AR Digital Telescope Azimuth Angle Generation Method Based on Spatial Positioning Analysis

Abstract

The present invention discloses a method for generating the azimuth angle of an AR digital telescope based on spatial positioning analysis, which relates to the field of augmented reality. The method includes: obtaining the first position and the first attitude information of the AR digital telescope through a spatial positioning system; collecting a target image through an image acquisition device; performing preprocessing, extracting image features, and obtaining a target feature descriptor; obtaining a panoramic view of the target scene, performing feature extraction, and obtaining a panoramic feature descriptor; performing feature matching to obtain matching points; determining the second position of the matching points according to the position of the matching points in the earth coordinate system of the panoramic view; and fusing the first position, the first attitude information, and the second position to calculate and obtain the azimuth angle of the target image. It solves the technical problem of low calculation accuracy existing in the existing method for generating the azimuth angle of an AR digital telescope, and achieves the technical effect of improving the accuracy of azimuth angle calculation.

Description

Technical Field

[0001] This application relates to the field of augmented reality, and particularly to a method for generating the azimuth angle of an AR digital telescope based on spatial positioning analysis. Background Art

[0002] In the rapid development of augmented reality (AR) technology, the AR digital telescope, as an innovative interactive device, provides users with an unprecedented immersive experience. Such devices not only combine the long-distance observation function of traditional telescopes but also integrate AR technology, enabling users to obtain rich digital information while observing. However, in AR applications, a key technical challenge is how to accurately correspond the target image observed by the user through the AR digital telescope to the actual position in the real world, that is, to determine the precise azimuth of the target image in the real world. Currently, in augmented reality applications, the method for determining the azimuth angle of the target image mainly relies on a single image recognition technology. This method finds the most similar matching item by comparing the target image with the images in the pre-stored image library and infers the approximate position and direction of the target image accordingly. However, the single image recognition technology is easily affected by factors such as lighting conditions, viewing angle changes, and image quality, resulting in inaccurate matching results and thus affecting the calculation accuracy of the azimuth angle.

[0003] In the related technologies at the present stage, there is a technical problem that the calculation accuracy of the AR digital telescope azimuth angle generation method is not high. Summary of the Invention

[0004] This application provides a method for generating the azimuth angle of an AR digital telescope based on spatial positioning analysis. By using a high-precision spatial positioning system to obtain the precise position and attitude information of the AR digital telescope in the earth coordinate system in real time, capturing the target image through the image acquisition device of the AR digital telescope and performing preprocessing, extracting features from the preprocessed target image to obtain the target feature descriptor. At the same time, obtaining the panoramic view of the target scene and performing feature extraction to obtain the panoramic feature descriptor, finding the matching points between the two through the feature matching algorithm, and based on the position of the matching points in the earth coordinate system of the panoramic view, combining the position and attitude information of the AR digital telescope, and finally obtaining the precise azimuth angle of the target image through spatial geometric calculation and other technical means, achieving the technical effect of improving the accuracy of azimuth angle calculation.

[0005] This application provides a method for generating the azimuth angle of an AR digital telescope based on spatial positioning analysis, including:

[0006] Obtain the first position and the first attitude information of the AR digital telescope in the Earth coordinate system through a spatial positioning system; collect a target image through an image acquisition device of the AR digital telescope; preprocess the target image and extract image features to obtain a target feature descriptor; obtain a panoramic view of the target scene and perform feature extraction on the panoramic view to obtain a panoramic feature descriptor; perform feature matching on the target feature descriptor and the panoramic feature descriptor to obtain matching points; determine the second position of the matching points in the Earth coordinate system according to the position of the matching points in the Earth coordinate system of the panoramic view; fuse the first position, the first attitude information and the second position, and calculate to obtain the azimuth angle of the target image.

[0007] In a possible implementation, the following processing is performed:

[0008] The first attitude information includes the pitch angle, yaw angle and roll angle of the AR digital telescope.

[0009] In a possible implementation, for the step of fusing the first position, the first attitude information and the second position and calculating to obtain the azimuth angle of the target image, the following processing is performed:

[0010] Determine the basic direction vector between the first position and the second position; convert the pitch angle, yaw angle and roll angle in the first attitude information into a rotation matrix; fuse the basic direction vector with the rotation matrix to obtain a corrected direction vector; use spherical trigonometry to calculate the angle of the corrected direction vector relative to the geographic north to obtain the azimuth angle of the target image.

[0011] In a possible implementation, the following processing is performed:

[0012] The formula for calculating the azimuth angle of the target image is as follows: α' = arctan2(V y ”, V x ”) + Δα dynamic ; where α' is the azimuth angle of the target image, V y ” and V x ” are the x and y components of the corrected direction vector V” in the Cartesian coordinate system respectively, and Δα dynamic is the dynamic error correction factor of the sensor; the formula for calculating the corrected direction vector V” is as follows: V” = R × V', where R is the rotation matrix and V' is the corrected direction vector; the formula for calculating the rotation matrix R is as follows: Where, is the rotation matrix of the yaw angle around the Z axis, is the rotation matrix of the roll angle around the Y axis, is the pitch angle The rotation matrix about the X-axis, The rotation matrix of is as follows:

[0013] The calculation formula of the corrected direction vector V' is as follows: where V is the base direction vector, ΔR is the ellipsoidal radius error correction amount based on the WGS84 ellipsoid model, is the normal vector of the earth's surface, Δh is the height difference between the first position and the second position, is the unit vector in the height direction; the calculation formula of the base direction vector V is as follows: V = (x2 - x1, y2 - y1, z2 - z1), where x1, y1, z1 are the Cartesian coordinates of the first position, and x2, y2, z2 are the Cartesian coordinates of the second position.

[0014] In a possible implementation, preprocess the target image, extract image features, obtain a target feature descriptor, and perform the following processing:

[0015] Perform noise removal, contrast enhancement, edge enhancement, and scale normalization preprocessing on the target image to obtain a standard target image; perform adaptive feature extraction on the standard target image to obtain the target feature descriptor.

[0016] In a possible implementation, perform adaptive feature extraction on the standard target image to obtain the target feature descriptor, and perform the following processing:

[0017] Based on the LBP algorithm, SIFT algorithm, Canny algorithm, and RefineNet model respectively, construct a texture feature extraction channel, a corner feature extraction channel, an edge feature extraction channel, and a high-level semantic feature extraction channel; combine the texture feature extraction channel with any one of the corner feature extraction channel, edge feature extraction channel, and high-level semantic feature extraction channel to construct a first feature extraction channel, combine the texture feature extraction channel with any two of the corner feature extraction channel, edge feature extraction channel, and high-level semantic feature extraction channel to construct a second feature extraction channel, and combine the texture feature extraction channel with the corner feature extraction channel, edge feature extraction channel, and high-level semantic feature extraction channel to construct a third feature extraction channel; connect the first feature extraction channel, the second feature extraction channel, and the third feature extraction channel in parallel to construct an adaptive feature extraction channel; through the adaptive feature extraction channel, perform feature extraction on the standard target image to obtain the target feature descriptor.

[0018] In a possible implementation manner, when performing feature extraction on the standard target image through the adaptive feature extraction channel to obtain the target feature descriptor, the following processing is performed:

[0019] Evaluate the standard target image to generate a target image evaluation result; construct an adaptive mapping link for the feature extraction channel; according to the target image evaluation result, map and extract a target adaptation relationship from the adaptive mapping link of the feature extraction channel; based on the target adaptation relationship, call a target feature extraction channel from the adaptive feature extraction channel to perform feature extraction on the standard target image to obtain the target feature descriptor.

[0020] In a possible implementation manner, when constructing the adaptive mapping link for the feature extraction channel, the following processing is performed:

[0021] Obtain a first sample image of the target scene, evaluate the first sample image to generate a first sample image evaluation result; input the first sample image into three feature extraction channels respectively for feature extraction to obtain multiple feature extraction results; score the multiple feature extraction results, and establish a first adaptation relationship between the feature extraction channel corresponding to the maximum score and the first sample image evaluation result; obtain multiple sample images of the target scene, and establish multiple adaptation relationships between multiple image evaluation results and the feature extraction channels corresponding to multiple maximum scores according to the multiple sample images; cluster the multiple adaptation relationships to construct an adaptive mapping link for the feature extraction channel.

[0022] It is intended to use the azimuth generation method of the AR digital telescope based on spatial positioning analysis proposed in this application. First, obtain the first position and the first attitude information of the AR digital telescope in the earth coordinate system through the spatial positioning system, then collect a target image through the image acquisition device of the AR digital telescope, then preprocess the target image and extract image features to obtain a target feature descriptor. Similarly, obtain a panoramic image of the target scene and perform feature extraction on the panoramic image to obtain a panoramic feature descriptor. Furthermore, perform feature matching on the target feature descriptor and the panoramic feature descriptor to obtain matching points, and then determine the second position of the matching points in the earth coordinate system according to the position of the matching points in the earth coordinate system of the panoramic image. Finally, fuse the first position, the first attitude information and the second position to calculate the azimuth of the target image, achieving the technical effect of improving the accuracy of azimuth calculation. Brief Description of the Drawings

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the methods according to the embodiments of this application. It should be understood that the operations in the front or below do not necessarily need to be precisely executed in sequence. On the contrary, according to the need, various steps can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several operations can be removed from these processes.

[0024] Figure 1 It is a schematic flowchart of the azimuth angle generation method of the AR digital telescope based on spatial positioning analysis provided by the embodiments of this application.

[0025] Figure 2 It is a schematic flowchart of obtaining the target image azimuth angle in the azimuth angle generation method of the AR digital telescope based on spatial positioning analysis provided by the embodiments of this application. Detailed implementation manners

[0026] The above description is only an overview of the technical solutions of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of this application more obvious and understandable, the following specifically gives the detailed implementation manners of this application.

[0027] In order to make the purpose, technical solutions and advantages of this application clearer, the following will further describe this application in detail with reference to the accompanying drawings. The described embodiments should not be regarded as limitations of this application. All other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of this application.

[0028] In the following description, "some embodiments" are involved, which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict. The terms "first\second" involved are only used to distinguish similar objects and do not represent a specific order for the objects. The terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or server including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules not clearly listed or inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application.

[0029] The embodiment of this application provides a method for generating the azimuth angle of an AR digital telescope based on spatial positioning analysis, as follows Figure 1 shown, the method includes:

[0030] Step S100, obtaining the first position and the first attitude information of the AR digital telescope in the earth coordinate system through a spatial positioning system. Specifically, the AR digital telescope is integrated with a spatial positioning system, which is a system that uses satellites or other wireless signal sources to determine the position of a certain point on the earth, such as GPS, Beidou or GLONASS, etc. The spatial positioning system is turned on and in a normal working state. The spatial positioning system receives the positioning signals from the satellites and uses the received signals to calculate the current position information of the AR digital telescope, such as longitude, latitude, altitude, etc., and the attitude information, such as pitch angle, yaw angle, roll angle, etc.

[0031] In a possible implementation manner, step S100 further includes step S110, and the first attitude information includes the pitch angle, yaw angle and roll angle of the AR digital telescope.

[0032] Step S200, collecting a target image through the image acquisition device of the AR digital telescope. Specifically, start the camera or other image acquisition device on the AR digital telescope. Automatically or manually adjust the parameters of the camera, such as focal length, exposure, etc., according to the ambient light and the shooting target. The camera takes a picture to obtain the target image within the current field of view.

[0033] Step S300, preprocessing the target image and extracting image features to obtain a target feature descriptor. Specifically, perform processing such as denoising and enhancing the contrast on the target image to improve the image quality and the analysis accuracy. Use an image processing algorithm (such as SIFT, etc.) to detect key points (such as corner points, edge points, etc.) in the target image, and calculate the feature descriptors (feature vectors) of these key points.

[0034] In a possible implementation, preprocess the target image and extract image features to obtain a target feature descriptor. Step S300 further includes step S310 of performing preprocessing such as noise removal, contrast enhancement, edge enhancement, and scale normalization on the target image to obtain a standard target image. Specifically, according to the type of image noise (such as Gaussian noise, salt-and-pepper noise, etc.), select a denoising algorithm, such as median filtering, Gaussian filtering, or bilateral filtering, etc. Apply the selected denoising algorithm to each pixel or pixel block of the target image to reduce or eliminate the noise in the image. Calculate the histogram of the target image and apply histogram equalization technology to expand the contrast range of the image to make the details of the image clearer. Use an edge detection algorithm to identify the edges in the image and perform enhancement processing on the detected edges, such as increasing the brightness or contrast of the edge pixels to make the edges more prominent. Scale the processed image to a unified size so that the subsequent feature extraction and matching processes are carried out on the same scale. When adjusting the size, ensure that the aspect ratio of the image remains unchanged to avoid image distortion. Step S320 is to perform adaptive feature extraction on the standard target image to obtain the target feature descriptor. Specifically, analyze the characteristics of the standard target image, such as texture complexity, color distribution, shape features, lighting conditions, etc. According to the characteristics of the image, adaptively select the most suitable feature extraction method or combine multiple methods. For example, if the image has rich texture information, a texture-based feature extraction method (such as LBP, etc.) can be selected. Apply the selected feature extraction method or method combination to the standard target image for feature extraction to generate the target feature descriptor. This implementation uses an adaptive feature extraction method to extract the feature descriptor of the standard target image. This method makes full use of the useful information in the image, is more flexible and effective than fixedly using a certain feature extraction method, thus generating a more discriminative and robust feature descriptor. The feature descriptor obtained in this way more accurately reflects the inherent characteristics of the target image and provides strong support for subsequent feature matching.

[0035] In a possible implementation, for adaptively extracting features from the standard target image to obtain the target feature descriptor, step S320 further includes step S321 of respectively constructing a texture feature extraction channel, a corner feature extraction channel, an edge feature extraction channel, and a high-level semantic feature extraction channel based on the LBP algorithm, the SIFT algorithm, the Canny algorithm, and the RefineNet model. Specifically, the Local Binary Patterns (LBP) algorithm is selected to construct the texture feature extraction channel, and the LBP algorithm generates a local texture descriptor by comparing the gray value of the central pixel with those of its surrounding pixels. The Scale-Invariant Feature Transform (SIFT) algorithm is selected to construct the corner feature extraction channel, and the SIFT algorithm can detect scale-invariant key points in the image and generate corresponding feature descriptors. The Canny edge detection algorithm is selected to construct the edge feature extraction channel, and the Canny algorithm can identify strong edges in the image. The RefineNet model is selected to construct the high-level semantic feature extraction channel, and RefineNet is a deep learning architecture for image segmentation and edge detection that can extract high-level semantic information in the image. Step S322: Combine the texture feature extraction channel with any one of the corner feature extraction channel, the edge feature extraction channel, and the high-level semantic feature extraction channel to construct a first feature extraction channel; combine the texture feature extraction channel with any two of the corner feature extraction channel, the edge feature extraction channel, and the high-level semantic feature extraction channel to construct a second feature extraction channel; and combine the texture feature extraction channel with the corner feature extraction channel, the edge feature extraction channel, and the high-level semantic feature extraction channel to construct a third feature extraction channel. Specifically, select any one of the texture feature extraction channel, the corner feature extraction channel, the edge feature extraction channel, and the high-level semantic feature extraction channel for combination to construct three first feature extraction channels. Select any two of the texture feature extraction channel, the corner feature extraction channel, the edge feature extraction channel, and the high-level semantic feature extraction channel for combination to construct three second feature extraction channels. Combine the texture feature extraction channel with all other feature extraction channels to construct one third feature extraction channel.

[0036] Step S323: Connect the first feature extraction channel, the second feature extraction channel, and the third feature extraction channel in parallel to construct an adaptive feature extraction channel. Specifically, connect the three first feature extraction channels, the three second feature extraction channels, and the one third feature extraction channel constructed in step S322 in parallel to form a seven-channel adaptive feature extraction architecture. During the feature extraction process, each channel will independently process the input image and extract features of its respective type. Step S324: Through the adaptive feature extraction channel, perform feature extraction on the standard target image to obtain the target feature descriptor. Specifically, conduct a preliminary analysis of the standard target image to understand its basic characteristics. Based on the results of the pre-analysis, select the most suitable feature extraction channel to perform feature extraction on the image and generate the target feature descriptor. This implementation method, by constructing multiple feature extraction channels and allowing selection according to different image characteristics, has high flexibility and adaptability and can effectively extract and utilize the information in the image.

[0037] In a possible implementation, the step of extracting features from the standard target image through the adaptive feature extraction channels to obtain the target feature descriptor, step S324 further includes step S3241 of evaluating the standard target image to generate a target image evaluation result. Specifically, a comprehensive analysis of the standard target image is performed, including but not limited to characteristics such as the texture, shape, edges, color distribution, brightness, and contrast of the image. According to the image characteristics, a set of evaluation criteria is formulated, including multiple dimensions such as the clarity, complexity, and information content of the image. Based on the above analysis and evaluation criteria, the standard target image is classified to generate a target image evaluation result, which is the basis for subsequent selection of the feature extraction channels. Step S3242 is to construct an adaptive mapping link for the feature extraction channels. Specifically, according to the image characteristics, a set of mapping rules between the feature extraction channels and the image evaluation result is designed, and these rules define how different evaluation results should correspond to different feature extraction channels. Based on the mapping rules, an adaptive mapping link for the feature extraction channels is constructed, and this link is a logical connection relationship used to guide how to select the most suitable feature extraction channel given the image evaluation result. Step S3243 is to map and extract the target adaptation relationship from the adaptive mapping link of the feature extraction channels according to the target image evaluation result. Specifically, the target image evaluation result generated in step S3241 is matched with the rules in the adaptive mapping link of the feature extraction channels. According to the matching result, the feature extraction channel corresponding to the target image evaluation result is extracted from the mapping link, that is, the target adaptation relationship. Step S3244 is to call the target feature extraction channel from the adaptive feature extraction channels based on the target adaptation relationship to extract features from the standard target image to obtain the target feature descriptor. Specifically, according to the target adaptation relationship extracted in step S3243, the corresponding feature extraction channel is selected from the adaptive feature extraction channels. The selected feature extraction channel is used to extract features from the standard target image to generate a target feature descriptor. This implementation method first evaluates the standard target image, and then dynamically selects the most suitable feature extraction channel for feature extraction according to the evaluation result, significantly improving the pertinence and effectiveness of feature extraction, avoiding the waste of computing resources and feature redundancy problems caused by blindly using all feature extraction channels, and better adapting to the changes in different image characteristics at the same time.

[0038] In a possible implementation, the step of constructing the feature extraction channel adaptive mapping link, step S3242 further includes step S32421 of obtaining a first sample image of the target scene, evaluating the first sample image, and generating a first sample image evaluation result. Specifically, the first sample image is collected from the target scene. The first sample image is comprehensively evaluated using a preset evaluation criterion. According to the evaluation result, a classification label is generated as the evaluation result of the first sample image. Step S32422 is to input the first sample image into three feature extraction channels respectively for feature extraction to obtain multiple feature extraction results. Specifically, the first sample image is input into the first feature extraction channel, the second feature extraction channel, and the third feature extraction channel respectively for feature extraction, and corresponding feature descriptors are extracted from each channel. Step S32423 is to score the multiple feature extraction results and establish a first adaptation relationship between the feature extraction channel corresponding to the maximum score and the first sample image evaluation result. Specifically, each feature extraction result is scored to evaluate its representativeness and effectiveness for the first sample image. The feature extraction result with the highest score (i.e., the maximum score) is found. The feature extraction channel corresponding to the extremely high score is associated with the evaluation result of the first sample image to form a first adaptation relationship. Step S32424 is to obtain multiple sample images of the target scene and establish multiple adaptation relationships between multiple image evaluation results and the feature extraction channels corresponding to multiple maximum scores according to the multiple sample images. Specifically, more sample images in the target scene other than the first sample image are obtained. The processes of steps S32421 to S32423 are repeated for each new sample image, and a corresponding adaptation relationship is established for each sample image and accumulated to form multiple adaptation relationships. Step S32425 is to cluster the multiple adaptation relationships to construct a feature extraction channel adaptive mapping link. Specifically, the similarities and differences between the multiple adaptation relationships are analyzed, and a clustering algorithm (such as K-means) is used to divide the multiple adaptation relationships into different groups or categories. According to the clustering result, an adaptive mapping link between the feature extraction channel and the image evaluation result is constructed, and this link guides how to dynamically select the most suitable feature extraction channel according to the image evaluation result. This implementation method establishes a correspondence relationship between image characteristics and the optimal feature extraction channel through the evaluation and feature extraction experiments of a large number of sample images. This correspondence relationship is simplified and generalized through clustering to form an adaptive mapping link, improving the accuracy of constructing the adaptive mapping link, and further improving the accuracy of feature extraction channel selection.

[0039] Step S400: Obtain a panoramic view of the target scene, and perform feature extraction on the panoramic view to obtain a panoramic feature descriptor. Specifically, the panoramic view is a wide-angle image that shows a complete 360-degree view of the target scene, and the panoramic view is pre-shot and stored in the device. Apply the same feature extraction algorithm as in step S300 to the obtained panoramic view, that is, preprocess the panoramic view (such as denoising, enhancing contrast, etc.), then detect key points in the image (such as corner points, edge points, etc.), and calculate feature descriptors for these key points.

[0040] Step S500: Perform feature matching on the target feature descriptor and the panoramic feature descriptor to obtain matching points. Specifically, use the FLANN feature matching algorithm to compare the target feature descriptor obtained in step S300 with the panoramic feature descriptor obtained in step S400, and find feature point pairs whose similarity exceeds a preset threshold. These feature point pairs are the matching points, and they represent the same physical position or feature in different images.

[0041] Step S600: Determine the second position of the matching point in the Earth coordinate system according to the position of the matching point in the Earth coordinate system in the panoramic view. Specifically, use the geographic coordinate mapping information of the panoramic view, such as the GPS position when the panoramic view is taken, to convert the pixel position in the panoramic view into the actual position in the Earth coordinate system. Convert the pixel position of the matching point obtained in step S500 in the panoramic view into the actual position in the Earth coordinate system through the geographic coordinate mapping relationship, and this position is the second position of the matching point in the Earth coordinate system.

[0042] Step S700: Integrate the first position, the first attitude information, and the second position to calculate the azimuth angle of the target image. Specifically, integrate the first position (the position of the AR digital telescope in the Earth coordinate system) and the first attitude information (pitch angle, yaw angle, roll angle, etc.) obtained in step S100 with the second position (the position of the matching point in the Earth coordinate system) obtained in step S600. Using the methods of geometric transformation and coordinate conversion, calculate the azimuth angle of the target image relative to the AR digital telescope based on the integrated data. This azimuth angle describes the direction and angle of the target image in space. Superimpose the calculated azimuth angle onto the field of view of the AR digital telescope user in a virtual manner to provide the positioning information of the target image relative to the AR digital telescope. In the embodiment of the present application, a high-precision spatial positioning system is used to obtain the accurate position and attitude information of the AR digital telescope in the Earth coordinate system in real time. The target image is captured by the image acquisition device of the AR digital telescope and preprocessed. Feature extraction is performed on the preprocessed target image to obtain the target feature descriptor. At the same time, the panoramic image of the target scene is obtained and feature extraction is performed to obtain the panoramic feature descriptor. The matching points between the two are found through the feature matching algorithm. Based on the position of the matching point in the Earth coordinate system of the panoramic image, combined with the position and attitude information of the AR digital telescope, the accurate azimuth angle of the target image is finally obtained through spatial geometric calculation and other technical means, achieving the technical effect of improving the accuracy of azimuth angle calculation.

[0043] Such as Figure 2As shown, in a possible implementation, to fuse the first position, the first attitude information, and the second position to calculate the azimuth angle of the target image, step S700 further includes step S710 of determining the basic direction vector between the first position and the second position. Specifically, obtain the specific coordinates of the first position and the second position. Based on the coordinates of these two positions, calculate a vector pointing from the first position to the second position, and this vector is the basic direction vector. Step S720 is to convert the pitch angle, yaw angle, and roll angle in the first attitude information into a rotation matrix. Specifically, the pitch angle is the angle of rotation around the X-axis, the yaw angle is the angle of rotation around the Z-axis, and the roll angle is the angle of rotation around the Y-axis. Use the conversion formula from Euler angles to a rotation matrix to convert these three angles into a 3x3 rotation matrix. Step S730 is to fuse the basic direction vector with the rotation matrix to obtain a corrected direction vector. Specifically, multiply the basic direction vector by the rotation matrix to obtain the corrected direction vector. The corrected direction vector takes into account the attitude of the AR digital telescope and more accurately reflects the actual direction of the target image relative to the AR digital telescope. Step S740 is to calculate the angle of the corrected direction vector relative to the geographic north using spherical trigonometry to obtain the azimuth angle of the target image. Specifically, use the formula in spherical trigonometry to calculate the angle between the corrected direction vector and the geographic north, which is the azimuth angle of the target image. This implementation comprehensively considers the first position, the second position, and the attitude of the AR digital telescope, and accurately calculates the azimuth angle of the target image.

[0044] In a possible implementation, step S740 further includes step S741, and the formula for calculating the azimuth angle of the target image is as follows:

[0045] α' = arctan2(V y ”, V x ”) + Δα dynamic ;

[0046] where α' is the azimuth angle of the target image, V y ” and V x ” are the x and y components of the corrected direction vector V” in the Cartesian coordinate system respectively, and Δα dynamic is the dynamic error correction factor of the sensor. The formula for calculating the corrected direction vector V” is as follows:

[0047] V” = R × V';

[0048] where R is the rotation matrix and V' is the corrected direction vector. The formula for calculating the rotation matrix R is as follows:

[0049]

[0050] where is the yaw angle The rotation matrix about the Z-axis, is the roll angle The rotation matrix about the Y-axis, is the pitch angle The rotation matrix about the X-axis. The rotation matrix of is as follows:

[0051]

[0052]

[0053]

[0054] The calculation formula of the corrected direction vector V' is as follows:

[0055]

[0056] where V is the base direction vector, ΔR is the ellipsoid radius error correction amount based on the WGS84 ellipsoid model, is the normal vector of the earth's surface, Δh is the height difference between the first position and the second position, is the unit vector in the height direction. The calculation formula of the base direction vector V is as follows:

[0057] V = (x2 - x1, y2 - y1, z2 - z1);

[0058] where x1, y1, z1 are the Cartesian coordinates of the first position, and x2, y2, z2 are the Cartesian coordinates of the second position. This implementation provides a calculation formula for the azimuth angle of the target image. The formula corrects the error caused by the earth's curvature and takes into account the height difference between positions, improving the accuracy of the position information. By introducing the dynamic error correction factor of the sensor, the influence of sensor error on azimuth angle calculation is reduced. The entire calculation process is based on a rigorous mathematical model (rotation matrix, vector operation, spherical trigonometry, etc.), ensuring the accuracy and reliability of the calculation results.

[0059] The above specific embodiments do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present application shall be included within the protection scope of the present application. In some cases, the actions or steps recited in the present application can be executed in a sequence different from that in the embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or consecutive order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method for generating an azimuth angle of an AR digital telescope based on spatial positioning analysis, characterized in that: The method comprises: Acquire a first position and a first posture information of the AR digital telescope in the earth coordinate system through a spatial positioning system; Collect target images through the image acquisition device of the AR digital telescope; Preprocessing the target image and extracting image features to obtain a target feature descriptor; Acquire a panoramic image of the target scene, and perform feature extraction on the panoramic image to obtain a panoramic feature descriptor; Perform feature matching on the target feature descriptor and the panoramic feature descriptor to obtain matching points; Determining a second position of the matching point in the earth coordinate system according to the position of the matching point in the earth coordinate system in the panoramic image; The first position, the first posture information and the second position are integrated to calculate the azimuth of the target image; The first attitude information includes the pitch angle, yaw angle and roll angle of the AR digital telescope; The fusing the first position, the first posture information and the second position to calculate the azimuth of the target image includes: determining a base direction vector between the first position and the second position; Convert the pitch angle, yaw angle and roll angle in the first posture information into a rotation matrix; The basic direction vector is merged with the rotation matrix to obtain a correction direction vector; The angle of the correction direction vector relative to geographic north is calculated using spherical trigonometry to obtain the azimuth of the target image.

2. The method for generating an azimuth angle of an AR digital telescope based on spatial positioning analysis according to claim 1, characterized in that: The azimuth angle calculation formula of the target image is as follows: α'=arctan2(V y ",V x ”)+Da dynamic ; Among them, α' is the azimuth angle of the target image, V y ”,V x " are the x and y components of the correction direction vector V" in the Cartesian coordinate system, Δα dynamic is the dynamic error correction factor of the sensor; The calculation formula of the correction direction vector V" is as follows: V"=R×V', where R is the rotation matrix and V' is the correction direction vector; The calculation formula of the rotation matrix R is as follows: in, is the yaw angle The rotation matrix around the Z axis, is the roll angle The rotation matrix around the Y axis, R x (θ) is the rotation matrix of the pitch angle θ around the X axis, R x (θ), The rotation matrix is ​​as follows: The calculation formula of the correction direction vector V' is as follows: Wherein, V is the basic direction vector, ΔR is the ellipsoid radius error correction value based on the WGS84 ellipsoid model, is the normal vector to the earth's surface, Δh is the height difference between the first position and the second position, is the unit vector in the height direction; The calculation formula of the basic direction vector V is as follows: V=(x2-x1, y2-y1, z2-z1), wherein x1, y1, z1 are the Cartesian coordinates of the first position, and x2, y2, z2 are the Cartesian coordinates of the second position.

3. The method for generating an azimuth angle of an AR digital telescope based on spatial positioning analysis according to claim 1, characterized in that: The preprocessing of the target image and extracting image features to obtain a target feature descriptor includes: Performing noise removal, contrast enhancement, edge enhancement and scale normalization preprocessing on the target image to obtain a standard target image; Adaptive feature extraction is performed on the standard target image to obtain the target feature descriptor.

4. The method for generating an azimuth angle of an AR digital telescope based on spatial positioning analysis according to claim 3, characterized in that: The step of performing adaptive feature extraction on the standard target image to obtain the target feature descriptor comprises: Based on the LBP algorithm, SIFT algorithm, Canny algorithm and RefineNet model, we construct the texture feature extraction channel, corner feature extraction channel, edge feature extraction channel and high-level semantic feature extraction channel respectively; Combining the texture feature extraction channel with any one of the corner feature extraction channel, the edge feature extraction channel, and the advanced semantic feature extraction channel to construct a first feature extraction channel; combining the texture feature extraction channel with any two of the corner feature extraction channel, the edge feature extraction channel, and the advanced semantic feature extraction channel to construct a second feature extraction channel; combining the texture feature extraction channel with the corner feature extraction channel, the edge feature extraction channel, and the advanced semantic feature extraction channel to construct a third feature extraction channel; Connecting the first feature extraction channel, the second feature extraction channel and the third feature extraction channel in parallel to construct an adaptive feature extraction channel; The feature extraction is performed on the standard target image through the adaptive feature extraction channel to obtain the target feature descriptor.

5. The method for generating an azimuth angle of an AR digital telescope based on spatial positioning analysis according to claim 4, characterized in that: The step of extracting features from the standard target image through the adaptive feature extraction channel to obtain the target feature descriptor includes: Evaluating the standard target image to generate a target image evaluation result; Construct feature extraction channel adaptive mapping link; Extracting a target adaptation relationship from the feature extraction channel adaptive mapping link mapping according to the target image evaluation result; Based on the target adaptation relationship, a target feature extraction channel is called from the adaptive feature extraction channel to perform feature extraction on the standard target image to obtain the target feature descriptor.

6. The method for generating an azimuth angle of an AR digital telescope based on spatial positioning analysis according to claim 5, characterized in that: The step of constructing a feature extraction channel adaptive mapping link comprises: Acquire a first sample image of the target scene, evaluate the first sample image, and generate a first sample image evaluation result; Inputting the first sample image into three feature extraction channels to perform feature extraction respectively, to obtain multiple feature extraction results; Scoring the plurality of feature extraction results, and establishing a first adaptation relationship between the feature extraction channel corresponding to the maximum score and the first sample image evaluation result; Acquire a plurality of sample images of the target scene, and establish a plurality of adaptation relationships between a plurality of image evaluation results and a plurality of feature extraction channels corresponding to the maximum score values ​​according to the plurality of sample images; The multiple adaptation relationships are clustered to construct a feature extraction channel adaptive mapping link.

Images