Satellite image uncontrolled block adjustment gross error image detection method and device

Abstract

The invention relates to a satellite image control-free block adjustment gross error image detection method and device, and the method comprises the steps: introducing external source digital elevation model data to quantitatively express the relative geometric positioning precision and change conditions of a satellite image in a block, and carrying out the calculation of the relative geometric positioning precision under the constraint of an initial seed image, and the relative positioning precision between the images is converted into the relative positioning precision relative to the initial seed image, so that gross error image detection is realized. Therefore, the problems that an existing block adjustment gross error detection method excessively depends on a large amount of manual post-check for gross error image detection, gross error images cannot be detected before block adjustment and the like are solved.

Description

Technical Field

[0001] The present application relates to the technical field of remote sensing satellite image regional network adjustment, and in particular to a method and device for detecting gross error images of satellite images without control regional network adjustment. Background Art

[0002] Block adjustment without ground control is the core link to achieve high-precision positioning and mapping of images in areas without ground control abroad. It uses the connection points between matching satellite images as the main observation values ​​to perform overall adjustment and improve the geometric accuracy of satellite images. However, the connection points obtained by automatic image matching technology usually contain a large number of gross errors, and it is necessary to detect and eliminate gross error connection points to ensure the accuracy and reliability of block adjustment.

[0003] However, due to the influence of various factors, such as image quality and image acquisition conditions, not only may there be gross errors in the observation values ​​(connection points), but the satellite images themselves may also have significant positioning deviations compared to the overall regional network. Such images are gross error images. If gross error images are directly included in the uncontrolled regional network adjustment, it will not only lead to an overall decrease in the overall regional network adjustment accuracy, but may also conflict with the additional observations introduced to overcome the rank deficiency of the adjustment model, causing relative changes in the strength of local observation conditions within the network, resulting in distortion and deformation within the regional network. Therefore, it is necessary to effectively and accurately detect gross error images in the regional network.

[0004] However, the existing methods for detecting gross errors in block adjustment mainly focus on the observation values ​​of the connection points of the block. The detection of gross error images still relies on a large amount of manual post-inspection, and it is impossible to detect gross error images before block adjustment, which needs to be solved urgently. Summary of the invention

[0005] The present application provides a method and device for detecting gross error images in uncontrolled regional network adjustment of satellite images, so as to solve the problems that the existing regional network adjustment gross error detection method relies too much on a large number of manual post-inspections for gross error image detection and cannot detect gross error images before regional network adjustment.

[0006] The first aspect of the present application provides a method for detecting gross error images of satellite images without control area network adjustment, comprising the following steps: obtaining multi-view image data corresponding to a target area, connection point data corresponding to the multi-view image data, and digital elevation model data covering the target area, and determining an initial seed image in the multi-view image data according to the digital elevation model data, the multi-view image data, and the connection point data; extracting initial connection point data of adjacent images corresponding to the initial seed image, and calculating relative positioning residuals corresponding to the initial connection point data, and statistically calculating residual means and residual variances of image square positioning residuals of the initial connection point data in row and column directions, Based on the residual mean and the residual variance, the gross error connection points in the initial connection point data are eliminated; a new seed image in the multi-scene image data is iteratively determined, and a new relative positioning residual of the non-seed image relative to the initial seed image is calculated in each iteration process, and based on the new relative positioning residual, a new residual mean and a new residual variance of the non-initial seed image in the multi-scene image data relative to the initial seed image in the row and column directions are determined, and a residual mean space is constructed according to the new residual mean and the new residual variance, and a confidence ellipse corresponding to the residual mean space is calculated, so as to use the confidence ellipse to perform gross error image detection for each scene image data.

[0007] Optionally, in one embodiment of the present application, the acquiring of multi-scene image data corresponding to the target area, the connection point data corresponding to the multi-scene image data, and the digital elevation model data covering the target area, and determining the initial seed image in the multi-scene image data according to the digital elevation model data, the multi-scene image data, and the connection point data, includes: selecting any one scene image data from the multi-scene image data, and determining a pending initial seed image according to the any one scene image data; calculating, based on the digital elevation model data, the connection point residuals and the number of connection points corresponding to the pending initial seed image and the adjacent images of the pending initial seed image; comparing the connection point residuals with a preset residual threshold, and when the connection point residuals are greater than the preset residual threshold and the number of connection points is less than a preset number threshold, determining that the pending initial seed image is the initial seed image.

[0008] Optionally, in one embodiment of the present application, the initial connection point data of the adjacent images corresponding to the initial seed image are extracted, and the relative positioning residuals corresponding to the initial connection point data are calculated, and the residual means and residual variances of the image-side positioning residuals of the initial connection point data in the row and column directions are counted, so as to eliminate the gross error connection points in the initial connection point data based on the residual means and the residual variances, including: determining a set of connection point differences between the initial seed image and the kth image in the multi-scene image data, where k is a positive integer; calculating the residual means and residual variances of the connection points in the connection point difference set in the row and column directions; judging whether the connection point exists outside a preset gross error discrimination range according to the residual means and the residual variance, and if the connection point exists outside the gross error discrimination range, judging the connection point to be the gross error connection point, and eliminating the gross error connection point.

[0009] Optionally, in one embodiment of the present application, the iteration determines a new seed image in the multi-scene image data, and calculates a new relative positioning residual of a non-seed image relative to the initial seed image during each iteration, including: setting the processing status of the initial seed image and all adjacent images of the initial seed image to a processed status, and setting the processing status of other images in the multi-scene image data to an unprocessed status, and constructing a first image group corresponding to the image data in the processed status; setting all adjacent images of the initial seed image as seed images, and selecting any adjacent image from all the adjacent images as a new seed image, and calculating a new relative positioning residual of an image connection point corresponding to an image that is adjacent to the new seed image and has an unprocessed status. And according to the new relative positioning residual, the gross error connection points in the image connection points are eliminated, and the processing status of the new seed image and the adjacent images corresponding to the new seed image are set to the processed status; the next scene image in the clockwise direction of any scene seed image in the first image group is determined, and the relative positioning residual calculation and gross error connection point elimination operation are re-performed on the next scene image to generate a second image group; the non-seed image located at the outermost periphery of the second image group is selected, and the relative positioning residual calculation and gross error connection point elimination operation are iterated according to the non-seed image to generate a new image group, and the image group generation operation is iterated n times until all image data are included in the generated nth image group to obtain the relative positioning residual corresponding to each scene image data, wherein n is a positive integer.

[0010] Optionally, in one embodiment of the present application, based on the new relative positioning residual, a new residual mean and a new residual variance of the non-initial seed image in the multi-scene image data relative to the initial seed image in the row and column directions are determined, and a residual mean space is constructed according to the new residual mean and the new residual variance, and a confidence ellipse corresponding to the residual mean space is calculated, so as to use the confidence ellipse to perform gross error image detection for each scene image data, including: counting the residual mean and residual variance of the relative positioning residual of each scene image data in the target area in the column and row directions, and constructing the residual mean space according to the residual mean; constructing the confidence ellipse according to the residual mean and the residual variance, and judging whether the relative positioning residual corresponding to each scene image data is within the confidence ellipse, wherein, if there is image data whose corresponding relative positioning residual is not within the confidence ellipse, the image data is judged to be a gross error image, and the gross error image is discarded.

[0011] The second aspect of the present application provides a satellite image without a control area network adjustment gross error image detection device, including: an acquisition module, used to acquire multi-scene image data corresponding to a target area, connection point data corresponding to the multi-scene image data and digital elevation model data covering the target area, and determine the initial seed image in the multi-scene image data according to the digital elevation model data, the multi-scene image data and the connection point data; a elimination module, used to extract the initial connection point data of the adjacent images corresponding to the initial seed image, and calculate the relative positioning residual corresponding to the initial connection point data, and count the residual mean and residual variance of the image square positioning residual of the initial connection point data in the row and column directions respectively. , so as to eliminate the gross error connection points in the initial connection point data based on the residual mean and the residual variance; a detection module, used to iteratively determine the new seed image in the multi-scene image data, and calculate the new relative positioning residual of the non-seed image with respect to the initial seed image in each iteration process, and based on the new relative positioning residual, determine the new residual mean and the new residual variance of the non-initial seed image in the multi-scene image data relative to the initial seed image in the row and column directions, and construct a residual mean space according to the new residual mean and the new residual variance, and calculate the confidence ellipse corresponding to the residual mean space, so as to use the confidence ellipse to perform gross error image detection for each scene image data.

[0012] Optionally, in one embodiment of the present application, the acquisition module includes: a first determination unit, used to select any one scene of image data from the multiple scene image data, and determine a pending initial seed image based on the any one scene of image data; a first calculation unit, used to calculate the connection point residuals and the number of connection points corresponding to the pending initial seed image and the adjacent images of the pending initial seed image based on the digital elevation model data; a judgment unit, used to compare the connection point residuals with a preset residual threshold, and when the connection point residuals are greater than the preset residual threshold and the number of connection points is less than a preset number threshold, judge that the pending initial seed image is the initial seed image.

[0013] Optionally, in one embodiment of the present application, the elimination module includes: a second determination unit, used to determine a set of connection point differences between the initial seed image and the kth image in the multi-scene image data, wherein k is a positive integer; a second calculation unit, used to calculate the residual mean and residual variance of the connection points in the connection point difference set in the row direction and the column direction; a first judgment unit, used to judge whether the connection point exists outside a preset gross error judgment range based on the residual mean and the residual variance; if the connection point exists outside the gross error judgment range, the connection point is judged to be the gross error connection point, and the gross error connection point is eliminated.

[0014] Optionally, in one embodiment of the present application, the detection module includes: a construction unit, used to set the processing status of the initial seed image and all adjacent images of the initial seed image to a processed state, and set the processing status of other images in the multi-scene image data to an unprocessed state, and construct a first image group corresponding to the image data in the processed state; a random unit, used to set all adjacent images of the initial seed image as seed images, and select any adjacent image from all the adjacent images as a new seed image, and calculate new relative positioning residuals of image connection points corresponding to images that are adjacent to the new seed image and have an unprocessed processing state, and eliminate gross errors in the image connection points based on the new relative positioning residuals. A connection point is formed, and the processing status of the new seed image and the adjacent image corresponding to the new seed image is set to a processed state; a generation unit is used to determine the next scene image in the clockwise direction of any scene seed image in the first image group, and re-perform relative positioning residual calculation and gross error connection point removal operations on the next scene image to generate a second image group; an iteration unit is used to select a non-seed image located at the outermost periphery of the second image group, and iterate the relative positioning residual calculation and gross error connection point removal operations according to the non-seed image to generate a new image group, and iterate the image group generation operation n times until all image data are included in the generated nth image group to obtain the relative positioning residual corresponding to each scene image data, wherein n is a positive integer.

[0015] Optionally, in one embodiment of the present application, the detection module also includes: a statistical unit, used to count the residual mean and residual variance of the relative positioning residuals in the column and row directions of each scene of image data in the target area, and construct the residual mean space based on the residual mean; a second judgment unit, used to construct the confidence ellipse based on the residual mean and the residual variance, and judge whether the relative positioning residuals corresponding to each scene of image data are within the confidence ellipse, wherein if there is an image data whose corresponding relative positioning residuals are not within the confidence ellipse, the image data is judged to be a gross error image, and the gross error image is discarded.

[0016] The third aspect of the present application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for detecting gross error images of satellite images without control area network adjustment as described in the above embodiment.

[0017] The fourth aspect of the present application provides a computer-readable storage medium, which stores a computer program. When the program is executed by a processor, it implements the above-mentioned satellite image without control area network adjustment gross error image detection method.

[0018] The fifth aspect of the present application provides a computer program product, including a computer program, which is executed to implement the above-mentioned satellite image without control area network adjustment gross error image detection method.

[0019] Therefore, the embodiments of the present application have the following beneficial effects:

[0020] The embodiments of the present application can obtain multi-view image data corresponding to the target area, connection point data corresponding to the multi-view image data, and digital elevation model data covering the target area, and determine the initial seed image in the multi-view image data according to the digital elevation model data, the multi-view image data, and the connection point data; extract the initial connection point data of the adjacent images corresponding to the initial seed image, and calculate the relative positioning residuals corresponding to the initial connection point data, and count the residual mean and residual variance of the image square positioning residuals of the initial connection point data in the row and column directions, so as to obtain the initial seed image based on the residual mean and the residual variance. Eliminate gross error connection points in the initial connection point data; iteratively determine new seed images in the multi-view image data, and calculate the new relative positioning residuals of the non-seed images relative to the initial seed images in each iteration process, and based on the new relative positioning residuals, determine the new residual mean and new residual variance of the non-initial seed images in the multi-view image data relative to the initial seed images in the row and column directions, and construct a residual mean space based on the new residual mean and the new residual variance, and calculate the confidence ellipse corresponding to the residual mean space, so as to use the confidence ellipse to detect gross error images for each scene image data. The present application realizes the detection of gross error images in uncontrolled regional network adjustment by selecting seed images, introducing exogenous digital elevation model data, and constructing a gross error detection model. Thus, it solves the problem that the existing regional network adjustment gross error detection method relies too much on a large number of manual post-inspections for gross error image detection, and cannot detect gross error images before regional network adjustment.

[0021] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0023] Figure 1 A flowchart of a method for detecting gross error images of satellite images without control area network adjustment provided according to an embodiment of the present application;

[0024] Figure 2 A schematic diagram of the execution logic of a method for detecting gross error images of satellite images without control area network adjustment provided by an embodiment of the present application;

[0025] Figure 3 A schematic diagram of seed image growth provided for one embodiment of the present application;

[0026] Figure 4 A schematic diagram of naturalizing a relative positioning error to an initial seed image provided by an embodiment of the present application;

[0027] Figure 5 This is an example diagram of a device for detecting gross errors in satellite image network adjustment without a control area according to an embodiment of the present application;

[0028] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0029] Among them, 10-satellite image without control area network adjustment gross error image detection device; 100-acquisition module, 200-removal module, 300-detection module; 601-memory, 602-processor, 603-communication interface. DETAILED DESCRIPTION

[0030] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0031] The following describes the method and device for detecting gross errors in satellite image network adjustment without control areas according to an embodiment of the present application with reference to the accompanying drawings. In response to the problems mentioned in the above background technology, the present application provides a method for detecting gross errors in satellite image network adjustment without control areas. In the method, multi-scene image data corresponding to the target area, connection point data corresponding to the multi-scene image data, and digital elevation model data covering the target area are obtained, and the initial seed image in the multi-scene image data is determined based on the digital elevation model data, the multi-scene image data, and the connection point data; the initial connection point data of the adjacent images corresponding to the initial seed image are extracted, and the relative positioning residuals corresponding to the initial connection point data are calculated, and the residuals of the image square positioning residuals of the initial connection point data in the row and column directions are counted. Mean and residual variance, based on the residual mean and residual variance, remove the gross error connection points in the initial connection point data; iteratively determine the new seed image in the multi-view image data, and calculate the new relative positioning residual of the non-seed image relative to the initial seed image in each iteration process, and based on the new relative positioning residual, determine the new residual mean and new residual variance of the non-initial seed image in the multi-view image data in the row and column direction relative to the initial seed image, and construct the residual mean space according to the new residual mean and the new residual variance, and calculate the confidence ellipse corresponding to the residual mean space, so as to use the confidence ellipse to detect gross error images for each scene image data. The present application realizes the detection of gross error images in the uncontrolled regional network adjustment by selecting seed images, introducing exogenous digital elevation model data and constructing a gross error detection model. Therefore, it solves the problem that the existing regional network adjustment gross error detection method relies too much on a large number of manual post-inspections for gross error image detection, and cannot detect gross error images before regional network adjustment.

[0032] Specifically, Figure 1 A flowchart of a method for detecting gross error images in satellite imagery without a control area network adjustment provided in an embodiment of the present application.

[0033] like Figure 1 As shown, the satellite image without control area network adjustment gross error image detection method includes the following steps:

[0034] In step S101, multi-view image data corresponding to the target area, connection point data corresponding to the multi-view image data, and digital elevation model (DEM) data covering the target area are obtained, and an initial seed image in the multi-view image data is determined according to the digital elevation model data, the multi-view image data, and the connection point data.

[0035] The embodiment of the present application first needs to prepare a set of image data and its corresponding connection point data for constructing a regional network, as well as a set of digital elevation model data that can cover the regional network range of the target area, such as Figure 2As shown; secondly, the embodiment of the present application also needs to select any image as the initial seed image, and according to the DEM-assisted single-piece space forward intersection, calculate the ground coordinates of the connection points based on the seed image and the adjacent image, and count the ground coordinate differences calculated from the seed image and the adjacent image, and set the quantity threshold and the difference threshold respectively, so as to determine whether the image can be used as a seed image.

[0036] Optionally, in one embodiment of the present application, multi-scene image data corresponding to the target area, connection point data corresponding to the multi-scene image data, and digital elevation model data covering the target area are obtained, and an initial seed image in the multi-scene image data is determined based on the digital elevation model data, the multi-scene image data, and the connection point data, including: selecting any one scene image data from the multi-scene image data, and determining a pending initial seed image based on the any one scene image data; based on the digital elevation model data, calculating the connection point residuals and the number of connection points corresponding to the pending initial seed image and the adjacent images of the pending initial seed image; comparing the connection point residuals with a preset residual threshold, and when the connection point residuals are greater than the preset residual threshold and the number of connection points is less than the preset number threshold, determining that the pending initial seed image is the initial seed image.

[0037] Specifically, the embodiment of the present application can first obtain images of the target area to construct a regional network, and use corresponding matching algorithms and gross error elimination algorithms to obtain reliable connection points on each scene image; in addition, the embodiment of the present application also needs to obtain digital elevation model data that meets the preset accuracy requirements and can cover the entire area.

[0038] After that, the embodiment of the present application further needs to randomly select an image in the center of the region as a seed image, extract all the connection points in the image, and calculate the DEM elevation value of the object coordinates corresponding to the connection points (the elevation regularization translation parameter H of the image can be taken as off ), and the longitude and latitude of the connection point can be directly calculated using the RFM model to obtain the three-dimensional coordinates of the connection point. The RFM model is shown as follows:

[0039]

[0040] It is understandable that since the distance calculated by longitude and latitude is inconvenient for subsequent calculation, the embodiment of the present application can convert it into spatial rectangular coordinates, as shown in the following formula:

[0041] X=proj(P),Y=proj(L)

[0042] Among them, (X, Y) is the spatial rectangular coordinate, (P, L) is the longitude and latitude coordinate, and proj(·) represents the conversion relationship from longitude and latitude coordinate to spatial rectangular coordinate.

[0043] Secondly, the embodiment of the present application can continue to calculate the spatial rectangular coordinates of the ground points of the connection points on the k scene images adjacent to the seed image; if a set of spatial rectangular coordinates of the connection points calculated from the seed image is (X 0k ,Y 0k ,Z 0k ) 0 , then the spatial rectangular coordinate set of the same group of connection points calculated from the k-th adjacent image is (X 0k ,Y 0k ,Z 0k ) k .

[0044] Afterwards, the embodiment of the present application can calculate the residual between the seed image and all adjacent image connection points, and set the residual threshold ∈ res With the quantity threshold ∈ num ; If the residual of the connection point on the seed image is greater than the residual threshold ∈ res The number of connection points is less than the number threshold ∈ num , then the scene image is considered to be a seed image, otherwise the current seed image is discarded, and the above steps are repeated until a suitable initial seed image is selected.

[0045] In step S102, the initial connection point data of the adjacent images corresponding to the initial seed image are extracted, and the relative positioning residuals corresponding to the initial connection point data are calculated, and the residual means and residual variances of the image-side positioning residuals of the initial connection point data in the row and column directions are statistically analyzed, so as to eliminate the gross error connection points in the initial connection point data based on the residual means and residual variances.

[0046] Furthermore, the embodiments of the present application also need to determine the ground coordinates of all connection points located in the initial seed image through DEM-assisted forward intersection, and based on the image positioning model, calculate the image coordinates of these connection points in other adjacent images, and count the residuals between the object coordinates and the image coordinates of the connection points, and the sizes of the residuals are expressed in units of pixels of the initial seed image; in addition, the embodiments of the present application take a scene image as a unit, calculate the mean and variance of the row and column residuals of all connection points on it, and according to the prior condition that the positioning residuals obey the normal distribution, the connection points whose row and column residuals fall outside [μ-3σ,μ+3σ] are regarded as gross errors and are eliminated.

[0047] Optionally, in one embodiment of the present application, initial connection point data of adjacent images corresponding to the initial seed image are extracted, and relative positioning residuals corresponding to the initial connection point data are calculated, and residual means and residual variances of image-side positioning residuals of the initial connection point data in row and column directions are counted, so as to eliminate gross error connection points in the initial connection point data based on the residual means and residual variances, including: determining a set of connection point differences between the initial seed image and the kth image in the multi-scene image data, where k is a positive integer; calculating the residual means and residual variances of the connection points in the connection point difference set in the row and column directions; judging whether there are connection points outside a preset gross error discrimination range based on the residual means and residual variances, and if there are connection points outside the gross error discrimination range, judging the connection points to be gross error connection points, and eliminating the gross error connection points.

[0048] It should be noted that after determining the seed image, the embodiment of the present application further needs to extract all connection points located on the seed image and its adjacent images, which can be expressed in the following point set form:

[0049] {{x,y}|{x 0k ,y 0k} k , k=1,2,…,n} and {{x,y}|{x 0k ,y 0k} 0 ,k=1,2,…,n}

[0050] Among them, {x 0k ,y 0k} k represents the image coordinates of the connection point on the kth image, {x 0k ,y 0k} 0 It represents the image coordinates of the connection point on the seed image, and n represents the number of adjacent images.

[0051] Based on the DEM-assisted spatial forward intersection, the present embodiment can uniquely determine the ground coordinates of a pair of connection points. The specific calculation steps are as follows:

[0052] 1. Set the initial ground elevation value of the pair of connection points It is usually equal to the regularized translation parameter of the RFM model elevation The corresponding spatial rectangular coordinates;

[0053] 2. Connect the coordinates of the points on the seed image (x 0k ,y 0k ) 0 And the RFM of the seed image, the spatial rectangular coordinates of the connection point can be calculated by spatial forward intersection And according to the plane coordinates Interpolate an elevation value from the corresponding position of the DEM

[0054] 3. Calculate the elevation value The rectangular coordinates of the connection point space The difference between ΔZ i1 ; If the difference is less than the given threshold δ z , it means that the spatial rectangular coordinate calculation of the connection point is completed, otherwise repeat the above steps until the difference is less than the corresponding threshold;

[0055] 4. Repeat the above steps until the spatial rectangular coordinates {X 0k ,Y 0k ,Z 0k} k All calculations completed.

[0056] At this time, the embodiment of the present application can obtain a set of spatial rectangular coordinate points, as shown below:

[0057] {{X,Y,Z}|{X 0k ,Y 0k ,Z 0k} k ,k=1,2,…,n}

[0058] In the embodiment of the present application, RFM can be used to back-project the point set onto the k-th adjacent image, and its coordinates {{x, y}|{x proj0k ,y proj0k} k , k=1,2,…,n}, and subtract it from the corresponding original matching point to obtain the corresponding difference value set {{Δx,Δy}|{Δx 0k ,Δy 0k} k ,k=1,2,…,n}; at this time, the element {Δx 0k ,Δy 0k} k That is, the relative positioning accuracy of the connection point of the kth image relative to the corresponding connection point of the seed image.

[0059] In a regional network, the resolution of each image is usually different. Therefore, for the convenience of subsequent statistics, the relative positioning accuracy of all non-seed image connection points is expressed in pixels of the initial seed image. Assume that the pixel size of the initial seed image is δ0, and the pixel size of the kth adjacent image is δ k , then the relative positioning accuracy of the tie point of the kth scene image relative to the corresponding tie point of the seed image can be converted to the pixel of the seed image by the following formula:

[0060]

[0061] At this time, the relative positioning accuracy of the connection points can be obtained by the point set express.

[0062] In the actual implementation process, even if the gross errors are eliminated after matching, there is still a high probability that there are gross errors in the connection points. Therefore, the embodiment of the present application can construct a statistical model by counting the mean and variance of the back-projection errors of the connection points between the seed image and the adjacent images of each scene to eliminate the gross errors therein. The specific steps are as follows:

[0063] 1. Extract the connection point difference set {Δx 0k ,Δy 0k} k ;

[0064] 2. Calculate the mean and variance of the difference between the row and column directions of the group of connection points. The calculation method is shown in the following formula:

[0065]

[0066] Among them, μ x0k represents the mean of column-wise differences, represents the variance of the column direction difference, n is the number of connection points; and the mean μ of the row direction difference y0k and standard deviation The calculation method is the same as that in the row direction;

[0067] 3. Except for a small amount of gross errors, the back-projection error is a random error and follows a Gaussian distribution. According to statistical laws, if the column direction difference Δx of a certain connection point 0k Falling In addition, or the row direction difference Δy 0k Falling If it is not, it is determined to be a gross error connection point and is removed;

[0068] 4. Repeat the above steps until the connection point difference set {Δx 0k ,Δy 0k} k All the gross errors in the are eliminated;

[0069] 5. Repeat the above steps until the gross connection points between the seed image and all adjacent images are eliminated.

[0070] Therefore, the embodiments of the present application provide reliable data and theoretical support for the subsequent elimination of gross error images through the connection point relative positioning residual calculation and single-view image gross error point elimination operations.

[0071] In step S103, a new seed image in the multi-scene image data is iteratively determined, and a new relative positioning residual of the non-seed image with respect to the initial seed image is calculated in each iteration process. Based on the new relative positioning residual, a new residual mean and a new residual variance of the non-initial seed image in the multi-scene image data in the row and column directions relative to the initial seed image are determined, and a residual mean space is constructed according to the new residual mean and the new residual variance, and a confidence ellipse corresponding to the residual mean space is calculated, so as to use the confidence ellipse to perform gross error image detection for each scene image data.

[0072] Furthermore, after completing the calculation of the relative positioning residuals of the connection points between the initial seed image and the adjacent images and eliminating the gross error points, the embodiments of the present application can also continuously select new seed images based on the neighborhood operation method, and calculate the relative positioning residuals between all non-seed images, and finally convert them into the relative positioning residuals of the non-seed images relative to the initial seed image.

[0073] Afterwards, since it cannot be guaranteed that the RFM of each seed image does not contain gross errors, after calculating the relative positioning residuals of all images in the regional network relative to the seed image, the embodiment of the present application needs to normalize the benchmark of the relative positioning residuals of all images to the initial seed image, and on this basis calculate the mean of the relative positioning residuals in the row and column directions of all non-initial seed image connection points in the regional network, and construct a residual mean space to detect gross error images through confidence ellipses.

[0074] Therefore, the embodiments of the present application introduce exogenous digital elevation model data to quantitatively express the relative geometric positioning accuracy and changes of satellite images in the regional network, and distinguish between connection point gross errors and gross error images through scene-constrained gross error point detection, and under the constraint of the initial seed image, the relative positioning accuracy between images is converted into the relative positioning accuracy relative to the reference image (i.e., the initial seed image), thereby realizing the detection of gross error images.

[0075] Optionally, in one embodiment of the present application, iteratively determining a new seed image in the multi-view image data, and calculating a new relative positioning residual of the non-seed image relative to the initial seed image in each iteration process, includes: setting the processing status of the initial seed image and all adjacent images of the initial seed image to a processed state, and setting the processing status of other images in the multi-view image data to an unprocessed state, and constructing a first image group corresponding to the image data in the processed state; setting all adjacent images of the initial seed image as seed images, and selecting any adjacent image from all adjacent images as a new seed image, and calculating a new relative positioning residual of an image connection point corresponding to an image that is adjacent to the new seed image and has an unprocessed processing state, And according to the new relative positioning residual, the gross error connection points in the image connection points are eliminated, and the processing status of the new seed image and the adjacent images corresponding to the new seed image is set to the processed state; the next scene image in the clockwise direction of any scene seed image in the first image group is determined, and the relative positioning residual calculation and gross error connection point elimination operation are re-performed on the next scene image to generate a second image group; the non-seed image located at the outermost periphery of the second image group is selected, and the relative positioning residual calculation and gross error connection point elimination operation are iterated according to the non-seed image to generate a new image group, and the image group generation operation is iterated n times until all image data are included in the generated nth image group to obtain the relative positioning residual corresponding to each scene image data, wherein n is a positive integer.

[0076] In order to obtain the relative positioning accuracy of all non-seed images in the regional network relative to the initial seed image, the embodiment of the present application can set the adjacent image adjacent to the initial seed image as a new seed image based on the morphological seed point growth algorithm, and continue to perform the connection point relative positioning residual calculation and the single-view image coarse error point elimination operation; each subsequent iteration uses the outermost image as the new seed image, re-calculates the connection point relative positioning residual and re-eliminates the single-view image coarse error point operation until the relative positioning accuracy of all images is calculated, such as Figure 3 As shown in the figure, the specific steps of the seed image generation strategy are as follows:

[0077] Step 1: After the initial seed image and its adjacent images are connected with point relative positioning residual calculation and single-view image coarse error point elimination operation, the embodiment of the present application can set the processing status flag=1 of the above-mentioned images, and the status of the remaining unprocessed images is flag=0; at this time, the processed images form an image group {I} i , i represents the i-th image group, which is 1 at this time;

[0078] Step 2: Set all adjacent images as seed images, and randomly select a seed image. Calculate the relative positioning errors of the image connection points adjacent to it and with the processing status of flag = 0 through the connection point relative positioning residual calculation and single-view image coarse error point elimination operation, and complete the elimination of coarse error connection points;

[0079] Step 3: After the processing of a seed image and its adjacent images is completed, update its processing status to flag = 1; according to the regularized translation parameter (B off ,L off ,H off ), select the next scene image in the clockwise direction of the seed image, and complete the processing of the remaining seed images according to step 2. At this time, a new image group {I} is formed. i+1 ;

[0080] Step 4: After completing the calculation of all seed images and their adjacent images in this iteration, continue to use the image RFM regularization translation parameter (B off ,L off ,H off ) Select the non-seed image at the outermost edge of the image group as the seed image for the next round of iteration.

[0081] Therefore, the embodiment of the present application can repeat the above steps until the image group {I} n includes all images in the regional network (n is the number of iterations), thereby completing the relative positioning accuracy calculation of each scene image in the regional network.

[0082] Optionally, in one embodiment of the present application, based on the new relative positioning residuals, a new residual mean and a new residual variance of the non-initial seed images in the multi-scene image data relative to the initial seed image in the row and column directions are determined, and a residual mean space is constructed based on the new residual mean and the new residual variance, and a confidence ellipse corresponding to the residual mean space is calculated, so as to use the confidence ellipse to perform gross error image detection for each scene of image data, including: counting the residual mean and residual variance of the relative positioning residuals of each scene of image data in the column and row directions in the target area, and constructing a residual mean space based on the residual mean; constructing a confidence ellipse based on the residual mean and the residual variance, and judging whether the relative positioning residual corresponding to each scene of image data is within the confidence ellipse, wherein, if there is image data whose relative positioning residual is not within the confidence ellipse, the image data is judged to be a gross error image, and the gross error image is eliminated.

[0083] After the seed image generation process is completed, the relative positioning residual {Δx jk ,Δy jk} k , where j represents the seed image and k represents the adjacent non-seed image.

[0084] According to the residual set, the embodiment of the present application can calculate the relative positioning residual between the non-seed image and the adjacent seed image, which is expressed as the mean of the residual in the row direction and the column direction, and the calculation method is as follows:

[0085]

[0086] in, Respectively represent the mean of the relative positioning residuals between the non-seed image and the adjacent seed image in the column and row directions.

[0087] Since most of the images in the regional network are not directly adjacent to the initial seed image, their relative positioning benchmarks need to be unified and naturalized to the initial seed image, such as Figure 4 Assume that a scene image in the regional network is the kth scene starting from the initial image, then its relative positioning accuracy in the column and row directions relative to the initial seed image is δ x0k , δ y0k Can be expressed as:

[0088]

[0089] Among them, δ x(i-1,i) and δ y(i-1,i) Represents the relative positioning residual in the column and row directions between the current image and its adjacent seed images.

[0090] According to the mean of the relative positioning residuals based on the initial seed image, the embodiment of the present application can construct relevant statistics to calculate the confidence ellipse to eliminate the gross error image. The specific implementation steps are as follows:

[0091] 1. Count the mean of the relative positioning residuals of each scene image in the regional network after normalization in the column and row directions and variance And construct the residual mean space according to the normalized relative positioning residual mean;

[0092] 2. Based on the above two statistics, a biased error ellipse is constructed. The mathematical expression of the error ellipse model is as follows:

[0093]

[0094] 3. Due to It follows a normal distribution, so the statistic constructed from it It obeys the chi-square distribution, and the embodiment of the present application believes that due to The confidence interval is 95%, at this time s = 5.991 ≈ 6, so if the coordinates of the normalized relative positioning residual in the residual space fall outside the range determined by the following formula, the image is considered to be a gross error image and should be eliminated:

[0095]

[0096] Therefore, the embodiments of the present application can automatically and accurately identify possible gross error images before regional network adjustment, avoiding the problem of manually screening gross error images after adjustment, which consumes a lot of time and manpower costs. In addition, external digital elevation model data is introduced during gross error detection, and seed image growth is performed based on the idea of ​​neighborhood operation, which greatly reduces the amount of calculation and improves the speed and efficiency of regional network gross error image detection. In addition, the embodiments of the present application perform gross error detection at the connection point level and the image level based on statistical principles, thereby ensuring the accuracy and reliability of the gross error detection results.

[0097] In summary, the embodiment of the present application fully considers the terrain information corresponding to the regional network, and based on the RFM model, performs spatial forward intersection of the connection points on each scene image before free network adjustment taking into account the elevation information, thereby obtaining the ground coordinates of the connection points on each scene image before the regional network adjustment. In addition, since the embodiment of the present application incorporates the elevation information for forward intersection, it can guarantee the accuracy of the connection point positioning result to a large extent. At the same time, the embodiment of the present application fully considers the spatial statistical information of the connection points, and performs gross error detection at the connection point level and the image level respectively, and by setting the seed image, the image square positioning residuals between adjacent image connection points are counted in the form of seed point growth; afterwards, the embodiment of the present application realizes gross error detection and elimination at the connection point level by constructing an affine transformation model based on the residual. On this basis, the embodiment of the present application can count the mean of the positioning residuals of the connection points of each image in the row and column directions, and determine the residual space accordingly; finally, the embodiment of the present application can calculate the confidence ellipse under the space, mark the positioning residuals that fall outside the ellipse as gross errors, and the corresponding images as gross error images, thereby realizing step-by-step automatic detection of gross error images in the regional network in a seed image growth manner based on the terrain-related spatial forward intersection method.

[0098] According to the satellite image uncontrolled area network adjustment gross error image detection method proposed in the embodiment of the present application, by obtaining multi-view image data corresponding to the target area, connection point data corresponding to the multi-view image data and digital elevation model data covering the target area, and determining the initial seed image in the multi-view image data according to the digital elevation model data, the multi-view image data and the connection point data; extracting the initial connection point data of the adjacent images corresponding to the initial seed image, and calculating the relative positioning residuals corresponding to the initial connection point data, and statistically calculating the residual mean and residual variance of the image square positioning residuals of the initial connection point data in the row and column directions, so as to obtain the initial seed image of the satellite image uncontrolled area network adjustment gross error image detection method according to the embodiment of the present application, and obtaining the initial seed image of the satellite image uncontrolled area network adjustment gross error image detection method according to the embodiment of the present application, and obtaining the initial seed image of the satellite image uncontrolled area network adjustment gross error image detection method according to the embodiment of the present application, Based on the residual mean and residual variance, the gross error connection points in the initial connection point data are eliminated; the new seed images in the multi-view image data are iteratively determined, and the new relative positioning residuals of the non-seed images relative to the initial seed images are calculated in each iteration process, and based on the new relative positioning residuals, the new residual means and new residual variances of the non-initial seed images in the multi-view image data relative to the initial seed images in the row and column directions are determined, and the residual mean space is constructed according to the new residual mean and the new residual variance, and the confidence ellipse corresponding to the residual mean space is calculated, so as to use the confidence ellipse to detect gross error images for each view image data. This application realizes the detection of gross error images in the adjustment of the uncontrolled area network by selecting seed images, introducing exogenous digital elevation model data and constructing a gross error detection model.

[0099] Next, a device for detecting gross errors in satellite image network adjustment without control area proposed in an embodiment of the present application will be described with reference to the accompanying drawings.

[0100] Figure 5 It is a block diagram of a device for detecting gross errors in satellite image network adjustment without control area according to an embodiment of the present application.

[0101] like Figure 5 As shown, the satellite image without control area network adjustment gross error image detection device 10 includes: an acquisition module 100, a removal module 200 and a detection module 300.

[0102] Among them, the acquisition module 100 is used to obtain the multi-view image data corresponding to the target area, the connection point data corresponding to the multi-view image data and the digital elevation model data covering the target area, and determine the initial seed image in the multi-view image data according to the digital elevation model data, the multi-view image data and the connection point data.

[0103] The elimination module 200 is used to extract the initial connection point data of the adjacent images corresponding to the initial seed image, calculate the relative positioning residuals corresponding to the initial connection point data, and count the residual mean and residual variance of the image-side positioning residuals of the initial connection point data in the row and column directions, so as to eliminate the gross error connection points in the initial connection point data based on the residual mean and residual variance.

[0104] The detection module 300 is used to iteratively determine a new seed image in the multi-scene image data, and calculate a new relative positioning residual of the non-seed image relative to the initial seed image in each iteration process, and based on the new relative positioning residual, determine a new residual mean and a new residual variance of the non-initial seed image in the multi-scene image data in the row and column directions relative to the initial seed image, and construct a residual mean space based on the new residual mean and the new residual variance, and calculate a confidence ellipse corresponding to the residual mean space, so as to use the confidence ellipse to perform gross error image detection for each scene image data.

[0105] Optionally, in one embodiment of the present application, the acquisition module 100 includes: a first determination unit, a first calculation unit and a determination unit.

[0106] The first determining unit is used to select any one scene of image data from the multiple scene image data, and determine the to-be-determined initial seed image according to the any one scene of image data.

[0107] The first calculation unit is used to calculate the connection point residuals and the number of connection points corresponding to the pending initial seed image and the adjacent images of the pending initial seed image based on the digital elevation model data.

[0108] The determination unit is used to compare the connection point residual with a preset residual threshold, and when the connection point residual is greater than the preset residual threshold and the number of connection points is less than the preset number threshold, determine that the pending initial seed image is the initial seed image.

[0109] Optionally, in one embodiment of the present application, the elimination module 200 includes: a second determination unit, a second calculation unit and a first judgment unit.

[0110] The second determining unit is used to determine a set of connection point difference values ​​between the initial seed image and the k-th scene image in the multi-scene image data, wherein k is a positive integer.

[0111] The second calculation unit is used to calculate the residual mean and residual variance of the connection points in the connection point difference set in the row direction and the column direction.

[0112] The first judgment unit is used to judge whether there is a connection point outside the preset gross error judgment range according to the residual mean and the residual variance. If there is a connection point outside the gross error judgment range, the connection point is judged to be a gross error connection point and the gross error connection point is eliminated.

[0113] Optionally, in one embodiment of the present application, the detection module 300 includes: a construction unit, a random unit, a generation unit and an iteration unit.

[0114] The construction unit is used to set the processing status of the initial seed image and all adjacent images of the initial seed image to a processed state, set the processing status of other images in the multi-view image data to an unprocessed state, and construct a first image group corresponding to the processed image data.

[0115] The random unit is used to set all adjacent images of the initial seed image as seed images, select any adjacent image of the scene from all adjacent images as a new seed image, calculate the new relative positioning residuals of the image connection points corresponding to the images that are adjacent to the new seed image and are in an unprocessed state, eliminate the gross error connection points among the image connection points according to the new relative positioning residuals, and set the processing states of the new seed image and the adjacent images corresponding to the new seed image to the processed state.

[0116] The generating unit is used to determine the next image in the clockwise direction of any seed image in the first image group, and recalculate the relative positioning residual and remove the gross error connection points on the next image to generate the second image group.

[0117] The iterative unit is used to select a non-seed image located at the outermost periphery of the second image group, and iterate the relative positioning residual calculation and the gross error connection point elimination operation based on the non-seed image to generate a new image group, and iterate the image group generation operation n times until the generated n-th image group contains all the image data, so as to obtain the relative positioning residual corresponding to each scene of image data, wherein n is a positive integer.

[0118] Optionally, in an embodiment of the present application, the detection module 300 further includes: a statistical unit and a second judgment unit.

[0119] The statistical unit is used to count the residual mean and residual variance of the relative positioning residual of each scene image data in the target area in the column direction and the row direction, and construct the residual mean space according to the residual mean.

[0120] The second judgment unit is used to construct a confidence ellipse according to the residual mean and the residual variance, and judge whether the relative positioning residual corresponding to each scene image data is within the confidence ellipse. If the relative positioning residual corresponding to the image data is not within the confidence ellipse, the image data is judged to be a gross error image, and the gross error image is discarded.

[0121] It should be noted that the above explanation of the embodiment of the method for detecting gross errors in satellite image network adjustment without control area is also applicable to the device for detecting gross errors in satellite image network adjustment without control area of ​​this embodiment, and will not be repeated here.

[0122] The satellite image non-control area network adjustment gross error image detection device proposed in the embodiment of the present application includes an acquisition module for acquiring multi-view image data corresponding to the target area, connection point data corresponding to the multi-view image data, and digital elevation model data covering the target area, and determining the initial seed image in the multi-view image data according to the digital elevation model data, the multi-view image data, and the connection point data; a elimination module for extracting the initial connection point data of the adjacent images corresponding to the initial seed image, and calculating the relative positioning residuals corresponding to the initial connection point data, and statistically analyzing the residual mean and residual square of the image square positioning residuals of the initial connection point data in the row and column directions respectively. The difference is to remove the gross error connection points in the initial connection point data based on the residual mean and residual variance; the detection module is used to iteratively determine the new seed image in the multi-view image data, and calculate the new relative positioning residual of the non-seed image relative to the initial seed image in each iteration process, and based on the new relative positioning residual, determine the new residual mean and new residual variance of the non-initial seed image in the multi-view image data in the row and column directions relative to the initial seed image, and construct the residual mean space according to the new residual mean and the new residual variance, and calculate the confidence ellipse corresponding to the residual mean space, so as to use the confidence ellipse to detect gross error images for each scene image data. The present application realizes the detection of gross error images in the uncontrolled area network adjustment by selecting seed images, introducing exogenous digital elevation model data and constructing a gross error detection model.

[0123] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0124] A memory 601 , a processor 602 , and a computer program stored in the memory 601 and executable on the processor 602 .

[0125] When the processor 602 executes the program, the method for detecting gross error images in satellite image network adjustment without control area provided in the above embodiment is implemented.

[0126] Furthermore, the electronic device further comprises:

[0127] The communication interface 603 is used for communication between the memory 601 and the processor 602 .

[0128] The memory 601 is used to store computer programs that can be executed on the processor 602 .

[0129] The memory 601 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0130] If the memory 601, the processor 602 and the communication interface 603 are implemented independently, the communication interface 603, the memory 601 and the processor 602 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0131] Optionally, in a specific implementation, if the memory 601, the processor 602 and the communication interface 603 are integrated on a chip, the memory 601, the processor 602 and the communication interface 603 can communicate with each other through an internal interface.

[0132] The processor 602 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0133] The embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned method for detecting gross error images of satellite images without control area network adjustment.

[0134] The embodiment of the present application also provides a computer program product, including a computer program, which, when executed, is used to implement the above-mentioned satellite image without control area network adjustment gross error image detection method.

[0135] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0136] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0137] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0138] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or N wirings (electronic devices), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways as necessary and then storing it in a computer memory.

[0139] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0140] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0141] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0142] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for detecting gross error images of satellite images without control area network adjustment, characterized in that: The following steps are involved: Acquire multi-view image data corresponding to a target area, connection point data corresponding to the multi-view image data, and digital elevation model data covering the target area, and determine an initial seed image in the multi-view image data according to the digital elevation model data, the multi-view image data, and the connection point data; Extracting initial connection point data of adjacent images corresponding to the initial seed image, calculating relative positioning residuals corresponding to the initial connection point data, and counting residual means and residual variances of image-side positioning residuals of the initial connection point data in row and column directions, so as to eliminate gross error connection points in the initial connection point data based on the residual means and the residual variances; Iteratively determine a new seed image in the multi-scene image data, and calculate a new relative positioning residual of a non-seed image with respect to the initial seed image in each iteration process, and based on the new relative positioning residual, determine a new residual mean and a new residual variance of the non-initial seed image in the multi-scene image data in row and column directions relative to the initial seed image, and construct a residual mean space according to the new residual mean and the new residual variance, and calculate a confidence ellipse corresponding to the residual mean space, so as to use the confidence ellipse to perform gross error image detection for each scene image data.

2. The method according to claim 1, characterized in that The step of acquiring multi-view image data corresponding to a target area, connection point data corresponding to the multi-view image data, and digital elevation model data covering the target area, and determining an initial seed image in the multi-view image data according to the digital elevation model data, the multi-view image data, and the connection point data, includes: Selecting any one scene of image data from the multiple scenes of image data, and determining a pending initial seed image according to the any one scene of image data; Based on the digital elevation model data, calculating the connection point residuals and the number of connection points corresponding to the pending initial seed image and the adjacent images of the pending initial seed image; The connection point residual is compared with a preset residual threshold, and when the connection point residual is greater than the preset residual threshold and the number of connection points is less than a preset number threshold, the pending initial seed image is determined to be the initial seed image.

3. The method according to claim 2, characterized in that The extracting of the initial connection point data of the adjacent images corresponding to the initial seed image, calculating the relative positioning residuals corresponding to the initial connection point data, and counting the residual mean and residual variance of the image-side positioning residuals of the initial connection point data in the row and column directions, so as to eliminate the gross error connection points in the initial connection point data based on the residual mean and the residual variance, includes: Determine a set of connection point difference values ​​between the initial seed image and the k-th image in the multi-image data, where k is a positive integer; Calculating the residual mean and residual variance of the connection points in the connection point difference set in the row direction and the column direction; It is determined whether the connection point exists outside the preset gross error determination range according to the residual mean and the residual variance; if the connection point exists outside the gross error determination range, the connection point is determined to be the gross error connection point and the gross error connection point is eliminated.

4. The method according to claim 3, characterized in that: The iterative determination of a new seed image in the multi-view image data and calculation of a new relative positioning residual of a non-seed image relative to the initial seed image in each iteration process include: Setting the processing status of the initial seed image and all adjacent images of the initial seed image to a processed status, setting the processing status of other images in the multi-view image data to an unprocessed status, and constructing a first image group corresponding to the image data in the processed status; All adjacent images of the initial seed image are set as seed images, and any adjacent image is selected from all the adjacent images as a new seed image, and new relative positioning residuals of image connection points corresponding to images adjacent to the new seed image and in an unprocessed state are calculated, and gross error connection points among the image connection points are eliminated according to the new relative positioning residuals, and the processing states of the new seed image and the adjacent images corresponding to the new seed image are set to a processed state; Determine the next scene image in the clockwise direction of any scene seed image in the first image group, and re-calculate the relative positioning residual and remove the gross error connection points for the next scene image to generate a second image group; A non-seed image located at the outermost periphery of the second image group is selected, and a new image group is generated by iterative relative positioning residual calculation and gross error connection point elimination operations based on the non-seed image, and the image group generation operation is iterated n times until the generated nth image group contains all image data, so as to obtain the relative positioning residual corresponding to each scene of image data, wherein n is a positive integer.

5. The method according to claim 4, characterized in that The method further comprises: determining a new residual mean and a new residual variance of the non-initial seed image in the multi-view image data relative to the initial seed image in the row and column directions based on the new relative positioning residual, constructing a residual mean space according to the new residual mean and the new residual variance, and calculating a confidence ellipse corresponding to the residual mean space, so as to perform gross error image detection on each view image data using the confidence ellipse, including: Counting the residual mean and residual variance of the relative positioning residual of each scene image data in the target area in the column direction and the row direction, and constructing the residual mean space according to the residual mean; The confidence ellipse is constructed according to the residual mean and the residual variance, and it is determined whether the relative positioning residual corresponding to each scene of the image data is within the confidence ellipse. If there is an image data whose relative positioning residual is not within the confidence ellipse, the image data is determined to be a gross error image, and the gross error image is discarded.

6. A satellite image without control area network adjustment gross error image detection device, characterized in that: include: an acquisition module, used to acquire multi-view image data corresponding to a target area, connection point data corresponding to the multi-view image data, and digital elevation model data covering the target area, and determine an initial seed image in the multi-view image data according to the digital elevation model data, the multi-view image data, and the connection point data; A removal module is used to extract the initial connection point data of the adjacent images corresponding to the initial seed image, calculate the relative positioning residuals corresponding to the initial connection point data, and count the residual means and residual variances of the image-side positioning residuals of the initial connection point data in the row and column directions, so as to remove the gross error connection points in the initial connection point data based on the residual means and the residual variance; A detection module is used to iteratively determine a new seed image in the multi-scene image data, and calculate a new relative positioning residual of a non-seed image relative to the initial seed image in each iteration process, and based on the new relative positioning residual, determine a new residual mean and a new residual variance of the non-initial seed image in the multi-scene image data relative to the initial seed image in the row and column directions, and construct a residual mean space according to the new residual mean and the new residual variance, and calculate a confidence ellipse corresponding to the residual mean space, so as to use the confidence ellipse to perform gross error image detection on each scene image data.

7. The device according to claim 6, characterized in that The acquisition module comprises: A first determining unit, configured to select any one scene of image data from the multiple scene image data, and determine a pending initial seed image according to the any one scene of image data; A first calculation unit is used to calculate the connection point residuals and the number of connection points corresponding to the undetermined initial seed image and the adjacent images of the undetermined initial seed image based on the digital elevation model data; A determination unit is used to compare the connection point residual with a preset residual threshold, and when the connection point residual is greater than the preset residual threshold and the number of connection points is less than a preset number threshold, determine that the pending initial seed image is the initial seed image.

8. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for detecting gross error images of satellite images without control area network adjustment as described in any one of claims 1 to 5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the method for detecting gross error images of satellite images without control area network adjustment as described in any one of claims 1 to 5.

10. A computer program product, comprising a computer program, characterized in that The computer program is executed to implement the method for detecting gross error images in satellite image network adjustment without control area as described in any one of claims 1 to 5.

Images