Scene image positioning integrated application system and method based on Beidou high-resolution remote sensing

Abstract

The invention relates to the technical field of geographic information and remote sensing, in particular to a scene image positioning integrated application system and method based on Beidou high-resolution remote sensing. Comprising a Beidou remote sensing chip module, a space-time registration module and a scene semantic analysis module. Wherein the Beidou remote sensing chip module is used for acquiring positioning image information and sending the positioning image information to the space-time registration module and the scene semantic analysis module; the space-time registration module is used for mapping a pixel coordinate # imgabs0 # of the positioning image information to a geographic coordinate # imgabs1 # based on a Beidou space-time reference and according to a mapping calculation formula so as to form mapping image information and send the mapping image information to the scene semantic analysis module; and the scene semantic analysis module is used for generating a scene feature tag in real time according to positioning image learning and mapping image information so as to output position-image-semantic integrated information. According to the method, the pixel coordinate # imgabs2 # is mapped to the geographic coordinate # imgabs3 # through a mapping calculation formula, so that the matching precision of the image and the actual geographic coordinate is improved.

Description

Technical Field

[0001] The present invention relates to the field of geographic information and remote sensing technology, and in particular to a scene image positioning integrated application system and method based on Beidou high-resolution remote sensing. Background Art

[0002] In the field of geographic information and remote sensing technology, traditional technologies face many bottlenecks in scene image positioning and information processing. Existing geographic information systems and image acquisition systems are mostly independent of each other. When processing remote sensing images and geolocation data, there are problems such as low data fusion efficiency and poor registration accuracy. On the one hand, traditional remote sensing image acquisition equipment lacks high-precision positioning coordination functions, and the collected images are difficult to accurately match with actual geographic coordinates, resulting in large errors in the mapping of image pixel coordinates to geographic coordinates, seriously affecting subsequent image-based geographic information analysis and applications; on the other hand, existing scene semantic parsing technologies, due to the lack of accurate spatiotemporal benchmarks and high-resolution image data support, have difficulty in quickly generating accurate scene semantic labels when identifying and analyzing scene features, and cannot achieve efficient output of "location-image-semantics" integrated information.

[0003] In summary, the technical problem actually solved by the present invention is how to improve the matching accuracy between images and actual geographic coordinates. Summary of the Invention

[0004] In order to overcome the technical defect of low matching accuracy between images and actual geographic coordinates in the above-mentioned prior art, the purpose of the present invention is to provide an integrated application system and method for scene image positioning based on Beidou high-resolution remote sensing, which converts pixel coordinates into Mapping to geographic coordinates , to improve the matching accuracy between the image and the actual geographic coordinates.

[0005] The present invention discloses an integrated application system for scene image positioning based on Beidou high-resolution remote sensing, comprising a Beidou remote sensing chip module, a spatiotemporal registration module and a scene semantic analysis module, wherein the scene semantic analysis module is connected to the Beidou remote sensing chip module and the spatiotemporal registration module respectively, and the Beidou remote sensing chip module is connected to the spatiotemporal registration module; wherein,

[0006] The Beidou remote sensing chip module is used to obtain positioning image information and send the positioning image information to the spatiotemporal registration module and the scene semantic analysis module;

[0007] The spatiotemporal registration module is used to locate the pixel coordinates of the image information based on the Beidou spatiotemporal reference and the mapping calculation formula. Mapping to geographic coordinates , to form the mapping image information and send it to the scene semantic analysis module. The mapping calculation formula is:

[0008]

[0009] in, It is represented by the coordinates of the main point of the image, H is the altitude, Expressed as the sensor focal length, Expressed as pixel size, is the BeiDou space-time reference coordinate;

[0010] The mapping error V of the spatiotemporal registration module is ≤ 0.2 pixels;

[0011] The scene semantic parsing module is used to generate scene feature labels in real time based on positioning image learning and mapping image information to output integrated "position-image-semantics" information.

[0012] Preferably, the Beidou remote sensing chip module is an integrated combination of a Beidou positioning unit and a high-resolution remote sensing sensor; wherein,

[0013] The Beidou positioning unit is used to obtain positioning information;

[0014] The high-resolution remote sensing sensor is used to obtain remote sensing image information in real time when the Beidou positioning unit obtains positioning information to form positioning image information.

[0015] Preferably, the high-resolution remote sensing sensor is a visible light CMOS sensor with 12 million pixels and a resolution of 0.5mGSD@100m altitude.

[0016] Preferably, the spatiotemporal registration module locates the pixel coordinates of the image information Mapping to geographic coordinates Finally, it is also used to correct the geographic coordinates through the calculation formula Perform precision correction, the correction formula is:

[0017]

[0018] in, , , and is the correction coefficient, and its value range is ; and are the corrected geographic coordinates.

[0019] Preferably, a storage module is further included, which is respectively connected to the Beidou remote sensing chip module, the spatiotemporal registration module and the scene semantic analysis module to store the positioning image information, the mapping image information and the "position-image-semantics" integrated information as historical data;

[0020] Correction factor and Based on statistical analysis of historical data.

[0021] Preferably, when the scene semantic parsing module generates scene feature labels in real time based on the positioning image information and the mapping image information, the scene semantic parsing module generates scene feature labels based on the geographic coordinates. Calculate the spatial distribution density of scene features. The calculation formula is:

[0022]

[0023] Where N is the geographic coordinate The number of specific scene features within a circular area with a center and a radius of R, is the area of ​​the circular region. The value of R is determined according to the actual application scenario, and the relationship between R and the mapping error V is: R ≥ 10 × V.

[0024] Preferably, the scene semantic parsing module includes an evaluation unit. When the scene semantic parsing module outputs the “position-image-semantics” integrated information, the evaluation unit performs a reliability evaluation on the “position-image-semantics” integrated information based on an evaluation formula. The evaluation formula is:

[0025]

[0026] in, is the reliability score of the integrated information of “location-image-semantics”, is the reliability score of the positioning information, and Related to the mapping error V, ; It is the reliability score of remote sensing image information, determined based on the clarity and completeness of the image; The reliability score of the semantic information is determined by the accuracy of the scene semantic parsing; 、 、 is the weight coefficient, and .

[0027] In view of this, the second object of the present invention is to provide an integrated application method for scene image positioning based on Beidou high-resolution remote sensing, comprising the steps of:

[0028] Obtain positioning image information based on Beidou remote sensing chip module;

[0029] Based on the Beidou space-time reference and according to the mapping calculation formula, the pixel coordinates of the positioning image information Mapping to geographic coordinates , to form the mapping image information, the mapping calculation formula is:

[0030]

[0031] in, It is represented by the coordinates of the main point of the image, H is the altitude, Expressed as the sensor focal length, Expressed as pixel size, is the BeiDou space-time reference coordinate;

[0032] Based on the positioning image learning and mapping image information, scene feature labels are generated in real time to output integrated "position-image-semantics" information.

[0033] Preferably, the pixel coordinates of the positioning image information are calculated based on the Beidou space-time reference and the mapping formula. Mapping to geographic coordinates , the step of forming the mapping image information further includes the following steps:

[0034] Correction calculation of geographic coordinates Perform precision correction, the correction formula is:

[0035]

[0036] in, , , and is the correction coefficient, and its value range is ; and are the corrected geographic coordinates.

[0037] Preferably, before the step of acquiring positioning image information based on the Beidou remote sensing chip module, the following steps are also included:

[0038] Initialize the Beidou remote sensing chip module.

[0039] After adopting the above technical solution, compared with the prior art, the beneficial effect of the present invention is that the pixel coordinates are converted into Mapping to geographic coordinates , in order to improve the matching accuracy between the image and the actual geographic coordinates; the geographic coordinates are corrected by the calculation formula Precision correction is performed to further reduce errors and further improve matching accuracy. The organically combined output of positioning, image, and semantic information changes the information separation found in traditional technologies, providing users with a more comprehensive and intuitive information display, facilitating quick understanding of the scene. The reliability of the integrated "position-image-semantic" information is assessed through evaluation formulas, providing users with a reference basis and preventing decisions made due to unreliable information. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1This is a schematic diagram of an integrated application system for scene image positioning based on Beidou high-resolution remote sensing according to the present invention;

[0041] Figure 2 The present invention is a flowchart of an integrated application method for scene image positioning based on Beidou high-resolution remote sensing. DETAILED DESCRIPTION

[0042] The advantages of the present invention are further described below with reference to the accompanying drawings and specific embodiments.

[0043] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.

[0044] The terms used in this disclosure are for the purpose of describing specific embodiments only and are not intended to limit the disclosure. As used in this disclosure and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0045] It should be understood that although the terms first, second, third, etc. may be used in this disclosure to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining."

[0046] In the description of the present invention, it should be understood that the terms "longitudinal", "transverse", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are 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 understood as limiting the present invention.

[0047] In the description of the present invention, unless otherwise specified and limited, it should be noted that the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a mechanical connection or an electrical connection, or it can be the internal communication between two components. It can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to the specific circumstances.

[0048] In the following description, the suffixes such as "module", "component" or "unit" used to represent elements are only used to facilitate the description of the present invention and have no specific meaning. Therefore, "module" and "component" can be used interchangeably.

[0049] This embodiment discloses an integrated application system for scene image positioning based on Beidou high-resolution remote sensing, including a Beidou remote sensing chip module, a spatiotemporal registration module and a scene semantic analysis module. The scene semantic analysis module is connected to the Beidou remote sensing chip module and the spatiotemporal registration module respectively, and the Beidou remote sensing chip module is connected to the spatiotemporal registration module; wherein the Beidou remote sensing chip module is used to obtain positioning image information and send the positioning image information to the spatiotemporal registration module and the scene semantic analysis module; the spatiotemporal registration module is used to calculate the pixel coordinates of the positioning image information based on the Beidou spatiotemporal reference and according to the mapping calculation formula. Mapping to geographic coordinates , to form the mapping image information and send it to the scene semantic analysis module. The mapping calculation formula is: ,in, It is represented by the coordinates of the main point of the image, H is the altitude, Expressed as the sensor focal length, Expressed as pixel size, is the Beidou space-time reference coordinate; the mapping error V of the space-time registration module is ≤ 0.2 pixels; the scene semantic parsing module is used to generate scene feature labels in real time based on positioning image learning and mapping image information to output "position-image-semantics" integrated information.

[0050] See Figure 1 As shown, this embodiment will provide a detailed description of an integrated application system for scene image positioning based on Beidou high-resolution remote sensing. The system is mainly composed of three core components, namely Beidou remote sensing chip module, spatiotemporal registration module and scene semantic parsing module. The Beidou remote sensing chip module is respectively connected to the spatiotemporal registration module and the scene semantic parsing module, and the spatiotemporal registration module is connected to the scene semantic parsing module.

[0051] Among them, the Beidou remote sensing chip module integrates the Beidou positioning unit and the high-resolution remote sensing sensor. The Beidou positioning unit is used to obtain positioning information, which includes latitude and longitude, altitude, and device attitude data (pitch angle). , roll angle , yaw angle Ψ), etc. High-resolution remote sensing sensors are used to collect high-resolution remote sensing imagery, clearly capturing every detail of a scene. The Beidou positioning unit and high-resolution remote sensing sensors operate synchronously, ensuring that the Beidou positioning unit acquires positioning information while the high-resolution remote sensing sensor simultaneously acquires remote sensing imagery. This synchronized data acquisition provides an accurate and time-aligned data foundation for subsequent processing.

[0052] The space-time registration module works based on the BeiDou space-time reference, which provides accurate time and space reference. Position the pixel coordinates on the image information Map converted to actual geographic coordinates In the mapping calculation formula, Expressed as the coordinates of the principal point of the image, it is the central reference point of the image coordinate system, which is determined by the characteristics of the sensor and is used to determine the pixel coordinates H represents the altitude, which affects the vertical mapping relationship between image pixels and geographic coordinates (X, Y). Expressed as the sensor focal length, it affects the imaging ratio and geometric relationship of the image. Sensors with different focal lengths will have different sizes and shapes of images of the same object. Expressed as pixel size, it determines the physical size represented by each pixel in the actual scene. It is the Beidou space-time reference coordinate, which is the geographic coordinate The starting reference point of the conversion is determined by the high-precision positioning information obtained by the Beidou positioning unit.

[0053] It should be noted that the image principal point coordinates will be determined for different types of sensors. For example, for visible light CMOS sensors, the image principal point coordinates are usually located at the center of the image, and the exact value can be obtained from the sensor specification sheet. If the sensor has manufacturing errors or installation deviations, specific calibration methods can be used to correct them, such as using a known standard pattern to shoot and using image processing algorithms to calculate the image principal point coordinates. , ensuring accuracy.

[0054] In addition, the sensor focal length (fx, fy) is determined by the manufacturing process during the production process and will be clearly marked when the product leaves the factory. In actual use, the sensor focal length (fx, fy) may change slightly due to factors such as ambient temperature and humidity, and regular calibration is required to maintain imaging accuracy. For example, a standard object of known size is used for shooting, and the accurate focal length value is calculated based on the ratio between the imaged size and the actual size, combined with parameters such as shooting distance. For pixel size , which has been determined by precise measurement at the factory and is clearly marked in the sensor specification. If a deviation in pixel size is found in actual use, the pixel size can be calculated by physical measurement of the sensor imaging area or by testing with a standard image of known resolution and the correspondence between the image pixels and the actual size. .

[0055] The mapping error is required to be ≤ 0.2 pixels, which is a key indicator to ensure the accuracy of the system. This error limit ensures that the image pixels are consistent with the actual geographic coordinates. The correspondence between them must be sufficiently accurate. For example, in urban planning applications, high-precision mapping enables planners to accurately match elements such as buildings and roads in positioning image information with their actual geographic locations, providing a reliable basis for urban layout planning. However, if the mapping error is too large, it may lead to misjudgment of the location of urban infrastructure, affecting the scientific and rational nature of the planning.

[0056] The scene semantic parsing module performs in-depth analysis of the positioning image information collected by the Beidou remote sensing chip module and the mapping image information generated by the spatiotemporal registration module. By building a knowledge base mapping "scene features, image information, and geographic elements," and utilizing a lightweight deep learning model, it identifies various landform types (such as buildings, vegetation, and water bodies) in the image in real time, calculates vegetation cover, and outlines buildings. This information is then integrated to generate scene feature labels. The final output is integrated information encompassing location, image, and semantics, providing users with a comprehensive and intuitive description of the scene to meet the needs of various application scenarios. For example, in emergency rescue scenarios, rescuers can use this integrated "location-image-semantics" information to quickly understand the terrain, building distribution, and population density of the affected area, thereby formulating more effective rescue plans.

[0057] It should be noted that the BeiDou time and space reference is the core element of the BeiDou satellite, providing precise time and space reference standards for various positioning, navigation, and timing applications. It mainly covers two aspects: time reference and space reference. It is used to determine the position of each pixel in the image on the two-dimensional image plane, and is used for positioning image information processing and geographic coordinates. The basis of mapping.

[0058] Furthermore, the Beidou remote sensing chip module is an integrated combination of a Beidou positioning unit and a high-resolution remote sensing sensor; wherein, the Beidou positioning unit is used to obtain positioning information; the high-resolution remote sensing sensor is used to obtain remote sensing image information in real time when the Beidou positioning unit obtains positioning information, so as to form positioning image information.

[0059] In this embodiment, the Beidou remote sensing chip module will be described as follows. The Beidou remote sensing chip is integrated with a Beidou positioning unit and a high-resolution remote sensing sensor. The Beidou positioning unit is used to obtain positioning information, and the high-resolution remote sensing sensor is used to obtain remote sensing image information in real time when the Beidou positioning unit obtains positioning information, thereby forming positioning image information.

[0060] It should be noted that the Beidou remote sensing chip module is an embeddable structure with dimensions ≤30mm × 20mm × 5mm, allowing it to be embedded in any carrier, such as eyeglass frames, backpacks, and mobile phones, with power consumption ≤50mW. In this embodiment, the Beidou remote sensing chip module is designed as an embeddable structure, enabling it to be easily integrated into a variety of carrier devices, expanding the scope of application of the present invention. Its small size and low power consumption make it suitable for devices with strict requirements on size and power consumption, such as wearable devices and handheld devices.

[0061] It should be noted that the BeiDou positioning unit integrates the BeiDou-3 high-precision positioning module to support BDS / GSP / GLONASS multi-system integration, achieving static positioning accuracy of ≤2cm and dynamic positioning accuracy of ≤5cm, thereby improving positioning accuracy and reliability. Multi-system integration ensures the system can stably acquire high-precision positioning data in various environments and application scenarios. For example, in precision agriculture, when agricultural machinery operates in the field, a single positioning system may be subject to interference due to complex environments. By integrating BDS, GPS, and GLONASS, the system's BeiDou positioning unit ensures that agricultural machinery maintains a dynamic positioning accuracy of ≤5cm, even in conditions such as crop obstruction, enabling precise sowing, fertilization, and other operations.

[0062] Furthermore, the high-resolution remote sensing sensor is a visible light CMOS sensor with 12 million pixels and a resolution of 0.5mGSD@100m altitude.

[0063] In this embodiment, a high-resolution remote sensing sensor is described in detail. The high-resolution remote sensing sensor is a visible light CMOS sensor with 12 million pixels and a resolution of 0.5mGSD@100m. It should be noted that the visible light CMOS sensor has high pixels and high resolution, and can collect clear and detailed remote sensing image information, providing a high-quality data foundation for subsequent image analysis and scene understanding. For example, during power line inspections, a high-resolution remote sensing sensor can clearly capture the details of the power line at a height of 100m, with a resolution of 0.5m. It can detect minor damage, aging, and other problems on the line. The 12 million pixels ensure image clarity, making it easier for staff to accurately judge the line condition. It should also be noted that the visible light CMOS sensor is an image sensor whose operating principle is based on the photoelectric effect. Through a series of processes, light signals are converted into electrical signals and ultimately form digital images.

[0064] It should be noted that the Beidou remote sensing chip module also integrates an edge computing unit, which can process the collected positioning image information in real time. This edge computing unit is an ARM Cortex-M4F chip with a computing power of 120DMIPS and a power consumption of ≤30mW.

[0065] Furthermore, the spatiotemporal registration module locates the pixel coordinates of the image information Mapping to geographic coordinates Finally, it is also used to correct the geographic coordinates through the calculation formula Perform precision correction, the correction formula is: ,in, , , and is the correction coefficient, and its value range is ; and are the corrected geographic coordinates.

[0066] In this embodiment, the spatiotemporal registration module will be described in detail again. In the above embodiment, the spatiotemporal registration module locates the pixel coordinates of the image information. Mapping to geographic coordinates After that, the geographic coordinates will be aligned with the spatiotemporal registration module. Optimize to reduce mapping errors. To geographic coordinates During the mapping process, the mapped geographic coordinates are affected by sensor parameter deviations, environmental factors, etc. There is a certain error, so the correction formula Adjust the coordinates, where , , and is the correction coefficient, and its value range is ; and are the corrected geographic coordinates.

[0067] It also includes a storage module, which is connected to the Beidou remote sensing chip module, the spatiotemporal registration module and the scene semantic analysis module to store the positioning image information, the mapping image information and the "position-image-semantics" integrated information as historical data; the correction coefficient and Based on statistical analysis of historical data.

[0068] In this embodiment, the system will be described in detail again. The system also includes a storage module, which is connected to the Beidou remote sensing chip module, the spatiotemporal registration module and the scene semantic analysis module respectively to store the positioning image information, the mapping image information and the "position-image-semantics" integrated information as historical data, and the correction coefficient and This is derived from statistical analysis of historical data.

[0069] It should be noted that the correction system and The acquisition process is to filter out historical data related to coordinate mapping from the positioning image information, mapping image information and "position-image-semantic" integrated information stored in the system in the past. The filtered historical data covers pixel coordinates at different times, different geographical locations and different environmental conditions. Mapping to geographic coordinates As well as BeiDou space-time reference coordinates. For example, historical data obtained from multiple remote sensing monitoring of the same urban area in different seasons and weather conditions, including the pixel coordinates at each monitoring time , geographic coordinates And accurate Beidou space-time reference coordinates. Then, the collected historical data is deeply analyzed to calculate the error of each mapping and identify the factors that cause the error and the underlying relationships and patterns. For example, statistical methods can be used to analyze the distribution characteristics and changing trends of errors under different terrains (mountains, plains) and different lighting conditions. Finally, based on the error patterns obtained from the analysis, the correction coefficient is determined through algorithms such as least squares method, gradient descent method, and Kalman filter algorithm. and .

[0070] For example, in urban topographic mapping, the geographic coordinates of a building are , but there may be a certain deviation, through the correction formula, the correction coefficient is determined according to historical data 、 , for geographic coordinates Correction to obtain more accurate , making the surveying and mapping results more accurate.

[0071] It should be noted that in this embodiment, the specific rules for filtering the positioning image information, mapping image information and "location-image-semantic" integrated information stored in the past of the system will be described in detail. The filtering can be performed according to multiple dimensions such as time, geographical location, and environmental conditions. For example, data within the past year or a specific time period are selected in chronological order. For geographical location, data from specific areas or different types of landforms can be selected according to the needs of the application scenario. In terms of environmental conditions, data under different weather conditions (sunny, cloudy, rainy, etc.) and different lighting intensities are filtered out. At the same time, it is necessary to ensure that the filtered historical data is representative and can cover various factors that may affect coordinate mapping.

[0072] In some embodiments, taking the least squares method as an example, it is described in detail how to use the filtered historical data to determine the correction coefficient. and . The pixel coordinates in the historical data Substitute the corresponding geographic coordinates (X, Y) into the mapping calculation formula to calculate the preliminary mapping result. Then, compare the preliminary mapping result with the Beidou space-time reference coordinates to calculate the coordinate deviation. and . Using these deviation values ​​as observation data, construct the correction coefficient and The error function is obtained by minimizing the error function to obtain the correction coefficient.

[0073] Furthermore, the scene semantic parsing module generates scene feature labels in real time based on the positioning image information and the mapping image information. Calculate the spatial distribution density of scene features. The calculation formula is: , where N is the geographic coordinate The number of specific scene features within a circular area with a center and a radius of R, is the area of ​​the circular region. The value of R is determined according to the actual application scenario, and the relationship between R and the mapping error V is: R ≥ 10 × V.

[0074] In this embodiment, the scene semantic parsing module will be described in detail again. When the scene semantic parsing module generates scene feature labels in real time based on the positioning image information and the mapping image information, the scene feature labels are generated based on the geographic coordinates. Calculate the spatial distribution density of scene features. The calculation formula for calculating the spatial distribution density is: .

[0075] The N in the calculation formula for calculating spatial distribution density is the geographic coordinates The number of specific scene features within a circular area with a center and a radius of R, The radius R is the area of ​​the circular region. The value of R is determined based on the actual application scenario, and the relationship between R and the mapping error V is: R ≥ 10 × V. In this embodiment, by calculating the spatial distribution density of specific scene features within a certain area, it is possible to more intuitively understand the concentration or dispersion of scene features in geographic space. The value of the radius R is related to the mapping error V in the aforementioned embodiment, ensuring that the calculation area covers scene features within a certain range.

[0076] For example, in agricultural land use monitoring, a certain geographical coordinate As the center, set the radius R = 20m (assuming the mapping error V = 2m), and count the number of farmlands N = 5 in the circular area. Then the spatial distribution density of the space in the area is , the distribution of farmland in the area can be evaluated.

[0077] Furthermore, the scene semantic parsing module includes an evaluation unit. When the scene semantic parsing module outputs the “position-image-semantics” integrated information, the evaluation unit performs a reliability evaluation on the “position-image-semantics” integrated information based on an evaluation formula. The evaluation formula is: ,in, is the reliability score of the integrated information of “location-image-semantics”, is the reliability score of the positioning information, and Related to the mapping error V, ; It is the reliability score of remote sensing image information, determined based on the clarity and completeness of the image; The reliability score of the semantic information is determined by the accuracy of the scene semantic parsing; 、 、 is the weight coefficient, and .

[0078] In this embodiment, the scene semantic parsing module will be described in detail again. The scene semantic parsing module also includes an evaluation unit. When the scene semantic parsing module outputs the "position-image-semantics" integrated information, the evaluation calculation formula in the evaluation unit is used to perform a reliability evaluation on the "position-image-semantics" integrated information. The evaluation calculation formula is: ,in, is the reliability score of the integrated information of “location-image-semantics”, is the reliability score of the positioning information, and Related to the mapping error V, ; It is the reliability score of remote sensing image information, determined based on the clarity and completeness of the image; The reliability score of the semantic information is determined by the accuracy of the scene semantic parsing; 、 、 is the weight coefficient, and .

[0079] It should be noted that The calculation of is related to the calculation error V in the above embodiment, specifically: , where V is the mapping error, mapping error Calculated, it reflects the geographical coordinates The accuracy of the mapping, when the mapping error V is 0, Reaching the maximum value of 1 means that the positioning information is completely reliable; as the mapping error V increases, Will gradually decrease, when the mapping error is 0.2 pixels, 0, indicating that the reliability of the positioning information is extremely low. For example, if a pixel coordinate Mapping to geographic coordinates The mapping error V is 0.1 and we can get , that is, the reliability score of this positioning information is 0.5.

[0080] It should be noted that It is determined based on factors such as image clarity and integrity. Clarity can be measured by indicators such as image resolution, edge sharpness, and detail richness. If a high-resolution image can clearly present the details of the ground objects, its clarity is high. Approaches 1; if the image is blurred and the texture is lost, then In terms of completeness, there is no missing or occluded complete image. High, there are large areas of data loss or blocked images, When conducting remote sensing monitoring of a city, some areas of the remote sensing image information obtained are blocked by clouds, affecting the observation of buildings. At this time, a comprehensive assessment of the image clarity and integrity may be made. Set to 0.7.

[0081] It should be noted that According to the accuracy setting of scene semantic analysis, the scene semantic analysis module builds a "scene feature-image learning-geographic element" mapping knowledge base, uses a lightweight deep learning model to identify information such as land feature type, vegetation coverage, building outline, and outputs scene semantic labels. If the analysis result accurately reflects the actual scene, such as correctly identifying an area as "commercial area-dense high-rise buildings-busy traffic", and is consistent with the actual situation, the accuracy is high. If there are many misjudgments, such as misjudging farmland as construction land, If 100 land parcels are identified for land feature type, and 90 are accurately identified, the accuracy rate is 90%. Set to 0.9.

[0082] It should be noted that In the calculation formula 、 、 is the weight coefficient, and , whose value is determined according to the actual application scenario. This formula combines the reliability scores of location, image, and semantics through weighted summation to obtain the reliability score of the integrated information of "location-image-semantics". For example, in the scenario of urban security monitoring, more attention will be paid to the accuracy of positioning information, and the weight can be set. =0.5, 、 , if the , , =0.6, then =0.5×0.8+0.3×0.7+0.2×0.6=0.73. In this embodiment, by Through the calculation formula, users can intuitively understand the reliability of the integrated "location-image-semantics" information and decide whether to take action based on the information.

[0083] In this embodiment, a lightweight deep learning model is described in detail. For example, a lightweight deep learning model can employ a convolutional neural network (CNN) architecture. For example, the lightweight deep learning model can consist of three convolutional layers, two pooling layers, and two fully connected layers. Each convolutional layer has a kernel size of 3×3 and a stride of 1, and so on. The convolutional layers extract local features of the image, the pooling layers reduce the data dimension while retaining key features, and the fully connected layers integrate the extracted features and output the final scene semantic label.

[0084] It should be noted that the training data for the lightweight deep learning model comes from multiple sources, including but not limited to public remote sensing image datasets and field-collected image data. The collected training data undergoes preprocessing, including image cropping, normalization, and annotation. Image cropping ensures consistent image dimensions for input to the lightweight deep learning model; normalization maps image pixel values ​​to a specific range, improving model training efficiency and stability; and annotation labels each image with true semantic information.

[0085] This embodiment also provides an integrated application method for scene image positioning based on Beidou high-resolution remote sensing, including the steps of: obtaining positioning image information based on the Beidou remote sensing chip module; calculating the pixel coordinates of the positioning image information based on the Beidou time-space reference and according to the mapping calculation formula; Mapping to geographic coordinates , to form the mapping image information, the mapping calculation formula is: ,in, It is represented by the coordinates of the main point of the image, H is the altitude, Expressed as the sensor focal length, Expressed as pixel size, It is the Beidou space-time reference coordinate; it generates scene feature labels in real time based on positioning image learning and mapping image information to output "position-image-semantic" integrated information.

[0086] In this embodiment, see Figure 2 As shown in the figure, a scene image positioning integrated application method based on Beidou high-resolution remote sensing is described in detail, which specifically includes the following steps:

[0087] Step S100: Acquire positioning image information based on the Beidou remote sensing chip module, wherein the Beidou remote sensing chip module includes a Beidou positioning unit and a high-resolution remote sensing sensor. The Beidou remote sensing chip module acquires positioning information based on the Beidou positioning unit and acquires remote sensing image information based on the high-resolution remote sensing sensor, so as to acquire positioning image information based on the positioning information and remote sensing image information.

[0088] Step S200: Calculate the pixel coordinates of the positioning image information obtained in step S100 based on the Beidou space-time reference and the mapping calculation formula. Mapping to geographic coordinates , thereby forming the mapping image information. The mapping calculation formula in this step S200 is ,in, It is represented by the coordinates of the main point of the image, H is the altitude, Expressed as the sensor focal length, Expressed as pixel size, It is the Beidou space-time reference coordinate.

[0089] Step S300: Generate scene feature labels in real time based on the positioning image information obtained in step S100 and the mapping image information obtained in step S200 to output “position-image-semantics” integrated information.

[0090] Furthermore, based on the Beidou space-time reference and according to the mapping calculation formula, the pixel coordinates of the positioning image information are Mapping to geographic coordinates The step of forming the mapping image information further includes the following steps: correcting the geographic coordinates by the calculation formula Perform precision correction, the correction formula is: ,in, , , and is the correction coefficient, and its value range is ; and are the corrected geographic coordinates.

[0091] In this example, the above step S200 will be described in detail. The above step S200 also includes the following steps:

[0092] Step S210: By calculating the geographic coordinates Perform precision correction. The precision correction calculation formula in this step S210 is: ,in, , , and is the correction coefficient, and its value range is ; and is the corrected geographic coordinate.

[0093] Furthermore, before the step of acquiring positioning image information based on the Beidou remote sensing chip module, the method further includes the step of initializing the Beidou remote sensing chip module.

[0094] In this embodiment, the method for integrated scene image positioning based on Beidou high-resolution remote sensing includes the following steps before step S100:

[0095] Step S400: Initialize the Beidou remote sensing chip module to ensure that all parameters and functional states of the Beidou remote sensing chip module are in normal state before starting to work.

[0096] It should be noted that the embodiments of the present invention have better practicability and do not impose any form of limitation on the present invention. Any technician familiar with the field may use the technical content disclosed above to change or modify it into an equivalent effective embodiment. However, any modification or equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A scene image positioning integrated application system based on Beidou high-resolution remote sensing, characterized in that: It includes a Beidou remote sensing chip module, a spatiotemporal registration module and a scene semantic analysis module, wherein the scene semantic analysis module is respectively connected to the Beidou remote sensing chip module and the spatiotemporal registration module, and the Beidou remote sensing chip module is connected to the spatiotemporal registration module; wherein, The Beidou remote sensing chip module is used to obtain positioning image information and send the positioning image information to the spatiotemporal registration module and the scene semantic analysis module; The spatiotemporal registration module is used to register the pixel coordinates of the positioning image information based on the Beidou spatiotemporal reference and according to the mapping calculation formula. Mapping to geographic coordinates , to form mapping image information and send it to the scene semantic analysis module. The mapping calculation formula is: in, It is represented by the coordinates of the main point of the image, H is the altitude, Expressed as the sensor focal length, Expressed as pixel size, is the BeiDou space-time reference coordinate; The mapping error V of the spatiotemporal registration module is ≤ 0.2 pixels; The scene semantic parsing module is used to generate scene feature labels in real time based on the positioning image learning and the mapping image information to output "position-image-semantics" integrated information.

2. The integrated application system for scene image positioning based on Beidou high-resolution remote sensing according to claim 1 is characterized in that: The Beidou remote sensing chip module is an integrated combination of a Beidou positioning unit and a high-resolution remote sensing sensor; wherein, The Beidou positioning unit is used to obtain positioning information; The high-resolution remote sensing sensor is used to obtain remote sensing image information in real time when the Beidou positioning unit obtains the positioning information, so as to form the positioning image information.

3. The integrated application system for scene image positioning based on Beidou high-resolution remote sensing according to claim 2 is characterized in that: The high-resolution remote sensing sensor is a visible light CMOS sensor with 12 million pixels and a resolution of 0.5mGSD@100m altitude.

4. The integrated application system for scene image positioning based on Beidou high-resolution remote sensing according to claim 1 is characterized in that: The spatiotemporal registration module calculates the pixel coordinates of the positioning image information Mapping to geographic coordinates Finally, it is also used to correct the geographic coordinates by the calculation formula Perform precision correction, the correction calculation formula is: in, , , and is the correction coefficient, the value range is 0.01~0.1, and is the corrected geographic coordinate.

5. The integrated application system for scene image positioning based on Beidou high-resolution remote sensing according to claim 4 is characterized in that: It also includes a storage module, which is respectively connected to the Beidou remote sensing chip module, the spatiotemporal registration module, and the scene semantic analysis module to store the positioning image information, the mapping image information, and the "position-image-semantics" integrated information as historical data; The correction factor and Derived from the statistical analysis of the historical data.

6. The integrated application system for scene image positioning based on Beidou high-resolution remote sensing according to claim 1 is characterized in that: When the scene semantic parsing module generates scene feature labels in real time according to the positioning image information and the mapping image information, the scene feature labels are generated according to the geographic coordinates. Calculate the spatial distribution density of scene features. The calculation formula is: Where N is the geographic coordinate The number of specific scene features within a circular area with a center and a radius of R, is the area of ​​the circular region, the value of R is determined according to the actual application scenario, and the relationship between R and the mapping error V is: R≥10×V.

7. The integrated application system for scene image positioning based on Beidou high-resolution remote sensing according to claim 2 is characterized in that: The scene semantic analysis module includes an evaluation unit. When the scene semantic analysis module outputs the "position-image-semantics" integrated information, the evaluation unit performs a reliability evaluation on the "position-image-semantics" integrated information based on an evaluation calculation formula. The evaluation calculation formula is: in, is the reliability score of the "position-image-semantics" integrated information, is the reliability score of the positioning information, and Related to the mapping error V, ; The reliability score of the remote sensing image information is determined based on the clarity and completeness of the image; The reliability score of the semantic information is determined based on the accuracy of the scene semantic parsing; 、 、 is the weight coefficient, and .

8. A scene image positioning integrated application method based on Beidou high-resolution remote sensing, characterized in that: Including steps: Obtain positioning image information based on Beidou remote sensing chip module; Based on the Beidou space-time reference and according to the mapping calculation formula, the pixel coordinates of the positioning image information are Mapping to geographic coordinates , to form the mapping image information, the mapping calculation formula is: in, It is represented by the coordinates of the main point of the image, H is the altitude, Expressed as the sensor focal length, Expressed as pixel size, is the BeiDou space-time reference coordinate; Scene feature labels are generated in real time based on the positioning image information and the mapping image information to output "position-image-semantics" integrated information.

9. The integrated application method for scene image positioning based on Beidou high-resolution remote sensing according to claim 8, characterized in that: Based on the Beidou space-time reference and according to the mapping calculation formula, the pixel coordinates of the positioning image information are Mapping to geographic coordinates , the step of forming the mapping image information further includes the following steps: The geographic coordinates are corrected by the calculation formula Perform precision correction, the correction calculation formula is: in, , , and is the correction coefficient, and its value range is ; and is the corrected geographic coordinate.

10. The integrated application method for scene image positioning based on Beidou high-resolution remote sensing according to claim 8, characterized in that: Before the step of obtaining positioning image information based on the Beidou remote sensing chip module, the following steps are also included: Initialize and set the Beidou remote sensing chip module.

Images