A coastal wetland remote sensing error detection method and system
Through multi-time phase remote sensing image preprocessing, classification algorithms and error propagation models, combined with high-precision GPS data and multi-source data fusion, the error detection problem in remote sensing monitoring in coastal wetlands is solved, and the accuracy and credibility of monitoring data are significantly improved. , provide a scientific basis for wetland protection.
Patent Information
- Application Number
- CN202411170566.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-26
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-08-26
AI Technical Summary
The traditional remote sensing monitoring method of coastal wetlands has difficulties in error detection, which has affected the accuracy and reliability of monitoring results, and efficient and accurate error detection methods and systems are urgently needed.
Using multi-time phase remote sensing image preprocessing, combined with high-precision GPS equipment to obtain ground measured data, error analysis and correction are performed through classification algorithms and error propagation models, multi-source data fusion and random forest classification algorithms are used to improve detection accuracy, and aerosol optical thickness is introduced Atmospheric correction with atmospheric visibility.
It significantly improves the accuracy and credibility of wetland remote sensing monitoring data, reduces information loss and error accumulation, enhances error analysis capabilities, and provides scientific basis for wetland protection and management decisions.
Smart Images

Figure CN118691980B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of digital image processing, and in particular to a method and system for detecting errors in remote sensing of coastal wetlands. Background Art
[0002] With the increasing prominence of global climate change and environmental issues, the protection and monitoring of coastal wetlands is particularly important. Traditional ground monitoring methods are not only time-consuming and labor-intensive, but also difficult to cover large areas. Therefore, remote sensing technology has gradually become the main means of coastal wetland monitoring. However, various errors will inevitably be introduced in the process of acquiring and processing remote sensing data. If these errors are not detected and corrected in time, they will seriously affect the accuracy and reliability of the monitoring results.
[0003] Therefore, there is an urgent need for an efficient and accurate coastal wetland remote sensing error detection method and system to meet the challenges in current remote sensing data processing and ensure the reliability and application value of monitoring data. Summary of the invention
[0004] In view of this, the present invention proposes a coastal wetland remote sensing error detection method and system, aiming to identify and quantify errors in remote sensing data through efficient and accurate means, so as to provide a scientific basis for subsequent wetland protection and management decisions.
[0005] The present invention proposes a coastal wetland remote sensing error detection method, comprising:
[0006] Acquire multi-temporal remote sensing images, pre-process the remote sensing images, and obtain the remote sensing distribution results of wetland water area and silt plain based on the pre-processed remote sensing images; at the same time, use high-precision GPS equipment to measure the actual distribution results of water body boundaries and silt plains to obtain ground measured data;
[0007] A classification algorithm is used to classify the remote sensing distribution results and the ground measured data, and a preliminary classification result is generated; the classified remote sensing monitoring results are compared and analyzed with the ground measured data, and the position error and area error are calculated; at the same time, the overall accuracy and Kappa coefficient are calculated based on the position error and area error to evaluate the accuracy of the classification results;
[0008] Through multi-source data fusion, the classification results are cross-validated to evaluate the errors between different sensors;
[0009] The error propagation model is used to analyze the influence of the error source of the error on the final result, and the remote sensing data is error corrected according to the analysis result to generate the corrected final monitoring result.
[0010] Preferably, the remote sensing images include optical images, radar images and lidar images.
[0011] Preferably, the remote sensing image is preprocessed, including radiation correction, geometric correction and atmospheric correction;
[0012] Wherein, the radiation correction is used to correct the radiation distortion of the remote sensing image;
[0013] The geometric correction is used to correct the geometric distortion of the remote sensing image;
[0014] The atmospheric correction is used to remove the influence of the atmosphere on the radiation to obtain the true reflectivity and / or the true radiation brightness.
[0015] Preferably, the radiation correction comprises:
[0016] Convert the original DN value observed by remote sensing equipment into surface reflectance or radiation brightness value;
[0017] Obtaining the solar altitude angle and the observation angle, and calculating the radiation correction coefficient according to the solar altitude angle and the observation angle;
[0018] The radiation correction coefficient is applied to correct the original DN value, and a radiation brightness image after radiation correction is generated.
[0019] Preferably, the atmospheric correction comprises:
[0020] Acquire real-time atmospheric parameters, and input the real-time atmospheric parameters into the MODTRAN model to calculate atmospheric transmittance and scattered radiation;
[0021] The atmospheric transmittance and scattered radiation are used to correct the radiation brightness image after radiation correction, and a real reflectance image after atmospheric correction is generated.
[0022] Preferably, the atmospheric transmittance and scattered radiation are used to correct the radiation brightness image after radiation correction, and the atmospheric corrected real reflectance image is generated, which includes the following steps:
[0023] Performing attenuation correction on the radiation brightness image after radiation correction by using the atmospheric transmittance to remove the absorption and scattering effects of the atmosphere on light;
[0024] Using the scattered radiation to compensate the radiation brightness image after radiation correction, so as to correct the radiation enhancement phenomenon caused by atmospheric scattering;
[0025] At the same time, the aerosol optical thickness and the atmospheric visibility are obtained, and the radiation brightness image after the radiation correction is corrected using the aerosol optical thickness and the atmospheric visibility.
[0026] Preferably, obtaining the aerosol optical thickness and the atmospheric visibility, and using the aerosol optical thickness and the atmospheric visibility to correct the radiation brightness image after the radiation correction includes:
[0027] The radiation brightness data after radiation correction is obtained according to the radiation brightness image after radiation correction, which is recorded as Lsur;
[0028] The scattered radiation intensity is estimated based on the aerosol optical thickness and atmospheric visibility. The calculation formula is:
[0029] Lsca = S × AOD × VIS-α;
[0030] Among them, AOD is aerosol optical depth; VIS is atmospheric visibility; Lsca is scattered radiation intensity; S is the scattering coefficient; α is an empirical coefficient related to atmospheric conditions;
[0031] The calculated scattered radiation is subtracted from the radiation-corrected radiance image to remove the atmospheric scattering effect. The calculation formula is:
[0032] Lcor=Lsur-Lsca;
[0033] Wherein, Lcor is the radiant brightness obtained by correcting the radiant brightness image after radiation correction using the aerosol optical thickness and atmospheric visibility.
[0034] Preferably, the geometric correction includes:
[0035] Geo-locating the atmospherically corrected true reflectance image, including establishing a correspondence between the radiance image and a geographic coordinate system using ground control base points with known geographic coordinates;
[0036] The image is resampled and reprojected through affine transformation to eliminate geometric distortion and obtain the geometrically corrected remote sensing image.
[0037] Preferably, a classification algorithm is used to classify the remote sensing distribution results and the ground measured data, and a preliminary classification result is generated; the classified remote sensing monitoring results are compared and analyzed with the ground measured data, and the position error and area error are calculated, including:
[0038] The sense distribution results and ground measured data are classified by random forest to generate classification results; the classification results are cross-validated using k-fold cross validation;
[0039] When the cross-validation result is accurate in classification, the classified remote sensing monitoring results are spatially compared with the ground measured data; the correctness of the classification results is analyzed by calculating the confusion matrix;
[0040] The position error is determined by calculating the spatial position difference between the classification result and the same feature in the ground measured data; the area error is calculated by comparing the classification result and the area in the ground measured data.
[0041] Preferably, the classification results, position error and area error are incorporated into the error propagation model analysis, and error correction is performed on the remote sensing data based on the error analysis results.
[0042] The present invention also proposes a coastal wetland remote sensing error detection system, which is used to implement the coastal wetland remote sensing error detection method mentioned above.
[0043] Compared with the prior art, the present invention has the following beneficial effects:
[0044] First, the present invention introduces multi-source data fusion technology, which is not limited to a single type of remote sensing image, but combines multi-source data such as optical images, radar images and lidar images. The complementarity of this multi-source data can more comprehensively capture the complex changes in the wetland environment and effectively reduce the problem of missing information or error accumulation caused by a single data source. Through cross-validation of multi-source data, the extraction accuracy of key parameters such as wetland water area and silt plain can be significantly improved, providing more reliable basic data for subsequent error analysis.
[0045] Secondly, in the radiation correction, geometric correction and atmospheric correction links, the present invention adopts a more refined processing flow. In particular, in the atmospheric correction process, in addition to the conventional MODTRAN model application, aerosol optical thickness and atmospheric visibility are innovatively introduced as correction factors to further adjust the radiation corrected data. This processing method can more accurately remove the influence of the atmosphere on light propagation, restore the true reflectivity or radiation brightness of the surface, and thus significantly improve the accuracy of remote sensing data.
[0046] Furthermore, the present invention adopts a more scientific method in the error analysis stage. By constructing an error propagation model, the classification results, position errors and area errors are systematically included in a unified framework for analysis, which can clearly reveal the source of errors and their influence on the final results. Based on this analysis result, the remote sensing data can be targeted for error correction to generate more accurate monitoring results. In addition, the combination of random forest classification algorithm and k-fold cross validation further improves the stability and reliability of the classification results.
[0047] Finally, the coastal wetland remote sensing error detection system proposed in this invention not only realizes the automatic execution of the above method, but also has high scalability and flexibility. The system can adjust parameter settings according to actual needs to adapt to wetland remote sensing monitoring tasks in different regions and time periods. At the same time, the system also has a good user interface and interactive functions, which facilitates users to perform operations such as data input, result viewing and error analysis.
[0048] In summary, the coastal wetland remote sensing error detection method and system of the present invention have achieved remarkable results in improving the accuracy of wetland remote sensing monitoring, optimizing data processing procedures, and enhancing error analysis capabilities. Its application will help improve the scientificity and effectiveness of wetland resource management. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:
[0050] Figure 1 This is a flow chart of the error detection method for remote sensing of coastal wetlands. DETAILED DESCRIPTION
[0051] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features described in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0052] This embodiment provides a coastal wetland remote sensing error detection method, including:
[0053] Acquire multi-temporal remote sensing images, pre-process the remote sensing images, and obtain the remote sensing distribution results of wetland water area and silt plain based on the pre-processed remote sensing images; at the same time, use high-precision GPS equipment to measure the actual distribution results of water body boundaries and silt plains to obtain ground measured data;
[0054] A classification algorithm is used to classify the remote sensing distribution results and the ground measured data, and a preliminary classification result is generated; the classified remote sensing monitoring results are compared and analyzed with the ground measured data, and the position error and area error are calculated; at the same time, the overall accuracy and Kappa coefficient are calculated based on the position error and area error to evaluate the accuracy of the classification results;
[0055] Through multi-source data fusion, the classification results are cross-validated to evaluate the errors between different sensors;
[0056] The error propagation model is used to analyze the influence of the error source of the error on the final result, and the remote sensing data is error corrected according to the analysis result to generate the corrected final monitoring result.
[0057] It is understood that this embodiment is intended to introduce an innovative coastal wetland remote sensing error detection technology. The method first involves collecting multi-temporal remote sensing image data and thoroughly preprocessing the image data to ensure its accuracy and availability. After the preprocessing is completed, the remote sensing distribution results of the wetland water area and the silt plain will be obtained based on the processed remote sensing image data. At the same time, a high-precision GPS device will be used to measure the actual distribution results of the water body boundary and the silt plain to obtain ground measured data as a reference for the remote sensing data.
[0058] Next, an advanced classification algorithm is used to classify the remote sensing distribution results and the ground measured data, and a preliminary classification result is generated. Subsequently, these classified remote sensing monitoring results are compared and analyzed in detail with the ground measured data to calculate the position error and area error, providing an error benchmark for subsequent data processing.
[0059] On this basis, the overall accuracy and Kappa coefficient are calculated according to the obtained position error and area error. These two indicators will be used to evaluate the accuracy of the classification results. In order to further improve the reliability of the detection results, the classification results are cross-validated through multi-source data fusion technology to evaluate the errors between different sensors.
[0060] Finally, the error propagation model is used to analyze the influence of the error sources of the above errors on the final results. According to the analysis results, the remote sensing data is subjected to necessary error correction processing to generate the corrected final monitoring results. Through this series of refined processing, the accuracy and credibility of coastal wetland remote sensing monitoring data can be significantly improved, providing strong data support for scientific research and management work in related fields.
[0061] In some embodiments of the present application, the remote sensing images include optical images, radar images and lidar images.
[0062] It is understandable that the remote sensing images involved in this embodiment can be subdivided into multiple types, including but not limited to optical images, radar images, and laser radar images. Optical images are images obtained based on optical principles, which can reflect information such as the color, brightness, and texture of the target object. Radar images use the reflection principle of radar waves to obtain geometric information and reflection characteristics of the target object. Laser radar images use laser radar technology to accurately measure information such as the three-dimensional structure and reflectivity of the surface of the target object. These different types of remote sensing images have different advantages and characteristics in practical applications, and appropriate remote sensing images can be selected for processing and analysis according to specific needs.
[0063] In some embodiments of the present application, the remote sensing image is preprocessed, including radiation correction, geometric correction and atmospheric correction;
[0064] Wherein, the radiation correction is used to correct the radiation distortion of the remote sensing image;
[0065] The geometric correction is used to correct the geometric distortion of the remote sensing image;
[0066] The atmospheric correction is used to remove the influence of the atmosphere on the radiation to obtain the true reflectivity and / or the true radiation brightness.
[0067] It is understandable that this embodiment mainly performs a series of preprocessing operations on remote sensing images to improve the quality and usability of remote sensing data. These preprocessing operations include three main steps: radiation correction, geometric correction, and atmospheric correction.
[0068] First, radiation correction is an important part of this embodiment, and its main purpose is to correct the radiation distortion in the remote sensing image. Radiation distortion is caused by the fact that when the remote sensing sensor receives the solar radiation energy reflected or radiated by the earth's surface, it may be affected by various factors, such as the performance of the sensor itself, the effect of the earth's atmosphere, etc., which leads to a deviation between the received radiation energy and the actual radiation energy. Through radiation correction, this radiation distortion can be eliminated as much as possible, thereby improving the accuracy and reliability of the remote sensing image.
[0069] Secondly, geometric correction is the process of correcting the geometric distortion of remote sensing images. During the acquisition process, remote sensing images may be affected by factors such as the curvature of the earth, changes in sensor attitude, and platform jitter, resulting in geometric distortion of remote sensing images, such as image distortion and position offset. Through geometric correction, these geometric distortions can be effectively eliminated, making the geographic information of remote sensing images more accurate and more consistent with the distribution of actual geographic scenes.
[0070] Finally, atmospheric correction is to remove the influence of the atmosphere on the radiation information of remote sensing images. Gases, water vapor and suspended particles in the earth's atmosphere absorb and scatter solar radiation energy, thereby affecting the radiation energy received by remote sensing sensors. Through atmospheric correction, the influence of the atmosphere on remote sensing images can be eliminated as much as possible, thereby obtaining the true reflectivity and true radiation brightness information of the earth's surface, and improving the accuracy and effectiveness of remote sensing data in the application process.
[0071] In summary, this embodiment aims to improve the quality of remote sensing data through preprocessing operations such as radiation correction, geometric correction and atmospheric correction, and provide a reliable data basis for subsequent remote sensing data applications.
[0072] In some embodiments of the present application, the radiation correction includes:
[0073] Convert the original DN value observed by remote sensing equipment into surface reflectance or radiation brightness value;
[0074] Obtaining the solar altitude angle and the observation angle, and calculating the radiation correction coefficient according to the solar altitude angle and the observation angle;
[0075] The radiation correction coefficient is applied to correct the original DN value, and a radiation brightness image after radiation correction is generated.
[0076] It can be understood that the radiation correction process involved in this embodiment involves the following steps: First, the original digital value (DN value) collected by the remote sensing device is converted to obtain the reflectivity or radiation brightness value of the surface. This step is to more accurately reflect the actual physical characteristics of the surface. Next, it is necessary to obtain the solar altitude angle and the observation angle information. These two sets of data are crucial for the subsequent radiation correction process. Then, based on the obtained solar altitude angle and observation angle, the corresponding radiation correction coefficient is calculated, which can effectively correct the radiation brightness deviation caused by different observation conditions. Finally, the calculated radiation correction coefficient is applied to the original DN value to correct them. After the correction is completed, the radiation brightness image after radiation correction can be obtained, which helps to improve the accuracy and practicality of remote sensing data and provide a more reliable basis for subsequent data analysis and application. Such a radiation correction process has a wide range of application prospects in the field of remote sensing and can improve the accuracy of surface remote sensing measurements.
[0077] In some embodiments of the present application, the atmospheric correction includes:
[0078] Acquire real-time atmospheric parameters, and input the real-time atmospheric parameters into the MODTRAN model to calculate atmospheric transmittance and scattered radiation;
[0079] The atmospheric transmittance and scattered radiation are used to correct the radiation brightness image after radiation correction, and a real reflectance image after atmospheric correction is generated.
[0080] It is understandable that the atmospheric correction process involved in this embodiment includes a series of delicate and orderly steps. First, through various sensors and monitoring equipment, key parameters in the atmosphere are obtained in real time. Then, these collected atmospheric parameters are accurately input into a complex radiation transmission model called MODTRAN. The MODTRAN model is a proven software tool widely used in the field of remote sensing, which can simulate the effects of atmospheric absorption, scattering and transmittance on electromagnetic waves. Through the operation of this model, the transmittance of the atmosphere to electromagnetic waves and the intensity of scattered radiation can be obtained, and these data are crucial for subsequent image processing.
[0081] Subsequently, after obtaining the data of atmospheric transmittance and scattered radiation, the method involved in this embodiment further uses these data to correct the radiation brightness image that has been radiation corrected. This step is to correct the image distortion caused by the atmospheric influence and restore more real and accurate radiation brightness information. Through a series of image processing algorithms, the influence of atmospheric scattering can be effectively removed or reduced, and the quality of the image can be improved. Finally, after this series of correction processing, the real reflectivity image after atmospheric correction can be obtained. This image not only contains the real reflection information of the ground object, but also because the atmospheric influence is removed, it is more conducive to subsequent image analysis and application. In general, the atmospheric correction method of this embodiment aims to improve the availability and accuracy of remote sensing data through precise parameter acquisition, model operation and image processing.
[0082] In some embodiments of the present application, the atmospheric transmittance and scattered radiation are used to correct the radiation brightness image after radiation correction, and the atmospheric corrected real reflectance image is generated, which includes the following steps:
[0083] Performing attenuation correction on the radiation brightness image after radiation correction by using the atmospheric transmittance to remove the absorption and scattering effects of the atmosphere on light;
[0084] Using the scattered radiation to compensate the radiation brightness image after radiation correction, so as to correct the radiation enhancement phenomenon caused by atmospheric scattering;
[0085] At the same time, the aerosol optical thickness and the atmospheric visibility are obtained, and the radiation brightness image after the radiation correction is corrected using the aerosol optical thickness and the atmospheric visibility.
[0086] It is understandable that, in this embodiment, for the atmospheric transmittance and scattered radiation, a series of correction processes are performed on the radiation brightness image after radiation correction, in order to generate a true reflectance image after atmospheric correction. This process includes the following main steps:
[0087] First, the radiation brightness image after radiation correction is attenuated by the atmospheric transmittance. The main purpose of this step is to remove the absorption and scattering effects of the atmosphere on light. Since the gas and aerosol in the atmosphere absorb and scatter the light, this will cause distortion of the radiation brightness image. Therefore, attenuation correction of the image by atmospheric transmittance can effectively remove this distortion.
[0088] Secondly, the scattered radiation is used to compensate the radiation brightness image after radiation correction. The main purpose of this step is to correct the radiation enhancement phenomenon caused by atmospheric scattering. In the atmosphere, scattered radiation will cause the distribution of radiation brightness to change, thereby affecting the accuracy of the reflectivity image. Therefore, by compensating for scattered radiation, this radiation enhancement phenomenon can be corrected and the accuracy of the reflectivity image can be improved.
[0089] Finally, the aerosol optical depth and atmospheric visibility are obtained at the same time. These two parameters are important parameters for describing atmospheric characteristics and can be used to further correct the radiation brightness image after radiation correction. By using aerosol optical depth and atmospheric visibility, the radiation brightness image can be corrected more accurately, thereby obtaining a more accurate atmospheric correction of the true reflectivity image.
[0090] In general, through the above three steps, the radiation brightness image after radiation correction can be effectively corrected, so as to obtain an accurate real reflectance image after atmospheric correction. This is of great significance for improving the accuracy and reliability of remote sensing images.
[0091] In some embodiments of the present application, obtaining aerosol optical thickness and atmospheric visibility, and using the aerosol optical thickness and atmospheric visibility to correct the radiation brightness image after radiation correction includes:
[0092] The radiation brightness data after radiation correction is obtained according to the radiation brightness image after radiation correction, which is recorded as Lsur;
[0093] The scattered radiation intensity is estimated based on the aerosol optical thickness and atmospheric visibility. The calculation formula is:
[0094] Lsca = S × AOD × VIS -α ;
[0095] Among them, AOD is aerosol optical depth; VIS is atmospheric visibility; Lsca is scattered radiation intensity; S is the scattering coefficient; α is an empirical coefficient related to atmospheric conditions;
[0096] The calculated scattered radiation is subtracted from the radiation-corrected radiance image to remove the atmospheric scattering effect. The calculation formula is:
[0097] Lcor=Lsur-Lsca;
[0098] Wherein, Lcor is the radiant brightness obtained by correcting the radiant brightness image after radiation correction using the aerosol optical thickness and atmospheric visibility.
[0099] It is understandable that this embodiment aims to obtain the optical thickness of aerosols and the visibility of the atmosphere. The obtained optical thickness of aerosols and the visibility of the atmosphere are used to perform corresponding correction processing on the radiation brightness image after radiation correction. The specific processing process is as follows:
[0100] First, the corresponding radiance data is obtained based on the radiance image that has been calibrated. Next, the intensity of the scattered radiation is estimated based on the optical thickness of the aerosol and the visibility of the atmosphere. In this formula, AOD represents the optical thickness of the aerosol, VIS represents the visibility of the atmosphere, Lsca represents the intensity of the scattered radiation, S represents the scattering coefficient, and α represents an empirical coefficient related to the atmospheric conditions. Among them, the settings of S and α can be adjusted according to the specific geographical environment and atmospheric conditions to ensure that the estimated scattered radiation intensity is more accurate. For example, the value of S may be adjusted according to the changes in season, geographical location and aerosol type. In humid and foggy coastal areas, aerosols are mainly composed of water droplets and salt particles. At this time, the S value will be relatively high to reflect the stronger scattering effect. On the contrary, in dry and less cloudy inland areas, aerosols are mainly dust and particulate matter, and the S value may be relatively low. As for the α coefficient, as an empirical correction of atmospheric conditions, it can be determined through long-term observation data and model simulation to better adapt to the scattering characteristics under different weather conditions.
[0101] Then, the calculated scattered radiation intensity is subtracted from the radiation brightness image that has been corrected to eliminate the influence of atmospheric scattering on the image. In the formula for calculating Lcor, Lcor represents the radiation brightness after the radiation brightness image that has been corrected is corrected using the optical thickness of aerosols and the visibility of the atmosphere.
[0102] Through such processing, the radiation brightness of the ground can be obtained more accurately, thus providing more precise data support for fields such as environmental monitoring and climate change research.
[0103] In some embodiments of the present application, the geometric correction includes:
[0104] Geo-locating the atmospherically corrected true reflectance image, including establishing a correspondence between the radiance image and a geographic coordinate system using ground control base points with known geographic coordinates;
[0105] The image is resampled and reprojected through affine transformation to eliminate geometric distortion and obtain the geometrically corrected remote sensing image.
[0106] It is understandable that the geometric correction process involved in this embodiment includes a series of delicate operation steps, the purpose of which is to accurately geolocate the real reflectivity image after atmospheric correction. This step is crucial because it ensures that the image data can correspond to the actual geographical location, thereby providing a solid foundation for subsequent data analysis and application. Specifically, this process involves the use of several ground control base points with known geographical coordinates, which are used as reference points to establish an accurate correspondence between the radiance image and the geographical coordinate system. The establishment of this correspondence provides a key basis for the accurate correction of remote sensing images.
[0107] In addition, in order to further improve the availability and accuracy of remote sensing images, this embodiment also involves the operation of performing affine transformation on the images. Affine transformation is a mathematical transformation by which images can be resampled and reprojected to effectively eliminate geometric distortion caused by factors such as shooting angle and lens distortion. The result of this step is to obtain a set of geometrically corrected remote sensing images, which are more accurate in spatial position and can better meet the needs of geographic information systems (GIS), environmental monitoring, urban planning and other fields. In general, these embodiments show how to improve the accuracy and applicability of remote sensing data by geolocating and geometrically correcting the real reflectance images after atmospheric correction.
[0108] In some embodiments of the present application, a classification algorithm is used to classify the remote sensing distribution results and the ground measured data, and a preliminary classification result is generated; the classified remote sensing monitoring results are compared and analyzed with the ground measured data, and the position error and area error are calculated, including:
[0109] The sense distribution results and ground measured data are classified by random forest to generate classification results; the classification results are cross-validated using k-fold cross validation;
[0110] When the cross-validation result is accurate in classification, the classified remote sensing monitoring results are spatially compared with the ground measured data; the correctness of the classification results is analyzed by calculating the confusion matrix;
[0111] The position error is determined by calculating the spatial position difference between the classification result and the same feature in the ground measured data; the area error is calculated by comparing the classification result and the area in the ground measured data.
[0112] It can be understood that this embodiment proposes a remote sensing data processing method based on a classification algorithm. The method first uses a classification algorithm to process the remote sensing data distribution results and the corresponding ground measured data, and then generates a preliminary classification result. Next, the remote sensing monitoring results after classification processing are compared and analyzed with the ground measured data to calculate various indicators of position error and area error.
[0113] The specific operations include: first, the remote sensing data distribution results and ground-measured data are classified and processed by the random forest algorithm to generate preliminary classification results. Then, the classification results are verified by the k-fold cross-validation method to ensure the accuracy and reliability of the classification results. When the cross-validation results show that the classification is accurate, the classified remote sensing monitoring results are further compared and analyzed spatially with the ground-measured data.
[0114] In addition, the correctness of the classification results is analyzed by calculating the confusion matrix to evaluate the performance of the classification algorithm. When calculating the position error, the position error is determined by analyzing the spatial position difference between the classification result and the same feature in the ground truth data. When calculating the area error, the area error is calculated by comparing the classification result with the area in the ground truth data.
[0115] Through the above steps, this embodiment provides an effective remote sensing data processing method, which can improve the accuracy of remote sensing monitoring results and provide strong support for research and practice in related fields. At the same time, this method also provides a convenient means for comparative analysis of ground measured data and remote sensing monitoring data, which helps to better understand and use remote sensing technology.
[0116] In some embodiments of the present application, the classification results, position error and area error are incorporated into the error propagation model analysis, and error correction is performed on the remote sensing data based on the error analysis results.
[0117] It is understandable that this embodiment takes into account the classification results, position error and area error together, and incorporates these factors into the error propagation model for analysis. In this way, we can perform more accurate error correction on remote sensing data. This process is mainly based on in-depth research and understanding of the error analysis results, so as to optimize and improve the remote sensing data and improve its accuracy and reliability. In this way, remote sensing data can be processed more effectively, providing strong support for research and practice in related fields.
[0118] The present invention also proposes a coastal wetland remote sensing error detection system, which is used to implement the coastal wetland remote sensing error detection method mentioned above.
[0119] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.
[0120] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0121] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0122] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.
Claims
1. A coastal wetland remote sensing error detection method, It is characterized in that include: Acquire multi-temporal remote sensing images, pre-process the remote sensing images, and obtain the remote sensing distribution results of wetland water area and silt plain based on the pre-processed remote sensing images; at the same time, use high-precision GPS equipment to measure the actual distribution results of water body boundaries and silt plains to obtain ground measured data; A classification algorithm is used to classify the remote sensing distribution results and the ground measured data, and a preliminary classification result is generated; the classified remote sensing monitoring results are compared and analyzed with the ground measured data, and the position error and area error are calculated; at the same time, the overall accuracy and Kappa coefficient are calculated based on the position error and area error to evaluate the accuracy of the classification results; Through multi-source data fusion, the classification results are cross-validated to evaluate the errors between different sensors; Using an error propagation model to analyze the influence of the error source of the error on the final result, performing error correction on the remote sensing data according to the analysis result, and generating a corrected final monitoring result; Use classification algorithms to classify remote sensing distribution results and ground-based measured data, and generate preliminary classification results; Comparing and analyzing the classified remote sensing monitoring results with the ground measured data, and calculating the position error and area error, including: classifying the remote sensing distribution results and the ground measured data through random forest to generate classification results; The classification results were cross-validated using k-fold cross validation. When the cross-validation result indicated that the classification was accurate, the classified remote sensing monitoring results were spatially compared with the ground measured data. The correctness of the classification results was analyzed by calculating the confusion matrix. The position error was determined by calculating the spatial position difference between the classification results and the same features in the ground measured data. The area error was calculated by comparing the classification results with the regional areas in the ground measured data.
2. The coastal wetland remote sensing error detection method according to claim 1, It is characterized in that The remote sensing images include optical images, radar images and lidar images.
3. The coastal wetland remote sensing error detection method according to claim 1, It is characterized in that Preprocess remote sensing images, including radiation correction, geometric correction and atmospheric correction; Wherein, the radiation correction is used to correct the radiation distortion of the remote sensing image; The geometric correction is used to correct the geometric distortion of the remote sensing image; The atmospheric correction is used to remove the influence of the atmosphere on the radiation to obtain the true reflectivity and / or the true radiation brightness.
4. The coastal wetland remote sensing error detection method according to claim 3, It is characterized in that The radiation correction includes: Convert the original DN value observed by remote sensing equipment into surface reflectance or radiation brightness value; Obtaining the solar altitude angle and the observation angle, and calculating the radiation correction coefficient according to the solar altitude angle and the observation angle; The radiation correction coefficient is applied to correct the original DN value, and a radiation brightness image after radiation correction is generated.
5. The coastal wetland remote sensing error detection method according to claim 4, It is characterized in that The atmospheric correction includes: Acquire real-time atmospheric parameters, and input the real-time atmospheric parameters into the MODTRAN model to calculate atmospheric transmittance and scattered radiation; The atmospheric transmittance and scattered radiation are used to correct the radiation brightness image after radiation correction, and a real reflectance image after atmospheric correction is generated.
6. The coastal wetland remote sensing error detection method according to claim 5, It is characterized in that The atmospheric transmittance and scattered radiation correct the radiation brightness image after the radiation correction, and generate the atmospheric corrected real reflectance image, including: Performing attenuation correction on the radiation brightness image after radiation correction by using the atmospheric transmittance to remove the absorption and scattering effects of the atmosphere on light; Using the scattered radiation to compensate the radiation brightness image after radiation correction, so as to correct the radiation enhancement phenomenon caused by atmospheric scattering; At the same time, the aerosol optical thickness and the atmospheric visibility are obtained, and the radiation brightness image after the radiation correction is corrected using the aerosol optical thickness and the atmospheric visibility.
7. The coastal wetland remote sensing error detection method according to claim 6, It is characterized in that Obtaining aerosol optical thickness and atmospheric visibility, and using the aerosol optical thickness and atmospheric visibility to correct the radiation brightness image after radiation correction, includes: The radiation brightness data after radiation correction is obtained according to the radiation brightness image after radiation correction, which is recorded as Lsur; The scattered radiation intensity is estimated based on the aerosol optical thickness and atmospheric visibility. The calculation formula is: Lsca = S × AOD × VIS-α; Among them, AOD is aerosol optical depth; VIS is atmospheric visibility; Lsca is scattered radiation intensity; S is the scattering coefficient; α is an empirical coefficient related to atmospheric conditions; The calculated scattered radiation is subtracted from the radiation-corrected radiance image to remove the atmospheric scattering effect. The calculation formula is: Lcor=Lsur-Lsca; Wherein, Lcor is the radiant brightness obtained by correcting the radiant brightness image after radiation correction using the aerosol optical thickness and atmospheric visibility.
8. The coastal wetland remote sensing error detection method according to claim 6, It is characterized in that The geometric correction includes: Geo-locating the atmospherically corrected true reflectance image, including establishing a correspondence between the radiance image and a geographic coordinate system using ground control base points with known geographic coordinates; The image is resampled and reprojected through affine transformation to eliminate geometric distortion and obtain the geometrically corrected remote sensing image.
9. A coastal wetland remote sensing error detection system, It is characterized in that The coastal wetland remote sensing error detection system is used to implement the coastal wetland remote sensing error detection method described in any one of claims 1-8.
