis module, a combination module, and an optimization algorithm module;

[0032] The data acquisition module is responsible for acquiring the Sentinel-2 L2A median composite image data of the training area and the test area;

[0033] The feature separability analysis module conducts feature separability analysis on spectral features and index features;

[0034] The combination module performs weighted random combination on highly separable features;

[0035] The optimization algorithm module constructs an optimization algorithm for searching for the optimal value of feature combination.

[0036] In the third aspect, an embodiment of the present invention provides a computer device, including a memory and a processor, where: when the processor executes the computer program, any step of the above-mentioned method for extracting photovoltaic panels from Sentinel-2 images is implemented.

[0037] In the fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, where: when the computer program is executed by the processor, any step of the above-mentioned method for extracting photovoltaic panels from Sentinel-2 images is implemented.

[0038] The beneficial effects of the present invention are as follows: By implementing efficient automated extraction of photovoltaic panels based on feature separability analysis and optimization algorithms, this significantly reduces manual inspection and modification. The present invention evaluates feature separability by introducing sample purity information within a threshold range; the combined feature method based on weights will accelerate the convergence speed of the optimization algorithm. In summary, the photovoltaic panel extraction method based on feature separability analysis and optimization algorithms provided by the present invention has significant advantages of being automatic, efficient, and self-explainable, which will provide an innovative solution for photovoltaic panel identification in the field of remote sensing image analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings. Among them:

[0040] Figure 1 It is a flowchart of the photovoltaic panel extraction method for Sentinel-2 images.

[0041] Figure 2 It is a schematic diagram of the training area and test area of the photovoltaic panel extraction method for Sentinel-2 images.

[0042] Figure 3 It is a diagram of the photovoltaic panel classification features and their separability analysis of the photovoltaic panel extraction method for Sentinel-2 images.

[0043] Figure 4 It is a schematic diagram of the photovoltaic panel results of the photovoltaic panel extraction method for Sentinel-2 images.

[0044] Figure 5 It is a schematic diagram of the feature thresholds and their corresponding classification accuracies of the photovoltaic panel extraction method for Sentinel-2 images. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0045] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0046] In the following description, many specific details are set forth to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0047] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure or characteristic that may be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that exclude each other with other embodiments.

[0048] The present invention is described in detail in conjunction with schematic diagrams. When detailing the embodiments of the present invention, for the convenience of explanation, the cross-sectional views showing the device structure will be locally enlarged out of the general proportion, and the schematic diagrams are only examples and should not limit the scope of protection of the present invention herein. In addition, the actual production should include three-dimensional spatial dimensions of length, width and depth.

[0049] At the same time, in the description of the present invention, it should be noted that the orientation or positional relationship indicated by terms such as "upper, lower, inner and outer" is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be construed as a limitation of the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0050] Unless otherwise clearly defined and limited in the present invention, the terms "installation, connection, and connection" shall be understood in a broad sense. For example: it may be a fixed connection, a detachable connection or an integral connection; similarly, it may be a mechanical connection, an electrical connection or a direct connection, or may be indirectly connected through an intermediate medium, or may be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0051] Embodiment 1

[0052] Referring to Figures 1 to 5 , this is the first embodiment of the present invention. This embodiment provides a method for extracting photovoltaic panels from Sentinel-2 images, including the following steps.

[0053] S1. Obtain the median composite image data of Sentinel-2 L2A level for the training area and the test area based on GEE.

[0054] Based on GEE, the training area and the test area are selected, that is Figure 2 As shown, Sentinel-2 L2A level data is selected for the image data. After cropping the obtained image, the image of the required area is finally obtained. It should be noted that the obtained image is the median composite image from January 2024 to September 2024.

[0055] The role of median synthesis is to effectively eliminate image interference caused by cloud cover, meteorological changes, or sensor noise. By calculating the median of pixel values at the same location over multiple time points, a "representative" value that reflects surface features during these time periods can be selected, thereby generating a clearer and more stable image.

[0056] On the Google Earth Engine (GEE) cloud platform, Sentinel-2 Level 2A data for training and testing is obtained. GEE provides powerful cloud computing capabilities, enabling efficient retrieval of multi-temporal satellite images covering the training area and validation area.

[0057] The Level 2A image data obtained in this step has been preprocessed, including radiometric calibration, atmospheric correction, and orthorectification. Completion of these preprocessing steps provides an accurate and consistent data basis for subsequent image analysis.

[0058] S2. Based on GEE, photovoltaic panel and non-photovoltaic panel samples in the training area are obtained, and their spectral features and index features are acquired.

[0059] Spectral features and index features include the blue band (B2), green band (B3), red band (B4), red edge bands (B5, B6, B7), near-infrared bands (B8, B8A), shortwave infrared bands (B11, B12);

[0060] These spectral data will be used for subsequent vegetation index calculation and feature analysis. The selected index features include ratio index features (B11 / B12, B12 / B11, B2 / B4, B4 / B2, B3 / B4, B4 / B3, B5 / B3, B6 / B3, B7 / B3, B4 / B5, B5 / B4, B6 / B5, B7 / B), normalized difference vegetation index (NDVI), normalized difference water index (NDWI), normalized difference moisture index (NDMI), canopy isolation index (CI), enhanced vegetation index (EVI), and normalized difference built-up index (NDBI).

[0061] The calculation formulas for each index feature are as follows:

[0062]

[0063] In the formula, B4 is the red band of the Sentinel-2 image, and B8 is the near-infrared band.

[0064]

[0065] In the formula, B3 is the green band of the Sentinel-2 image, and B4 is the red band.

[0066]

[0067] In the formula, B8 is the near-infrared band of the Sentinel-2 image, and B11 is the short-wave infrared band.

[0068]

[0069] In the formula, B7 is the red-edge band of the Sentinel-2 image, and B8 is the near-infrared band.

[0070]

[0071] In the formula, B2 is the blue band of the Sentinel-2 image, B4 is the red band, and B8 is the near-infrared band.

[0072] S3. Use ArcMap to label the pixels of photovoltaic panels and non-photovoltaic panels in the training and test areas.

[0073] The labeling result is an 8-bit binary image with values of 0 and 255.

[0074] S4. Conduct feature separability analysis on spectral features and index features and finally determine highly separable features.

[0075] The feature separability analysis is carried out in the Python environment;

[0076] Moreover, the output results of the feature separability analysis include the threshold interval and purity value of a single feature.

[0077] First, obtain the maximum and minimum values of each feature instead of sorting the samples, which will significantly reduce the consumption of computing resources. Second, determine the threshold interval that can cover at least 90% of the photovoltaic panel sample points by analyzing the maximum and minimum values of the features. To prevent the search for the threshold interval from falling into multiple solutions and local optimal solutions and to accelerate convergence, we keep the moving step size of the lower limit of the threshold interval synchronized with the moving step size of the upper limit. This search process is implemented based on the Python environment.

[0078] After the threshold interval is determined, it is necessary to calculate the ratio of the number of photovoltaic panel samples (Sample_Panel) to the number of non-photovoltaic panel samples (Sample_Non_Panel) within this interval. The larger the ratio, the higher the purity (N) of this feature in distinguishing between photovoltaic panel and non-photovoltaic panel samples, which means that the feature has higher separability. This method can effectively screen out the features that contribute more to the classification task and provide a scientific basis for subsequent feature combination and optimization algorithms. The calculation formula is as follows:

[0079]

[0080] In the formula, Sample_Panel is the number of photovoltaic panel samples within a specific threshold range of a certain feature, Sample_Non_Panel is the number of non-photovoltaic panel samples within the specific threshold range of this feature, and N is the purity of this feature.

[0081] This step finally selects the top 10 N values as alternative features for subsequent operations.

[0082] S5. Randomly combine the highly separable features with weights and perform an optimal threshold search relying on an optimization algorithm to determine the final optimal feature combination.

[0083] The inputs to the constructed optimization algorithm include,

[0084] feature combinations, feature threshold ranges, and training area images;

[0085] The feature threshold range will accelerate the convergence speed of the optimization algorithm;

[0086] while the input of the training area image will be restricted by different thresholds of the feature combination;

[0087] Finally, an early stopping strategy will be set.

[0088] Perform an optimal value search for feature combinations based on the optimization algorithm (genetic algorithm) constructed in the Python environment and the alternative feature combinations provided in step five. The inputs to the constructed genetic algorithm include three parts, namely feature combinations, the feature threshold ranges obtained in step four, and the training area images obtained in step one. The threshold ranges obtained in step four will accelerate the convergence speed of the genetic algorithm, while the input of the training area image will be restricted by different thresholds of the feature combination. The genetic algorithm will perform an optimal threshold search based on the classification result accuracy of the feature combination under different thresholds. Among the parameters of the genetic algorithm, the population size is 200, the number of iterations is 300 times, the fitness function is the IOU function, and other parameters remain default. The expression of the IOU function is:

[0089]

[0090] In the formula, A represents the classification result generated based on the feature combination, and B represents the annotation result obtained in step two.

[0091] This step obtains the optimal threshold of each feature combination and its maximum fitness result. By analyzing the fitness results, a method for extracting photovoltaic panels from Sentinel-2 images based on feature separability analysis and optimization algorithm is finally constructed, as Figure 3 and Figure 3 shown. The finally obtained feature combination is "B12 / B11 < 0.715 & B4 / B5 > 0.773 & CI3 < 0.148".

[0092] In summary, the efficient and automatic extraction of photovoltaic panels is achieved through feature separability analysis and optimization algorithms, which significantly reduces manual inspection and modification. The present invention evaluates feature separability by introducing sample purity information within a threshold range; the combined feature method based on weights will accelerate the convergence rate of the optimization algorithm. In summary, the photovoltaic panel extraction method based on feature separability analysis and optimization algorithms provided by the present invention has significant advantages of being automatic, efficient, and self-explanatory, which will provide an innovative solution for the identification of photovoltaic panels in the field of remote sensing image analysis.

[0093] Embodiment 2

[0094] Based on the first embodiment, this embodiment further provides a Sentinel-2 image photovoltaic panel extraction system, including a data acquisition module, a feature separability analysis module, a combination module, and an optimization algorithm module;

[0095] The data acquisition module is responsible for acquiring Sentinel-2 Level 2A median composite image data of the training area and the test area;

[0096] The feature separability analysis module performs feature separability analysis on spectral features and index features;

[0097] The combination module randomly combines highly separable features with weights;

[0098] The optimization algorithm module constructs an optimization algorithm for searching the optimal value of feature combination.

[0099] This embodiment also provides a computer device applicable to the case of the Sentinel-2 image photovoltaic panel extraction method, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the Sentinel-2 image photovoltaic panel extraction method proposed in the above embodiment.

[0100] The computer device may be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be achieved through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0101] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for extracting photovoltaic panels from Sentinel-2 images as proposed in the above embodiment.

[0102] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiment belong to the same inventive concept. Technical details not described in detail in this embodiment can be referred to in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0103] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not restrictive. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for extracting photovoltaic panels from Sentinel-2 images, characterized in that: The following steps are included: Based on GEE, the Sentinel-2L2A-level median synthetic image data of the training and test areas were obtained; Based on GEE, we obtain the samples of photovoltaic panels and non-photovoltaic panels in the training area, and obtain their spectral characteristics and index characteristics; The pixels of photovoltaic panels and non-photovoltaic panels in the training and test areas were labeled using ArcMap; Perform feature separability analysis on spectral features and index features and finally determine highly separable features; The highly separable features are randomly combined with weights, and the optimal threshold is searched by the optimization algorithm to determine the final optimal feature combination.

2. The method for extracting photovoltaic panels from Sentinel-2 images as claimed in claim 1, characterized in that: The spectral characteristics and index characteristics include a blue light band (B2), a green light band (B3), a red light band (B4), a red edge band (B5, B6, B7), a near infrared band (B8, B8A), and a short-wave infrared band (B11, B12).

3. The method for extracting photovoltaic panels from Sentinel-2 images as claimed in claim 2, characterized in that: The steps for characterizing the spectral features and index features include: First, obtain the maximum and minimum values ​​of each feature; By analyzing the maximum and minimum values ​​of the features, the threshold interval for searching the photovoltaic panel sample points is determined; Keep the lower limit moving step size of the threshold interval synchronized with the upper limit moving step size; After the threshold interval is determined, the ratio of the number of PV panel samples to the number of non-PV panel samples within the interval is calculated.

4. The method for extracting photovoltaic panels from Sentinel-2 images as claimed in claim 3, characterized in that: The calculation formula for the ratio of the number of photovoltaic panel samples to the number of non-photovoltaic panel samples is: Where Sample_Panel is the number of PV panel samples within a specific threshold interval of a feature, Sample-_Non_Panel is the number of non-PV panel samples within a specific threshold interval of the feature, and N is the purity of the feature.

5. The method for extracting photovoltaic panels from Sentinel-2 images as claimed in claim 4, characterized in that: The optimization algorithm inputs constructed include, Feature combinations, feature threshold intervals, and training area images; The feature threshold interval will accelerate the convergence speed of the optimization algorithm; The input of the training area image will be constrained by different thresholds of feature combinations; Finally, the early stopping strategy will be set.

6. The method for extracting photovoltaic panels from Sentinel-2 images as claimed in claim 5, characterized in that: Among the parameters of the optimization algorithm, the fitness function is the IOU function, which is specifically expressed as: In the formula, A represents the classification result generated based on feature combination, and B represents the annotation result obtained in step 2.

7. The method for extracting photovoltaic panels from Sentinel-2 images as claimed in claim 6, characterized in that: Feature separability analysis in Python environment; In addition, the output results of feature separability analysis include the threshold range and purity value of a single feature.

8. A Sentinel-2 image photovoltaic panel extraction system, based on the Sentinel-2 image photovoltaic panel extraction method according to any one of claims 1 to 7, characterized in that: It includes data acquisition module, feature separability analysis module, combination module, and optimization algorithm module; The data acquisition module is responsible for acquiring the Sentinel-2L2A-level median synthetic image data of the training area and the test area; The feature separability analysis module performs feature separability analysis on spectral features and index features; The combination module performs weighted random combination of high separability features; The optimization algorithm constructed by the optimization algorithm module is used to search for the optimal value of the feature combination.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the Sentinel-2 image photovoltaic panel extraction method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the Sentinel-2 image photovoltaic panel extraction method described in any one of claims 1 to 7 are implemented.

Images