A panoramic picture node link relationship automatic generation method

Abstract

The present application relates to a kind of panoramic picture node link relationship automatic generation method, belong to geographic spatial information technical field.The present application relates to a kind of panoramic picture node link relationship automatic generation method, the present application first changes plane panoramic picture into spherical panoramic picture model, and renders panoramic picture;Then establish panoramic image importance judgment standard, evaluate each panoramic image importance degree;Again using clustering algorithm, divide panoramic image, construct aerial panoramic image hierarchical model, generate hierarchical list;Finally, according to link generation rule, establish different image link relationship, and according to link label position, adjust link label rendering effect.Through the above process, the present application improves aerial panoramic image display effect, improves image spatial readability, can help observer to recognize geographic topography distribution.

Description

Technical Field

[0001] This invention relates to a method for automatically generating node link relationships in panoramic images, belonging to the field of geospatial information technology. Background Technology

[0002] With the continuous development of surveying and mapping technology, data acquisition methods and visualization forms are becoming increasingly diversified and professional. Aerial panoramic imagery, created by using rotary-wing drones to capture real-world images while hovering in the air, produces comprehensive, all-around panoramic maps. It shifts the observation perspective from the ground to the air, enabling panoramic observation from horizon to horizon and providing an aerial view and a sense of soaring through the sky. This is of great significance for understanding the overall spatial environment and has seen rapid development in recent years. As aerial photography equipment becomes increasingly sophisticated, the methods for capturing aerial panoramic images are gradually simplifying, and the amount of aerial panoramic image data is constantly increasing. The application of aerial panoramic imagery is developing towards diversification, multidimensionality, and professionalization, evolving from panoramic image display to photogrammetry based on aerial panoramic images. It is widely used in aerial photography, street view maps, target reconnaissance, and unmanned autonomous platforms. Aerial panoramic imagery has become a three-dimensional real-world tool, alongside traditional remote sensing images and virtual geographic environments, for recording the real world from an aerial perspective and transmitting information, providing a valuable way for people to understand the world. To meet users' needs for understanding their surroundings and roaming through panoramic aerial images, an image hierarchy is established, jump links are generated, and link tags are added to the images to accurately indicate spatial distribution, thereby improving the user experience from a cognitive perspective.

[0003] Image linking, based on aerial panoramic image data, establishes image navigation by adapting to user task requirements and spatial relationships. This enhances readability and interactivity, and can be used for multi-angle and multi-view surveys of scenes of varying sizes. With the development of aerial panoramic image acquisition technology, the amount of aerial panoramic image data is increasing daily. However, individual aerial panoramic images are independent, lacking effective integration methods and ways to express spatial relationships, which reduces the readability of panoramic images and significantly impacts their display quality. Summary of the Invention

[0004] The purpose of this invention is to provide a method for automatically generating the node link relationship of a panoramic image, so as to solve the problem of low readability of current aerial panoramic images.

[0005] To address the aforementioned technical problems, this invention provides a method for automatically generating panoramic image node link relationships, comprising the following steps:

[0006] 1) Obtain a planar panoramic image and perform spherical modeling to obtain a spherical panoramic image;

[0007] 2) Based on the task requirements, establish a criteria for judging the importance of panoramic images and evaluate the importance of each panoramic image;

[0008] 3) Use clustering algorithms to divide the images into clusters. Based on the distance of the images in each cluster to the center of that cluster, divide the images into different levels and obtain a list of different levels.

[0009] 4) Generate link relationships between panoramic images of different levels and within the same level according to the set link relationship generation rules and based on the importance of each panoramic image;

[0010] 5) Obtain the angle between each link direction, compare the angle with the angle threshold set to avoid rendering overlap, and adjust the rendering position of the link label according to the comparison result. The link label is the icon representation of the link, and the pixel coordinates of the link label on the panorama are calculated based on the orientation relationship between the two panoramas.

[0011] This invention first transforms a planar panoramic image into a spherical panoramic image model and renders the panoramic image. Then, it establishes a criterion for evaluating the importance of panoramic images, assessing the importance of each panoramic image. Next, it uses a clustering algorithm to segment the panoramic images, constructing a hierarchical model of aerial panoramic images and generating a hierarchical list. Finally, based on link generation rules, it establishes link relationships between different images and adjusts the rendering effect of link tags according to their positions. Through this process, this invention improves the display effect of aerial panoramic images, enhances image spatial readability, and helps observers understand geographical and topographical distribution.

[0012] The link relationship generation rules in step 4) include:

[0013] A. Links from high-level lists to low-level lists: For each cluster, with the line of sight from the center of the panoramic image of the high-level list as 0°, clockwise angles are positive and counterclockwise angles are negative, and each 90° interval is a separate interval, with only one link relationship established within the interval;

[0014] B. Links from lower-level list images to higher-level list images: Each image in the lower-level list establishes a link relationship with the image in the higher-level list.

[0015] C. Links between images in the same cluster and at the same level: For each cluster, with the center of the panoramic image and the line of sight as 0°, clockwise angles are positive and counterclockwise angles are negative. Images in non-lowest level lists are divided into intervals of 180°, and only one link is established within each interval; images in the lowest level list are divided into intervals of 90°, and only one link is established within each interval.

[0016] D. Linking images between sibling lists in different clusters: Calculate the distance between each aerial panoramic image in two sibling lists and link the two images with the smallest distance.

[0017] This invention establishes a link rule base based on the hierarchical list of images, which includes links from high-level lists to low-level lists, links from low-level list images to high-level list images, links between images in the same cluster and at the same level, and links between images in different clusters and at the same level, so that there are corresponding links within a hierarchy, between hierarchies, and between the same cluster and different clusters.

[0018] Furthermore, in step A, if there are at least two images in the next lower-level list within the interval, the image with the smallest distance is selected for linking.

[0019] When establishing links between high-level and low-level lists, this invention selects the image with the smallest distance when there are at least two images in the next low-level list within the interval, so that the established links can more accurately reflect the relationship between images.

[0020] Furthermore, in step C, if there are at least two images within the interval, the visibility of each image is determined, and the shortest distance is prioritized to establish a link relationship among the images with visible visibility.

[0021] When establishing links between images in the same cluster and at the same level, this invention selects the image with the shortest distance among the images that are accessible to establish the link when there are at least two images in the interval, so that the established link can more accurately reflect the relationship between the images.

[0022] Furthermore, in step 3), the image is divided into three levels, and the process for determining the list of each level is as follows:

[0023] First-level image list: Establish a planar coordinate system with the cluster center as the origin, the horizontal distance from the image to the center as the x-axis, and the difference between the image height and the center height as the y-axis. When the image... When the value is at its maximum, it becomes the central aerial panoramic image and is added to the first-level image list;

[0024] Second-level image list: Establish a planar coordinate system with the central aerial panoramic image as the origin, the horizontal distance from the image to the origin as the x-axis, and the height difference between the image and the central image as the y-axis, and calculate the average value of the images in the cluster. and If the distance is less than the average and the height difference is greater than the average, add it to the second-level image list;

[0025] Third-level image list: Images that do not belong to the first, second, or important lists are added to the third-level image list.

[0026] This invention determines a first-level image list based on the ratio of the height difference between images in each cluster and the height difference between the cluster center and the horizontal distance from the image in the cluster to the center. A second-level image list is determined based on the average horizontal distance and average height difference between images in each cluster to the cluster center.

[0027] Furthermore, the adjustment process in step 5) is as follows: if the included angle is within the set included angle threshold, it is determined that the position of the corresponding link mark needs to be adjusted, and the rendering state is updated according to the link tag avoidance formula.

[0028] In this invention, if the angle between each link direction is within a set angle threshold, the position of the link marker needs to be adjusted so that the angle between each link direction is not within the set angle threshold.

[0029] Furthermore, the calculation equation for avoiding link tag overlap is as follows:

[0030]

[0031] In the formula, (u1,v1) are the pixel coordinates of the left link marker, (u2,v2) are the pixel coordinates of the right link marker, and d is the length of the square link label.

[0032] Furthermore, the process for determining the visibility of an image is as follows:

[0033] Obtain the spatial rectangular coordinates (x′, y′, z′) of the ground point along the line connecting the two images. Compare the obtained z′ with z. If z ≥ z′, the two images are intervisible; otherwise, they are not intervisible. The formula for calculating z is as follows:

[0034]

[0035] Where (x0,y0,z0) are the spatial rectangular coordinates of this image, (x1,y1,z1) are the spatial rectangular coordinates of the linked image, and S is the Euclidean distance between the two images.

[0036] This invention determines whether two images are visible to each other based on the spatial rectangular coordinates of a ground point and the spatial rectangular coordinates of the two images, and can accurately determine the visibility between the images.

[0037] Furthermore, the clustering algorithm in step 2) adopts the K-means clustering algorithm.

[0038] Furthermore, in step 1), the spherical panoramic image is obtained by calculating the mapping relationship from the planar panoramic image to the spherical panoramic image. The calculation formula for this mapping relationship is as follows:

[0039]

[0040]

[0041]

[0042] Where (x) p ,y p ,zp Let (x′) be the coordinates of point P in the spherical coordinate system. p ,y′ p ,z′ p Let ) represent the coordinates of the projection point P′ on the sphere in the camera coordinate system, t be an intermediate parameter, f be the focal length of the panoramic camera, α be the angle between the x-axis of the spherical coordinate system and the x-axis of the camera coordinate system, and β be the angle between the y-axis of the spherical coordinate system and the y-axis of the camera coordinate system. c ,y c ,z c Let P be the coordinates of point P in the camera coordinate system.

[0043] This invention utilizes the mapping relationship between planar panoramic images and spherical panoramic images to convert planar panoramic images into spherical panoramic images, enabling fast and accurate conversion between the two. Attached Figure Description

[0044] Figure 1 This is a flowchart of the method for automatically generating panoramic image node link relationships according to the present invention;

[0045] Figure 2 This is a schematic diagram of spherical modeling in an embodiment of the present invention;

[0046] Figure 3 This is a projection diagram in an embodiment of the present invention;

[0047] Figure 4 This is a schematic diagram of an aerial panoramic image grading model in an embodiment of the present invention. Detailed Implementation

[0048] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0049] This invention first performs spherical modeling, utilizing the relationship between planar panoramic images and spherical panoramic images to obtain spherical panoramic images; based on image position and attitude parameters, it completes the positioning and orientation determination of aerial panoramic images; it establishes a panoramic image importance evaluation criterion to assess the importance of panoramic images; it uses the geospatial K-means clustering algorithm to divide the aerial panoramic image hierarchy list and establish an aerial panoramic image hierarchical model; it establishes a link building library, processes the image hierarchy list, generates image link relationships, and constructs a global link structure for aerial panoramic images; finally, it compares the link label angle with the label overlap angle threshold to determine whether to adjust the link label coordinates, thereby achieving automatic generation of panoramic image node link relationships. The implementation process of this method is as follows: Figure 1 As shown, the following is a detailed explanation.

[0050] 1. Obtain a planar panoramic image and perform spherical modeling to obtain a spherical panoramic image.

[0051] This embodiment utilizes a drone to capture panoramic images in the field. The camera's geodetic coordinates are stored in the platform's data file and can be exported via a data transfer cable. Based on the GPS / IMU module of the image capture platform (such as a drone), this invention obtains the geodetic coordinates (i.e., geodetic coordinates) and attitude of the panoramic image capture point, denoted as O(B). i ,L i H i In this embodiment, the specific content of the aerial panoramic image is shown in Table 1.

[0052] A panoramic image refers to a wide-view real-world image obtained by processing aerial image sequences using image stitching software. Simultaneously, the latitude and longitude coordinates (i.e., GPS data) of the panoramic image are obtained from the latitude and longitude coordinates of the sequence images. Based on the exported GPS data, the coordinates of the drone's shooting point are obtained. In this embodiment, the shooting point coordinates of the panoramic image of the Dawa Villa are (113.091420, 34.516967, 572.000000).

[0053] Table 1

[0054]

[0055] A spherical panoramic image is determined based on the mapping relationship from a planar panoramic image to a spherical panoramic image. The calculation principle is as follows:

[0056] Let the spherical coordinate system be XYZ, the camera coordinate system be xyz, the camera's shooting direction be (α,β), and the camera's focal length be f = 28mm. Figure 2 As shown, the corresponding point P′(x′,y′) on the sphere is a pixel P(x,y) in the planar panoramic image. The coordinates of point P in the camera coordinate system are (x1,y1,-f), where x1=xl / 2, y1=yh / 2, and l and h are the length and width of the panoramic image, respectively. The coordinates of point P in the XYZ coordinate system are (x1,y1,-f). p ,y p ,z p Then we have:

[0057]

[0058] In the formula, α is the horizontal deflection angle of the line of sight, and β is the vertical deflection angle of the line of sight.

[0059] from Figure 2 It can be seen from this that point P satisfies the parametric equation of the line formed by point P and point P′:

[0060]

[0061] In the formula, (x p ,y p ,zp Let (x′) be the coordinates of point P in the spherical coordinate system. p ,y′ p ,z′ p Let P' be the coordinates of the projection point P' onto the sphere in the camera coordinate system, satisfying the equation of the sphere:

[0062] x′ p 2 +y′ p 2 +z′ p 2 =f 2

[0063] Combining the two formulas above, we obtain the parameter t:

[0064]

[0065] Through the above process, the parametric coordinates (x′) of the projection point P′ on the sphere can be obtained. p ,y′ p ,z′ p ).

[0066] 2. Based on the task requirements, establish a criteria for judging the importance of panoramic images and evaluate the importance of each panoramic image.

[0067] To evaluate the importance of aerial panoramic images, this invention employs the following importance evaluation criteria:

[0068] Among them, GeoPosXY represents the geographic longitude and latitude of the aerial panoramic image, describing the image's geographical location and determining its relative planar position within the image set; Height represents the image's shooting height, affecting the spatial range recorded by the image, generally the greater the height, the larger the recorded area and the more links established; Time represents the image's shooting time, images taken at the same geographical location on different dates and within 24 hours will differ; HVT (Height-value target) indicates whether there is an important target in the image, this indicator varies depending on task requirements and largely determines the image's importance; Rank represents the administrative level of the province or city where the image is located; Environment indicates whether the image environment is urban or outdoor. More factors can also be considered depending on task requirements.

[0069] An importance hierarchy for panoramic images is established, dividing aerial panoramic images into two subsets based on importance evaluation criteria: Important = {Pano1, Pano2, Pano3, ...} and Normal = {Pano1, Pano2, Pano3, ...}. Important represents the set of panoramic images that must be linked in the first-level aerial panoramic image list; Normal represents the set of images linked according to the linking rules.

[0070] 3. Use clustering algorithms to divide the images into clusters, construct a hierarchical model for aerial panoramic images, and generate a hierarchical list of aerial panoramic images.

[0071] This invention employs the geospatial K-means clustering algorithm for clustering, the principle of which is as follows: For n aerial panoramic images X={X1,X2,,...,X... n}, using geographic coordinate attributes (x g ,y g ,z g This is a metric for measuring the similarity between objects. It groups n images with the highest similarity into k specified clusters, where each image belongs to exactly one cluster. The k cluster centers are initialized as follows: {C1, C2, ..., Cn...} k}(1<k≤n), calculate the cluster center C for each image. k The Euclidean distance is as follows:

[0072]

[0073] In the above formula, (x ig ,y ig ,z ig (x) represents the spatial rectangular coordinates of the i-th image, (x) jg ,y jg ,z jg ) represents the spatial rectangular coordinates of the j-th cluster center.

[0074] Compare the distance of each image to each cluster center, and assign each image to the cluster with the nearest cluster center, resulting in k clusters {S1, S2, ..., S...}. k Repeat the above formula calculation until any escape condition is met to obtain the final result. Escape conditions: No objects are reassigned to different clusters; cluster centers change; the sum of squared errors is locally minimized.

[0075] First-level image list: Establish a planar coordinate system with the cluster center as the origin, the horizontal distance from the image to the center as the x-axis, and the difference between the image height and the center height as the y-axis. When the image... When the value is at its maximum, it becomes the central aerial panoramic image and is added to the first-level image list.

[0076] Second-level image list: A planar coordinate system is established with the central aerial panoramic image as the origin, the horizontal distance from the image to the origin as the x-axis, and the height difference between the image and the central image as the y-axis. The average value of the images in the cluster is calculated. and If the distance is less than the average and the height difference is greater than the average, it should be added to the second-level image list.

[0077] Third-level image list: Images that do not belong to the first, second, or important lists are added to the third-level image list.

[0078] Establish a hierarchy among aerial panoramic images in different hierarchical lists within the same cluster.

[0079] In this embodiment, the image's geodetic latitude and longitude coordinates are first converted to WGS84 spatial rectangular coordinates. The specific method is as follows:

[0080] First, convert the geodetic latitude and longitude to spatial rectangular coordinates using the WGS84 coordinate datum:

[0081]

[0082] The major radius of the ellipsoid is a = 6378137m, and the first heart rate e of the WGS84 ellipsoid. 2 =0.00669437999013, (X,Y,Z) represent the WGS84 spatial rectangular coordinates, and (B,L,H) represent the geodetic coordinates, where H is altitude, L is latitude, and B is longitude. Based on the above formula, the spatial rectangular coordinates of the 16 panoramic images are obtained as shown in Table 2.

[0083] Table 2

[0084]

[0085] Using the geospatial K-means clustering algorithm, the number of clusters was set to 3, and the initial center coordinates were (113.103349, 34.524288, 450.0), (113.105613, 34.519028, 445.0), and (113.092190, 34.515462, 550.0). The distance from the panoramic image to each center coordinate was calculated using the following formula:

[0086]

[0087] In the above formula, (x ig ,y ig ,z ig (x) represents the spatial rectangular coordinates of the i-th image, (x) jg ,y jg ,zjg ) represents the spatial rectangular coordinates of the j-th cluster center.

[0088] Each image is clustered into the cluster with the smallest distance to obtain the initial clustering. The center coordinates and distances are repeatedly calculated and the images are clustered until the clustering no longer changes. Finally, the clustering results are shown in Table 3.

[0089] Table 3

[0090]

[0091]

[0092] The spatial rectangular coordinates of the cluster center are: C1(-2063639.296112,4839951.072679,3594042.360809), C2(-2064585.897152,4839119.001860,3594431.545798), C3(-2064435.612489,4838942.631189,3594720.535030).

[0093] A planar coordinate system is established with the cluster center as the origin, the horizontal distance from the image to the center in each cluster as the x-axis, and the difference between the image height and the center height as the y-axis. The values ​​of each image are then calculated. The image with the highest value is taken as the first panoramic image in the list. The specific results are: Zhifang Reservoir high altitude, Xuegou high altitude, and Dawa overall high altitude.

[0094] Using panoramic images centered on the following clusters—Pano1 (-2064186.476499, 4838937.005301, 3594919.102100) at Zhifang Reservoir, Pano2 (-2064611.318487, 4839130.929416, 3594437.559974) at Xuegou, and Pano3 (-2063657.894956, 4840041.155000, 3594146.986159) at Dawa—the average horizontal distance from each cluster of images to the central image is calculated. and average elevation difference If the distance is less than the average distance and the elevation difference is greater than the average elevation difference, add it to the second list, as detailed in Table 4. Figure 4 As shown in the figure, the rectangular border is the Normal set image, and the rounded rectangular border is the Important set image.

[0095] Table 4

[0096] Cluster number Second list 1 Zhifang Dam High Altitude 2 Xuegou Office 3 High-altitude areas of Dawa Square and Dawa Pavilion

[0097] According to the third list construction rules, the specific list is shown in Table 5.

[0098] Table 5

[0099] Cluster number Third list members 1 Northeast side of Zhifang Reservoir, southwest side of Zhifang Reservoir, middle section of Zhifang Reservoir, Zhifang Dam 2 Snow Valley Gate 3 Dawa Villas, Dawa Gate, Dawa Basketball Court, Dawa Dormitory Building

[0100] 4. Establish link relationship generation rules to generate link relationships between panoramic images of different and the same level.

[0101] (1) Links from high-level list to low-level list: For each cluster, the line of sight from the center of the panoramic image of the high-level list is 0°, the clockwise angle is positive and the counterclockwise angle is negative, and each 90° is an interval. Only one link relationship is established within the interval. If there are multiple images of the next low-level list within the interval, the image with the smallest distance is selected for linking.

[0102] (2) Links from low-level list images to high-level list images: Each image in the low-level list establishes a link relationship with the image in the higher-level list.

[0103] (3) Links between images in the same cluster and at the same level: For each cluster, the center of the panoramic image is taken as 0°, the clockwise angle is positive and the counterclockwise angle is negative. Images in the non-lowest level list are divided into intervals of 180°. Only one link relationship is established within the interval. If there are multiple images in the interval, the link with the shortest distance is established. Images in the lowest level list are divided into intervals of 90°. Only one link relationship is established within the interval. If there are multiple images in the interval, the visibility of each image is first determined. If the visibility is between images, the shortest distance is selected to establish the link relationship.

[0104] (4) Linking images in heterogeneous clusters: Calculate the distance between each aerial panoramic image in two clusters of the same level, and link the two images with the smallest distance.

[0105] (5) The important image list is only linked to images in the first-level image list of the same cluster;

[0106] (6) If an image is missing an entry or exit link, the link relationship is established with the nearest image. This rule has the highest priority.

[0107] In this embodiment, the angle between the vertical plane and the direction of the line connecting the panoramic image center viewpoint is calculated. The specific principle is as follows:

[0108] like Figure 3 As shown, let the direction of the view from the center of the panoramic image be n0 = (x0, y0, z0), and the direction of the connecting line between the images be n1 = (x1, y1, z1). Then the angle θ between the direction and the perpendicular plane is:

[0109]

[0110] Taking the Dawa basketball court as the base image and the Dawa dormitory building as the linked image, we calculate θ = 111.336442°. When linking using the linking rule library, we determine the number of panoramic images within the angle range by judging the angle between the vertical planes, and then further determine whether to establish a link.

[0111] It is also necessary to determine whether there is visual interoperability between panoramic images. The calculation formula is as follows:

[0112]

[0113] In the formula, (x0,y0,z0) are the spatial rectangular coordinates of the current image, (x1,y1,z1) are the spatial rectangular coordinates of the linked image, and S is the Euclidean distance between the two images.

[0114] Obtain the spatial rectangular coordinates (x′, y′, z′) of the ground point along the line connecting the two images, substitute them into the above formula, and compare the magnitudes of z and z′. If z ≥ z′, the two images are mutually visible; otherwise, they are not. Taking the Dawa basketball court as an example, substitute the two images of the Dawa dormitory building and a ground point (113.091890, 34.575879, 520.15) along the connecting line into the above formula, and calculate: zz′ = 12.92, indicating that the ground point cannot obscure the two images. Based on the third-level list constructed in step 3, use the link establishment rules to calculate the link angle and visibility, and obtain the image linking results.

[0115] 5. Obtain the angle between each link direction, compare the angle with the angle threshold set to avoid rendering overlap, and determine whether the rendering position of the link tag needs to be adjusted.

[0116] A link is an abstract logical relationship, which can be represented by a small icon. The pixel coordinates of the icon on the panoramic image are calculated based on the orientation relationship between the two linked panoramic images, similar to the forward and backward buttons in Baidu Street View. In this step, calculating the label position involves calculating the image coordinates of the rendered label on the spherical panoramic image based on the link direction. In this implementation case, the threshold for the overlapping vertical plane angle is set to 5°. The angle between the line of sight from the center of this image and the vertical plane between the two linked images is calculated according to the steps. The difference between the two angles gives the vertical plane angle of the link, calculated as follows:

[0117] Δθ=|θ1-θ2|

[0118] In the formula, θ1 is the angle between the line of sight from the center of this image and the vertical plane between one of the linked images, and θ2 is the angle between the line of sight from the center of this image and the vertical plane between the other linked image.

[0119] Taking the example image as the basketball court in Dawa, and the linked images as the dormitory building and the sky above Dawa Square, the angle between the link tags is calculated to be 21.122271°, so the two link tags do not overlap.

[0120] If link tags overlap, adjust the link tag coordinates according to the following formula:

[0121]

[0122] In the formula, (u1,v1) are the pixel coordinates of the left link label, and (u1,v1) are the pixel coordinates of the right link label.

[0123] 6. Set the label text attribute parameters and complete the rendering.

[0124] In this step, the text parameters set include: text font, text size, text color, whether the text has an outline, and the outline color, among others. Finally, based on the tag positions set in step 6, the renderable link tags and tag text are rendered.

[0125] The automatic generation method for panoramic image node link relationships of the present invention first obtains the geodetic coordinates and pose parameters of the planar panoramic image; performs spherical modeling, and obtains the spherical panoramic image by utilizing the relationship between the planar panoramic image and the spherical panoramic image; establishes an aerial panoramic image importance evaluation model, dividing the panoramic image data into two sets; uses the geospatial K-means clustering algorithm to cluster the aerial panoramic images, constructs an aerial panoramic image hierarchical model, and generates an aerial panoramic image hierarchy list; uses a link rule library to establish links between different image hierarchy lists; and determines whether the established aerial panoramic image link tags overlap. Based on the determination, the rendering position of the tags is adjusted; then, the tag text attribute parameters are set to complete the rendering of the link tags.

[0126] This invention solves the problem of automatic generation of aerial panoramic image links, improves the spatial readability of aerial spherical panoramic images, and can help observers understand the distribution of geographical terrain and their own location. It can be used in fields related to virtual geographical environment fusion display, environmental survey, map street view, and other aerial panoramic image display.

Claims

1. A method for automatically generating node link relationships in a panoramic image, characterized in that, The generation method includes the following steps: 1) Obtain a planar panoramic image and perform spherical modeling to obtain a spherical panoramic image; 2) Based on the task requirements, establish a criteria for judging the importance of panoramic images and evaluate the importance of each panoramic image; 3) Use clustering algorithms to divide the image into clusters. Based on the distance from the image in each cluster to the center of that cluster, the image is divided into three levels. The process of determining the list of each level is as follows: First-level image list: Establish a planar coordinate system with the cluster center as the origin, the horizontal distance from the image to the center as the x-axis, and the difference between the image height and the center height as the y-axis. When the image... When the value is at its maximum, it becomes the central aerial panoramic image and is added to the first-level image list; Second-level image list: Establish a planar coordinate system with the central aerial panoramic image as the origin, the horizontal distance from the image to the origin as the x-axis, and the height difference between the image and the central image as the y-axis, and calculate the average value of the images in the cluster. and If the distance is less than the average and the height difference is greater than the average, add it to the second-level image list; Third-level image list: Images that do not belong to the first, second, or important lists are added to the third-level image list; 4) Generate link relationships between panoramic images of different levels and within the same level according to the set link relationship generation rules and based on the importance of each panoramic image; 5) Obtain the angle between each link direction, compare the angle with the angle threshold set to avoid rendering overlap, and adjust the rendering position of the link label according to the comparison result. The link label is the icon representation of the link, and the pixel coordinates of the link label on the panorama are calculated based on the orientation relationship between the two panoramas.

2. The method for automatically generating panoramic image node link relationships according to claim 1, characterized in that, The link relationship generation rules in step 4) include: A. Links from high-level lists to low-level lists: For each cluster, with the line of sight from the center of the panoramic image of the high-level list as 0°, clockwise angles are positive and counterclockwise angles are negative, and each 90° interval is a separate interval, with only one link relationship established within the interval; B. Links from lower-level list images to higher-level list images: Each image in the lower-level list establishes a link relationship with the image in the higher-level list; C. Links between images in the same cluster and at the same level: For each cluster, with the center of the panoramic image and the line of sight as 0°, clockwise angles are positive and counterclockwise angles are negative. Images in non-lowest level lists are divided into intervals of 180°, and only one link is established within each interval; images in the lowest level list are divided into intervals of 90°, and only one link is established within each interval. D. Linking images in sibling lists of different clusters: Calculate the distance between each aerial panoramic image in two sibling lists and link the two images with the smallest distance.

3. The method for automatically generating panoramic image node link relationships according to claim 2, characterized in that, In step A, if there are at least two images in the next lower-level list within the interval, the image with the smallest distance is selected for linking.

4. The method for automatically generating panoramic image node link relationships according to claim 2, characterized in that, In step C, if there are at least two images within the interval, the visibility of each image is determined, and the shortest distance image is prioritized to establish a link relationship among the images with visible visibility.

5. The method for automatically generating panoramic image node link relationships according to claim 2, characterized in that, Step 3) also includes establishing the hierarchical relationship between aerial panoramic images of different levels within the same cluster.

6. The method for automatically generating panoramic image node link relationships according to claim 2, characterized in that, The adjustment process in step 5) is as follows: if the included angle is within the set included angle threshold, it is determined that the position of the corresponding link mark needs to be adjusted, and the rendering state is updated according to the link tag avoidance formula.

7. The method for automatically generating panoramic image node link relationships according to claim 6, characterized in that, The calculation equation for avoiding link tag overlap is as follows: ； In the formula, Mark the pixel coordinates for the left-hand link. The pixel coordinates of the right-hand link tag are given, and d is the length of the square link tag.

8. The method for automatically generating panoramic image node link relationships according to claim 4, characterized in that, The process for determining the visibility of an image is as follows: Obtain the spatial rectangular coordinates of the ground point along the line connecting the two images. , will get Compare z with z, if If z is perpendicular to the image, the two images are mutually visible; otherwise, they are not mutually visible. The formula for calculating z is as follows: ； in, These are the spatial rectangular coordinates of this image. Let S be the spatial rectangular coordinates of the linked images, and S be the Euclidean distance between the two images.

9. The method for automatically generating panoramic image node link relationships according to claim 1 or 5, characterized in that, The clustering algorithm in step 2) is the K-means clustering algorithm.

10. The method for automatically generating panoramic image node link relationships according to claim 1 or 5, characterized in that, In step 1), the spherical panoramic image is obtained by calculating the mapping relationship from the planar panoramic image to the spherical panoramic image. The calculation formula for this mapping relationship is as follows: ； ； ； in for The coordinates of a point in a spherical coordinate system Projection point on the sphere Coordinates in the camera coordinate system For intermediate parameters, The focal length of the panoramic camera. For spherical coordinate system Axis and camera coordinate system The angle between the axes, For spherical coordinate system Axis and camera coordinate system The angle between the axes, For point Coordinates in the camera coordinate system.

Images