Double-linear-array image uncontrolled positioning method and device based on laser elevation point

Abstract

A double-linear-array image uncontrolled positioning method based on laser elevation points belongs to the field of satellite image processing, and is characterized by comprising the following steps: acquiring image connection points through double-linear-array image matching; the laser elevation point is automatically stabbed, the laser elevation point is stabbed to the double-linear-array image, and a laser connection point is obtained; the encrypted point and the exterior orientation after adjustment are obtained through adjustment of the adaptive piecewise polynomial strict model; and performing rigorous model adjustment processing based on laser elevation point assistance by using the laser connection point, correcting the elevation of the encryption point, and outputting a final exterior orientation and an encryption point file. For a strip image in which a continuous adjustment model cannot be constructed, adaptive segmentation processing can be realized, and finally adaptive rigorous model adjustment processing of the strip image is realized. And meanwhile, elevation control is realized by adopting laser elevation point assisted rigorous model adjustment, laser elevation point assisted rigorous model adjustment processing is realized, and the stripe image elevation positioning precision is improved.

Description

Technical Field

[0001] This invention belongs to the field of satellite image processing, and particularly relates to a method and apparatus for uncontrolled positioning of dual-linear array images based on laser elevation points. Background Technology

[0002] The rapid development of high-resolution remote sensing and mapping satellite technology has brought new development opportunities to the surveying and mapping geographic information industry. With the rapid development of society and economy, the traditional technologies and methods for producing new geographic products can no longer fully meet the needs of my country's new generation of surveying and mapping informatization construction. Supported by high-resolution remote sensing satellites, how to adopt new technologies and utilize multi-payload data to produce surveying and mapping geographic information products of various scales and achieve high-precision uncontrolled positioning processing of stereoscopic surveying and mapping satellites is an urgent problem to be solved in current remote sensing satellite processing.

[0003] Rigorous model adjustment of satellite imagery is a crucial step in achieving high-precision positioning processing for remote sensing satellites. Traditional rigorous model adjustment mainly uses exterior orientation models such as polynomials, piecewise polynomials, and orientation patches. However, the applicability of these adjustment models decreases for large-area, long strip data. For cloud, water, and anomaly data, manual intervention in the adjustment algorithm for grouped adjustment is required. Furthermore, addressing the challenge of establishing and measuring elevation control points in complex terrain areas such as forests, deserts, mountains, and glaciers, and finding a way to replace traditional field elevation control point acquisition with laser altimetry data to achieve accurate elevation control from satellite imagery, is also a pressing issue that needs to be resolved. Summary of the Invention

[0004] The present invention aims to solve the above problems and provides a method and apparatus for uncontrolled positioning of dual linear array images based on laser elevation points.

[0005] In a first aspect, the present invention provides a method for uncontrolled positioning of dual-linear array images based on laser elevation points, comprising the following steps: Step 1: Obtain image connection points through dual-line array image matching; Step 2: Automatically rotate the laser elevation point onto the dual-line array image to obtain the laser connection point; Step 3: Obtain the adjusted densification points and exterior orientations through adaptive piecewise polynomial rigorous model adjustment; Step 4: Perform rigorous model adjustment based on laser elevation points using laser tie points to correct the elevation of the densified points, and output the final exterior orientation and densified point files.

[0006] Furthermore, the uncontrolled positioning method for dual-linear array images based on laser elevation points described in this invention includes the following steps in step 1: Dual-linear array image matching. Step 11: Select the rear view image as the reference image and the front view image as the image to be matched; Step 12: Perform regular gridding on the reference image and extract Moravec corner points in the gridded area; Step 13: Using the image imaging model, traverse and calculate the overlapping area between the image to be matched and the reference image. If the reference image and the image to be matched do not overlap, continue this step. Step 14: Based on the imaging model, predict the corner points obtained from the reference image onto the image to be matched in the overlapping area; Step 15: Check whether the resolution of the reference image and the image to be matched are consistent. If they are inconsistent, perform pyramid resolution consistency processing on the reference image and the image to be matched. Step 16: Perform correlation coefficient matching on the set of corner points corresponding to the acquired reference image and the image to be matched. If the correlation coefficient is greater than the threshold, perform least squares matching. Otherwise, repeat this step for the next corner point of the reference image. Step 17: Record the precise points obtained by least squares image matching. Repeat steps 15-16 for the next corner point of the reference image until all corner point sets of the reference image have been processed. Step 18: Use the RANSAC algorithm and forward intersection of corresponding points to remove gross errors from the obtained matching point set. At the same time, extract water points based on the global water mask and extract gross errors from cloud and snow and error data based on the matching threshold. Step 19: Mark the reference image as a matched image, then select the reference image from the images to be matched, and perform the processing steps 12-18 until the image set processing is complete.

[0007] Furthermore, in the uncontrolled positioning method for dual-linear array images based on laser elevation points described in this invention, step 2, the automatic laser elevation point rotation, includes the following steps: Step 21: Based on the longitude, latitude, and elevation values ​​of the laser elevation point, as well as the orientation parameters of the stereo image, rotate the center position of the footprint point to the rear view image to obtain the initial image point coordinates of the rear view image. Step 22: Using the initial image point coordinates as the center, extract the feature point set of the rear view image using the Fronster operator according to the preset spot size; Step 23: Select the feature points to be matched from the feature point set, and use the orientation parameters attached to the dual linear array image to predict the initial position of the corresponding points of the feature points. Step 24: Calculate the correlation coefficient of the image points to be matched. If the correlation coefficient is greater than the threshold, perform least squares point refinement processing; otherwise, calculate the next feature point of the reference image and repeat steps 23-24. Step 25: Record the matching point pairs after least squares processing; select the next feature point of the reference image and execute steps 23-25 ​​until all feature points of the reference image have been processed. Step 26: After removing erroneous points, output the matching results; Step 27: Mark the reference image as a matched image, then select the reference image from the unmatched images, and repeat steps 22-27 until the image set processing is complete.

[0008] Furthermore, the uncontrolled positioning method based on dual-linear array images using laser elevation points described in this invention, step 3 of which involves adaptive piecewise polynomial rigorous model adjustment, includes the following steps: Step 31: Generate adaptive adjustment units based on row time data and tie points; Step 32: Start processing a single adjustment unit. Based on the rigorous adjustment model of satellite imagery, fit the original attitude data using a polynomial to obtain the initial values ​​of the piecewise polynomial model parameters. Step 33: Using satellite stereo image pairs as units, calculate the object coordinates of the connection points through forward intersection in space based on the initial attitude and orbital parameters; Step 34: Establish the error equation and solve for the attitude polynomial coefficient correction values ​​of each view image by spatial resection. Step 35: Iterate through front intersection and back intersection until the residuals converge; Step 36: Process all adjustment units according to steps 32-35 until all adjustment units have been processed.

[0009] Furthermore, in the uncontrolled positioning method for dual-linear array images based on laser elevation points described in this invention, step 31, the generation of the adaptive adjustment unit, includes the following steps: Step 311: Read the foreground and background timelines of all scene images, sort them from smallest to largest according to the timeline timestamp, and form a scene list data; Step 312: Based on the sorted scene list data, sequentially read the scene image connection point files; Step 313: Determine the number of image connection points in step 312. If it is less than the threshold, the strip adjustment unit is interrupted and proceed to the next step; otherwise, put the current foreground data into the adjustment unit. Step 314: Determine whether the number of scenes in the current adjustment unit is greater than or equal to the scene number threshold of the adjustment unit; if it is greater than or equal to the scene number threshold of the adjustment unit, record the current adjustment unit; otherwise, proceed to the next scene until the adjustment unit meets the threshold. If the strip adjustment unit described in steps 315 and 313 is interrupted, determine whether the current number of scenes in the adjustment unit is greater than 0; if it is greater than 0, record the current number of scenes as an adjustment unit; otherwise, skip the current scene and proceed to the next scene judgment.

[0010] Furthermore, in the uncontrolled positioning method for dual-linear array images based on laser elevation points described in this invention, step 4, the rigorous model adjustment assisted by laser elevation points, includes the following steps: Step 41: After the piecewise polynomial adjustment iteration is completed, laser encryption points are selected from the encryption points; Step 42: Calculate the mean elevation difference DetH between the laser encryption point and the laser connection point; Step 43: Based on the mean elevation difference DetH, perform elevation correction on the object elevation H of all encrypted points to obtain the corrected encrypted point data; Step 44: Calculate the exterior orientation using the resection of the corrected encrypted points; Step 45: Output the encrypted point and exterior orientation data.

[0011] Furthermore, in the uncontrolled positioning method for dual-linear array images based on laser elevation points described in this invention, when calculating the mean elevation difference DetH between laser densification points and laser connection points in step 42, the difference between the object elevation information of laser densification points after piecewise polynomial adjustment and the actual elevation of laser connection points is calculated, all laser densification point elevation values ​​are calculated, outliers are removed, and the difference between the elevation accuracy after piecewise polynomial model adjustment and the actual elevation accuracy is obtained.

[0012] In a second aspect, the present invention provides a dual-linear array image uncontrolled positioning system based on laser elevation points, including an image tie point acquisition module, a laser tie point acquisition module, an adaptive piecewise polynomial adjustment module, and a laser elevation point auxiliary module. The image connection point acquisition module is used to acquire image connection points through dual-line array image matching; The laser connection point acquisition module is used to automatically rotate the laser elevation point onto the dual-line array image to obtain the laser connection point. The adaptive piecewise multi-form adjustment module is used to obtain the adjusted densification points and external orientation through adaptive piecewise polynomial rigorous model adjustment. The laser elevation point auxiliary module is used to perform rigorous model adjustment based on laser elevation points using laser connection points, correct the elevation of the densified points, and output the final exterior orientation and densified point files.

[0013] Thirdly, the present invention provides a dual-line array image uncontrolled positioning device based on laser elevation points, comprising a memory and a processor; the memory is used to store a computer program; the processor is used to implement the dual-line array image uncontrolled positioning method based on laser elevation points as described in the first aspect when the computer program is executed.

[0014] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the uncontrolled positioning method for dual-line array images based on laser elevation points as described in the first aspect.

[0015] The present invention discloses a method and apparatus for uncontrolled positioning of dual-linear array images based on laser elevation points. This method employs a rigorous model adjustment algorithm with an adaptive piecewise polynomial model to achieve rigorous model adjustment processing. For strip images where continuous adjustment models cannot be constructed due to water areas, clouds / snow, isolated islands, or data with bit errors, adaptive piecewise processing can be achieved, ultimately realizing adaptive rigorous model adjustment processing for long strip images. Simultaneously, elevation control is achieved through rigorous model adjustment assisted by laser elevation points. Laser connection points are included in the adjustment processing, and the elevation error correction of the entire strip data is performed using the consistency characteristics of strip data errors. This achieves rigorous model adjustment processing based on laser elevation points, improving the elevation positioning accuracy of strip images. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the uncontrolled positioning method for dual-linear array images based on laser elevation points according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the simulation data imaging trajectory described in an embodiment of the present invention; Figure 3 This is the simulated stereo image of the ground coverage map described in the embodiment of the present invention; Figure 4 This is a schematic diagram of the topography of the survey area described in an embodiment of the present invention. Detailed Implementation

[0017] The following detailed description of the uncontrolled positioning method and apparatus for dual-linear array images based on laser elevation points according to the present invention is provided with reference to the accompanying drawings and embodiments.

[0018] Example 1 This embodiment discloses a method for uncontrolled positioning of dual-linear array images based on laser elevation points, such as... Figure 1 As shown, it includes the following steps: Step 1: Obtain image connection points through dual-line array image matching; Step 2: Automatically rotate the laser elevation point onto the dual-line array image to obtain the laser connection point; Step 3: Obtain the adjusted densification points and exterior orientations through adaptive piecewise polynomial rigorous model adjustment; Step 4: Perform rigorous model adjustment based on laser elevation points using laser tie points to correct the elevation of the densified points, and output the final exterior orientation and densified point files.

[0019] Dual-line array image matching requires the rapid extraction of tie points between the front and rear view images. To meet the requirements of high-precision adjustment regarding the position and number of image points and to ensure that tie points are evenly distributed across the entire image area, a method combining correlation coefficient and least squares matching is adopted to achieve fast and high-precision matching.

[0020] In this embodiment of the disclosure, the dual-line array image matching in step 1 includes the following steps: Step 11: Select the rear view image as the reference image and the front view image as the image to be matched; Step 12: Perform regular gridding on the reference image and extract Moravec corner points in the gridded area; Step 13: Using the image imaging model, traverse and calculate the overlapping area between the image to be matched and the reference image. If the reference image and the image to be matched do not overlap, continue this step. Step 14: Based on the imaging model, predict the corner points obtained from the reference image onto the image to be matched in the overlapping area; Step 15: Check whether the resolution of the reference image and the image to be matched are consistent. If they are inconsistent, perform pyramid resolution consistency processing on the reference image and the image to be matched. Step 16: Perform correlation coefficient matching on the set of corner points corresponding to the acquired reference image and the image to be matched. If the correlation coefficient is greater than the threshold, perform least squares matching. Otherwise, repeat this step for the next corner point of the reference image. Step 17: Record the precise points obtained by least squares image matching. Repeat steps 15-16 for the next corner point of the reference image until all corner point sets of the reference image have been processed. Step 18: Use the RANSAC algorithm and forward intersection of corresponding points to remove gross errors from the obtained matching point set. At the same time, extract water points based on the global water mask and extract gross errors from cloud and snow and error data based on the matching threshold. Step 19: Mark the reference image as a matched image, then select the reference image from the images to be matched, and perform the processing steps 12-18 until the image set processing is complete.

[0021] The automatic laser elevation point repositioning function is responsible for automatically repositioning laser elevation points onto the dual-line array stereo image based on their longitude, latitude, and elevation. This provides the rigorous model adjustment calculation plugin with laser elevation control point information that meets the requirements for point distribution and accuracy. Its main function is to fully utilize image information and its accompanying orientation parameters to automatically predict the center position of the laser elevation point in the stereo image, and simultaneously use automatic correlation matching technology to automatically match corresponding points in the dual-line array within the laser spot area.

[0022] In this embodiment of the disclosure, step 2, the automatic laser elevation point rotation, includes the following steps: Step 21: Based on the longitude, latitude, and elevation values ​​of the laser elevation point, as well as the orientation parameters of the stereo image, automatically rotate the center position of the footprint point to the rear view image to obtain the initial image point coordinates of the rear view image. Step 22: Using the initial image point coordinates as the center, extract the feature point set of the rear view image using the Fronster operator according to the preset spot size of 30m×30m; Step 23: Select the feature points to be matched from the feature point set, and use the orientation parameters attached to the dual linear array image to predict the initial position of the corresponding points of the feature points. Step 24: Calculate the correlation coefficient of the image points to be matched. If the correlation coefficient is greater than the threshold, perform least squares point refinement processing; otherwise, calculate the next feature point of the reference image and repeat steps 23-24. Step 25: Record the matching point pairs after least squares processing; select the next feature point of the reference image and execute steps 23-25 ​​until all feature points of the reference image have been processed. Step 26: After removing erroneous points, output the matching results; Step 27: Mark the reference image as a matched image, then select the reference image from the unmatched images, and repeat steps 22-27 until the image set processing is complete.

[0023] Based on the characteristics of stereo mapping satellites, this invention constructs an adaptive piecewise polynomial model for rigorous model adjustment. Step 3, the adaptive piecewise polynomial rigorous model adjustment, includes the following steps: Step 31: Generate adaptive adjustment units based on row time data and tie points; Step 32: Start processing a single adjustment unit. Based on the rigorous adjustment model of satellite imagery, fit the original attitude data using a polynomial to obtain the initial values ​​of the piecewise polynomial model parameters. Step 33: Using satellite stereo image pairs (adjustment units) as units, calculate the object coordinates of the connection points through forward intersection in space based on the initial attitude and orbital parameters; Step 34: Establish the error equation and solve for the attitude polynomial coefficient correction values ​​of each view image by spatial resection. Step 35: Iterate through front intersection and back intersection until the residuals converge; Step 36: Process all adjustment units according to steps 32-35 until all adjustment units have been processed.

[0024] The adaptive piecewise polynomial model adjustment algorithm includes adaptive adjustment cell generation, initial value calculation of exterior orientation elements, forward intersection, and backward intersection. First, adjustment cells are constructed through adaptive adjustment cell generation. Then, piecewise polynomial adjustment processing (forward intersection and backward intersection) is performed based on the order of the adjustment cells until all adjustment cells have been processed.

[0025] Adaptive adjustment unit generation is a prerequisite for adaptive piecewise polynomial model adjustment. Multiple images within a strip are sorted according to the capture time of each image. Adjustment units are adaptively generated based on the input adjustment unit threshold, constrained by the number of image tie points and laser tie points. When considering tie point matching in the dual-line array, gross errors in water areas, clouds / snow, and bit error data have already been removed. Therefore, tie point data information can be used for adaptive adjustment unit generation.

[0026] In this embodiment of the disclosure, the adaptive adjustment unit generation in step 31 includes the following steps: Step 311: Read the foreground and background timelines of all scene images, sort them from smallest to largest according to the timeline timestamp, and form a scene list data; Step 312: Based on the sorted scene list data, sequentially read the scene image connection point files; Step 313: Determine the number of image tie points in step 312. If it is less than the threshold of 200 (by default, when the number of image tie points is less than 200, the image quality is very poor and may be water or cloud / snow areas, which cannot be included in the adjustment), then the strip adjustment unit will be interrupted and proceed to the next step; otherwise, the current foreground data will be added to the adjustment unit. Step 314: Determine whether the number of scenes in the current adjustment unit is greater than or equal to the adjustment unit scene number threshold 3 (i.e., 3 scenes constitute one adjustment unit); if it is greater than or equal to the adjustment unit scene number threshold, record the current adjustment unit; otherwise, proceed to the next scene until the adjustment unit meets the threshold. If the strip adjustment unit described in steps 315 and 313 is interrupted, determine whether the number of scenes in the current adjustment unit is greater than 0; if it is greater than 0, record the current number of scenes as an adjustment unit (the number of scenes in the current adjustment unit is less than the threshold); otherwise, skip the current scene and proceed to the next scene judgment.

[0027] The adaptive adjustment unit described in this invention can adaptively generate multiple adjustment units based on a threshold number of adjustment units, automatically skipping cloud data or continuous cloud data, thus achieving automated grouping of adjustment units. Table 1 shows a simulated example of adaptive adjustment unit generation. The simulation demonstrates the automated adjustment grouping results after clouds and snow appeared in 15 images from a strip dataset. (In the table, "positive" represents a normal image, free of clouds, snow, water areas, and error data, and whose image matching tie points meet the threshold requirements; "cloud" and "water" represent images containing clouds, snow, water areas, error data, etc., which would prevent image matching tie points from meeting the threshold requirements.) Table 1 As can be seen from Table 1, regardless of whether there is anomalies in cloud and water in the beginning, middle, or end of the strip image, whether it is a single scene interruption or a continuous scene interruption, adaptive adjustment units can be generated. The maximum number of adjustment units generated is n scenes (n is the adjustment unit threshold, which is 3 by default) and the minimum is 1 scene (that is, a single scene can also be adjusted to ensure that effective data is not lost). In this embodiment, the initial values ​​of the exterior orientation elements are calculated using the timing data, precise orbit determination data, and precise attitude determination data of the dual-line array image. Based on the coordinates of the image point measurement connection points, the initial values ​​of the exterior orientation elements corresponding to the pixels are calculated. Specifically, Lagrange interpolation is used for the exterior orientation line elements, and linear interpolation is used for the exterior orientation angle elements.

[0028] In this embodiment, the exterior orientation adjustment model adopts a piecewise polynomial model. The piecewise polynomial adjustment model (PPM) divides the entire flight path into several segments according to a certain time interval, and each segment uses a low-order polynomial to describe the exterior orientation elements. The piecewise polynomial adjustment model has no limitation on the flight path length and is suitable for adjustment processing of longer flight paths.

[0029] For the time within the i-th orbital segment, PPM can be expressed as: (1) (2) in,( X Si , Y Si , Z Si , f i , oh i , k i () represents the initial exterior orientation element of the i-th segment of the orbit; , and (i =1, 2, ...) are the coefficients of the piecewise polynomial model; t i This indicates the initial imaging time of the i-th segment of the orbit.

[0030] The linear CCD sensor employs push-broom imaging to obtain continuous image strips. Each scan line image has a strict central projection relationship with the subject and possesses its own exterior orientation element. The forward intersection establishes an instantaneous image plane coordinate system with the scan line direction as the x-axis and the flight direction as the y-axis. Let the exterior orientation element of the i-th scan line be... X Si , Y Si , Z Si , f i , oh i , k i Then the instantaneous imaging equation is: (3) in, a 1 ，a 2 ，a 3, b 1 ，b 2 ，b 3, c 1 ，c 2 ，c 3. The rotation matrix elements calculated for the attitude angle elements corresponding to the image point coordinates. f is the perpendicular distance from the camera center to the image. x 0, y 0 is the principal point.

[0031] a 1 = cos f i cos k i - sin f i sin oh i sin k i a 2 = -cos f i sin k i - sin f i sin oh i cos k i a 3 = - sin f i cos oh i b 1 = cos oh i sin k i b 2 = cos oh i cos k i b 3 = -sin oh i c 1=sin f i cos k i + cos f i sin oh i sin k i c 2 = -sin f i sin k i + cos f i sin oh i cos k i c 3 = cos f i cos oh i Based on the constructed model, the coordinate values ​​of the connection points ( x,y Read the corresponding line-time information, and combine it with the already acquired quadratic polynomial model to solve for the attitude angle elements corresponding to the connection points. f i , oh i , k i Joint orbit determination parameters X Si , Y Si , Z SiThis forms six external orientation elements.

[0032] The collinear equation (3) can be simplified to obtain: (4) In the formula, X, Y, and Z are the coefficients of the linear terms, which can be expressed as: in, a 1 ，a 2 ，a 3, b 1 ，b 2 ，b 3, c 1 ，c 2 ，c 3. To solve for the initial attitude angle elements based on equation (1) and the timing information corresponding to the coordinates of the connection points. f i , oh i , k i The elements of the constructed rotation matrix.

[0033] If n images contain the same spatial point, then there are a total of n systems of equations of the form (4) to solve for the ground coordinates. X , Y , Z Based on the least squares principle, solve for the ground coordinates of the connection point.

[0034] The re-intersection function uses ground point coordinates, interior orientation elements, and exterior orientation element information to inversely calculate its image coordinates.

[0035] The ground coordinates of all connection points obtained from the forward intersection ( X , Y , Z ) is treated as the true value and remains unchanged, while the corresponding image point coordinates ( x,y The value is considered as an observation and the corresponding correction is added. v x , v y Substituting into the collinearity equation (3) and combining it with the quadratic polynomial exterior orientation element model (1), the Taylor expansion is taken to linearize it and obtain the general form of the error equation: (5) in, v x , v y The residual vector of the image point coordinate observations; Δ f 0, Δ f 1. Δ f2. Δ oh 0, Δ oh 1. Δ oh 2. Δ k 0, Δ k 1. Δ k 2 are both corrections for exterior orientation elements; a 14 , a 15 , a 16 , a 24 , a 25 , a 26 All matrices are designed using exterior orientation angle elements; l x 、l y This represents the difference between the observed and calculated values ​​of the image point.

[0036] At the segment boundary, the adjacent piecewise polynomials i and i The external orientation elements (considering only angular elements here) should satisfy the equality constraint, thus we have: (6) Using the polynomial coefficients as observed values ​​and taking into account the constraints, the mathematical model for PPM model adjustment is as follows: (7) in, X 1 、X 2 represents the coefficient vector of the first and second polynomials. V 1. V 2 represents the residual vector of the image point coordinates corresponding to the first and second polynomials. V Y1 The residual vector of continuous observations P 1. P 2 and P Y1 for V 1. V 2. V Y1 The corresponding weight matrix, A 1. A 2. B 1. B 2 represents the corresponding design matrix.

[0037] For a single polynomial model, it can be represented in matrix form as follows: V = AX - L (8) In the formula: V for Image point coordinate observation residual vector; X These are the coefficients of the exterior azimuth element; A Design a matrix for the exterior orientation elements; L The difference between the observed and calculated values ​​of the image point is given by the image coordinates of each pair of connected points, which can be used to formulate the system of equations as shown in equation (8).

[0038] According to the principle of least squares indirect adjustment, the normal equation can be constructed as follows: A T PAX = A T PL (9) In the formula, P The weight matrix of the observations, A T For matrix A The transpose of .

[0039] The x-direction weights can be obtained using the residual values ​​in the x-direction of the image side of the connection point. v x With all residuals v x The mean error value σ0 is determined by adopting the principle that a large residual has a small weight and a small residual has a large weight. (10) Similarly, the weights in the y-direction are based on the residual values. v y This is determined by the same principles as described above. In the formula... k 0= s 0; k 1=3 s 0, set the initial weight matrix P=E ( E (the identity matrix) v The residuals of each image point can be obtained by resection calculation and used as the basis for the next weighting, so as to eliminate gross errors.

[0040] According to equation (9), we can obtain: X = ( A T PA ) -1A T PL (11) According to equation (11), the correction number is calculated using the method of gradual iterative approach. X Combined with the exterior orientation model (1), the corrected exterior orientation attitude angle elements are obtained.

[0041] To further improve the elevation accuracy of strip-based rigorous model adjustment, laser elevation point data is introduced to further enhance the elevation accuracy of stereo satellite imagery. This invention employs laser elevation point-assisted rigorous model adjustment. After piecewise polynomial adjustment, the object-space elevation values ​​of the laser elevation points are used to further correct the elevation values ​​of the densified points after adjustment, thereby improving elevation accuracy. Based on this correction, resection is then performed to generate the exterior orientation of the foreground and background images.

[0042] In this embodiment of the disclosure, the laser elevation point-assisted rigorous model adjustment in step 4 includes the following steps: Step 41: After the piecewise polynomial adjustment iteration is completed, laser densification points are selected from the densification points. The selection of laser densification points is mainly based on the adaptive piecewise polynomial adjustment results. Laser connection points are selected from the laser densification points of each scene for subsequent elevation value calculation. Step 42: Calculate the mean elevation difference DetH between the laser densification points and the laser connection points; use the difference between the object elevation information of the laser densification points after piecewise polynomial adjustment and the actual elevation of the laser connection points to calculate the elevation values ​​of all laser densification points, remove outliers, and calculate the difference between the elevation accuracy after piecewise polynomial model adjustment and the actual elevation accuracy. Step 43: Based on the mean elevation difference DetH, perform elevation correction on the object elevation H of all encrypted points to obtain the corrected encrypted point data; Step 44: Calculate the exterior orientation using the resection of the corrected densified points; calculate the exterior orientation elements of the dual-line array image using the resection of the corrected densified point data, and finally obtain high-precision densified points and image exterior orientation elements, realizing rigorous model adjustment processing based on laser elevation assistance.

[0043] Step 45: Output the encrypted point and exterior orientation data.

[0044] To verify the effect of laser-assisted adjustment on improving positioning accuracy, simulation data was used for experimental verification in this embodiment. The experimental verification includes three parts: experimental data, experimental methods, and experimental results.

[0045] 1) Experimental data The laser-assisted rigorous model adjustment simulation experiment was validated using simulation data. Simulations were performed on two stereo image pairs (forward and backsight) from the same region, located on the Korean Peninsula, with longitude ranging from 127.0226 to 127.6382, latitude from 36.8650 to 37.9352, and elevation from 10 to 610 meters. One stereo image pair had a 0-degree side tilt, while the other had a -22-degree side tilt.

[0046] The satellite parameters used in the data simulation are as follows: Orbital altitude: 500km; Ground width: 20km; Resolution: 0.25 meters; Pixel size: 5µm; Focal length: 10m; Linear array pixel count: 80,000 pixels; Camera angles: -10 degrees for the front-view camera and 5 degrees for the rear-view camera.

[0047] Each stereo pair includes ideal attitude data, orbit data, travel time data, image point measurement data, and checkpoint data. For example... Figure 2 The image shows the imaging times of the two-track data; the ground coverage is as follows. Figure 3 As shown; the topography of the survey area is as follows: Figure 4 As shown.

[0048] 2) Test methods In this embodiment of the disclosure, the error type and error magnitude of the test setup are as follows: a. Attitude error; 0.1 arcsecond (1σ) random error is added to each attitude angle.

[0049] b. Track error; 0.1 m (1σ) random error is added to each track data.

[0050] c. Image point measurement error; 0.2 pixel random error is added to the image point matching coordinates.

[0051] d. Angle error between the satellite and ground cameras; a 1 arcsecond system error is added to the angle between the satellite and ground cameras.

[0052] e. Laser altimeter data error; 0.3-meter random measurement error is added to the laser control data.

[0053] Based on the simulation data, the aforementioned errors were sequentially introduced to examine the positioning accuracy of the laser-assisted rigorous model before and after adjustment, under both non-swinging and swinging conditions. Furthermore, the processing accuracy of joint adjustment of heterogeneous stereo images relative to single-strip adjustment was further examined.

[0054] The plane coordinates used for accuracy verification are Gaussian projection (6-degree zone), and the elevation datum is CGCS2000 geodetic height.

[0055] 3) Test Results Using simulation data of 0 degrees and -22 degrees of lateral sway, attitude, orbit, image point measurement error, and star-ground camera angle error were added. A random distance measurement error of 0.3 meters was added to the laser elevation points, and the adjustment accuracy of adding different numbers of laser elevation points was analyzed.

[0056] Table 2 below shows the accuracy of laser control point auxiliary adjustment when the side swing is 0 degrees; Table 3 shows the accuracy of laser elevation point auxiliary adjustment when the side swing is -22 degrees.

[0057] Table 2 Table 3 As shown in Tables 2 and 3, the use of laser elevation points in the adjustment process can effectively offset the stereo positioning elevation error caused by changes in the angle between the satellite and ground cameras. Even with large side sway, the use of laser elevation points in the adjustment process still effectively eliminates the stereo positioning elevation error. When a small number of laser elevation points are added, the elevation accuracy does not change significantly due to the influence of laser ranging errors and various random errors. As the number of laser points increases, random errors are suppressed, and the elevation accuracy is significantly improved.

[0058] Example 2 Based on the above embodiment 1, this embodiment discloses a dual-linear array image uncontrolled positioning system based on laser elevation points, including an image tie point acquisition module, a laser tie point acquisition module, an adaptive piecewise polynomial adjustment module, and a laser elevation point auxiliary module; The image connection point acquisition module is used to acquire image connection points through dual-line array image matching; The laser connection point acquisition module is used to automatically rotate the laser elevation point onto the dual-line array image to obtain the laser connection point. The adaptive piecewise multi-form adjustment module is used to obtain the adjusted densification points and external orientation through adaptive piecewise polynomial rigorous model adjustment. The laser elevation point auxiliary module is used to perform rigorous model adjustment based on laser elevation points using laser connection points, correct the elevation of the densified points, and output the final exterior orientation and densified point files.

[0059] The specific operation steps of the uncontrolled positioning system based on laser elevation points using dual-line array images described in this embodiment are the same as those of the uncontrolled positioning method based on laser elevation points using dual-line array images described in Embodiment 1 above, and will not be repeated here.

[0060] Example 3 This embodiment discloses a controlless positioning device for dual-line array images based on laser elevation points, including a memory and a processor; the memory is used to store a computer program; the processor is used to implement the controlless positioning method for dual-line array images based on laser elevation points as described in Embodiment 1 when the computer program is executed. The specific positioning method steps are the same as those in Embodiment 1, and will not be repeated here.

[0061] Example 4 This embodiment discloses a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the uncontrolled positioning method for dual-line array images based on laser elevation points as described in Embodiment 1. The specific positioning method steps are the same as those in Embodiment 1, and will not be repeated here.

[0062] The computer described in this application embodiment can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. The computer-readable storage medium can be any usable medium that a computer can read, or a data storage device such as a server or data center that integrates one or more usable media. The usable medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., digital versatile optical disc (DVD)), or a semiconductor medium (e.g., solid-state drive (SSD)). The software formed by the computer's stored code can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other storage media that are mature in the art.

[0063] In the various embodiments of this application, the functional modules can be integrated into one processing unit or module, or each module can exist physically separately, or two or more modules can be integrated into one unit or module. In the above embodiments, they can be implemented entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, they can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated.

[0064] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for uncontrolled positioning of dual-linear array images based on laser elevation points, characterized in that... Includes the following steps: Step 1: Obtain image connection points through dual-line array image matching; Step 2: Automatically rotate the laser elevation point onto the dual-line array image to obtain the laser connection point; Step 3: Obtain the adjusted densification points and exterior orientations through adaptive piecewise polynomial rigorous model adjustment; Step 4: Perform rigorous model adjustment based on laser elevation points using laser tie points to correct the elevation of the densified points, and output the final exterior orientation and densified point files.

2. The uncontrolled positioning method for dual-linear array images based on laser elevation points according to claim 1, characterized in that, Step 1, the dual-linear array image matching, includes the following steps: Step 11: Select the rear view image as the reference image and the front view image as the image to be matched; Step 12: Perform regular gridding on the reference image and extract Moravec corner points in the gridded area; Step 13: Using the image imaging model, traverse and calculate the overlapping area between the image to be matched and the reference image. If the reference image and the image to be matched do not overlap, continue this step. Step 14: Based on the imaging model, predict the corner points obtained from the reference image onto the image to be matched in the overlapping area; Step 15: Check whether the resolution of the reference image and the image to be matched are consistent. If they are inconsistent, perform pyramid resolution consistency processing on the reference image and the image to be matched. Step 16: Perform correlation coefficient matching on the set of corner points corresponding to the acquired reference image and the image to be matched. If the correlation coefficient is greater than the threshold, perform least squares matching. Otherwise, repeat this step for the next corner point of the reference image. Step 17: Record the precise points obtained by least squares image matching. Repeat steps 15-16 for the next corner point of the reference image until all corner point sets of the reference image have been processed. Step 18: Use the RANSAC algorithm and forward intersection of corresponding points to remove gross errors from the obtained matching point set. At the same time, extract water points based on the global water mask and extract gross errors from cloud and snow and error data based on the matching threshold. Step 19: Mark the reference image as a matched image, then select the reference image from the images to be matched, and perform the processing steps 12-18 until the image set processing is complete.

3. The uncontrolled positioning method for dual-linear array images based on laser elevation points according to claim 2, characterized in that, Step 2, the automatic laser elevation point rotation, includes the following steps: Step 21: Based on the longitude, latitude, and elevation values ​​of the laser elevation point, as well as the orientation parameters of the stereo image, rotate the center position of the footprint point to the rear view image to obtain the initial image point coordinates of the rear view image. Step 22: Using the initial image point coordinates as the center, extract the feature point set of the rear view image using the Fronster operator according to the preset spot size; Step 23: Select the feature points to be matched from the feature point set, and use the orientation parameters attached to the dual linear array image to predict the initial position of the corresponding points of the feature points. Step 24: Calculate the correlation coefficient of the image points to be matched. If the correlation coefficient is greater than the threshold, perform least squares point refinement processing; otherwise, calculate the next feature point of the reference image and repeat steps 23-24. Step 25: Record the matching point pairs after least squares processing; select the next feature point of the reference image and execute steps 23-25 ​​until all feature points of the reference image have been processed. Step 26: After removing erroneous points, output the matching results; Step 27: Mark the reference image as a matched image, then select the reference image from the unmatched images, and repeat steps 22-27 until the image set processing is complete.

4. The uncontrolled positioning method for dual-linear array images based on laser elevation points according to claim 1, characterized in that, Step 3, the adaptive piecewise polynomial rigorous model adjustment, includes the following steps: Step 31: Generate adaptive adjustment units based on row time data and tie points; Step 32: Begin processing a single adjustment unit. Based on the rigorous adjustment model of satellite imagery, fit the original attitude data using a polynomial to obtain the initial values ​​of the piecewise polynomial model parameters. Step 33: Using satellite stereo image pairs as units, calculate the object coordinates of the connection points through forward intersection in space based on the initial attitude and orbital parameters; Step 34: Establish the error equation and solve for the attitude polynomial coefficient correction values ​​of each view image by spatial resection. Step 35: Iterate through front intersection and back intersection until the residuals converge; Step 36: Process all adjustment units according to steps 32-35 until all adjustment units have been processed.

5. The uncontrolled positioning method for dual-linear array images based on laser elevation points according to claim 4, characterized in that, Step 31, the generation of the adaptive adjustment unit, includes the following steps: Step 311: Read the foreground and background timelines of all scene images, sort them from smallest to largest according to the timeline timestamp, and form a scene list data; Step 312: Based on the sorted scene list data, sequentially read the scene image connection point files; Step 313: Determine the number of image connection points in step 312. If it is less than the threshold, the strip adjustment unit is interrupted and proceed to the next step; otherwise, put the current foreground data into the adjustment unit. Step 314: Determine whether the number of scenes in the current adjustment unit is greater than or equal to the scene number threshold of the adjustment unit; if it is greater than or equal to the scene number threshold of the adjustment unit, record the current adjustment unit; otherwise, proceed to the next scene until the adjustment unit meets the threshold. If the strip adjustment unit described in steps 315 and 313 is interrupted, determine whether the current number of scenes in the adjustment unit is greater than 0; if it is greater than 0, record the current number of scenes as an adjustment unit; otherwise, skip the current scene and proceed to the next scene judgment.

6. The uncontrolled positioning method for dual-linear array images based on laser elevation points according to claim 4 or 5, characterized in that, Step 4, the laser elevation point-assisted rigorous model adjustment, includes the following steps: Step 41: After the piecewise polynomial adjustment iteration is completed, laser encryption points are selected from the encryption points; Step 42: Calculate the mean elevation difference DetH between the laser encryption point and the laser connection point; Step 43: Based on the mean elevation difference DetH, perform elevation correction on the object elevation H of all encrypted points to obtain the corrected encrypted point data; Step 44: Calculate the exterior orientation using the resection of the corrected encrypted points; Step 45: Output the encrypted points and external orientation data.

7. The uncontrolled positioning method for dual-linear array images based on laser elevation points according to claim 6, characterized in that, In step 42, when calculating the mean elevation difference DetH between the laser densification points and the laser connection points, the elevation information of the laser densification points after piecewise polynomial adjustment is used to calculate the difference with the actual elevation of the laser connection points. The elevation values ​​of all laser densification points are calculated, outliers are removed, and the difference between the elevation accuracy after piecewise polynomial model adjustment and the actual elevation accuracy is calculated.

8. A controlless positioning system for dual-linear array images based on laser elevation points, characterized in that: It includes an image tie point acquisition module, a laser tie point acquisition module, an adaptive piecewise polynomial adjustment module, and a laser elevation point auxiliary module; The image connection point acquisition module is used to acquire image connection points through dual-line array image matching; The laser connection point acquisition module is used to automatically rotate the laser elevation point onto the dual-line array image to obtain the laser connection point. The adaptive piecewise multi-form adjustment module is used to obtain the adjusted densification points and external orientation through adaptive piecewise polynomial rigorous model adjustment. The laser elevation point auxiliary module is used to perform rigorous model adjustment based on laser elevation points using laser connection points, correct the elevation of the densified points, and output the final exterior orientation and densified point files.

9. A controlless positioning device for dual-linear array images based on laser elevation points, characterized in that: It includes a memory and a processor; the memory is used to store a computer program; the processor is used to implement the uncontrolled positioning method based on dual-line array images of laser elevation points according to any one of claims 1-7 when the computer program is executed.

10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, which, when executed by a processor, implements the uncontrolled positioning method for dual-line array images based on laser elevation points as described in any one of claims 1-7.

Images