CN114494429B - Large-scale uncontrolled three-dimensional adjustment net geometric positioning gross error detection and processing method - Google Patents
Large-scale uncontrolled three-dimensional adjustment net geometric positioning gross error detection and processing method Download PDFInfo
- Publication number
- CN114494429B CN114494429B CN202111592527.4A CN202111592527A CN114494429B CN 114494429 B CN114494429 B CN 114494429B CN 202111592527 A CN202111592527 A CN 202111592527A CN 114494429 B CN114494429 B CN 114494429B
- Authority
- CN
- China
- Prior art keywords
- image
- geometric positioning
- positioning
- error
- points
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
- G06T7/74—Determining position or orientation of objects or cameras using feature-based methods involving reference images or patches
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C11/00—Photogrammetry or videogrammetry, e.g. stereogrammetry; Photographic surveying
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/77—Determining position or orientation of objects or cameras using statistical methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
- G06T2207/10021—Stereoscopic video; Stereoscopic image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20076—Probabilistic image processing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Probability & Statistics with Applications (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a large-scale uncontrolled stereo adjustment net geometric positioning gross error detecting and processing method, which comprises the following steps: firstly, constructing a connection relation of a balancing net, analyzing geometric positioning errors of the balancing net, counting errors in the geometric positioning of the whole balancing net, taking a 3-time value of the errors as a geometric positioning gross error threshold value, and performing rough positioning image pair detection and geometric positioning gross error image identification. And then replacing the image with geometric positioning gross error by using the image with higher positioning precision, reconstructing a connection relation, analyzing the geometric positioning error of the adjustment net, using 2 times of the error in the geometric positioning of the adjustment net as a gross error threshold, and performing rough positioning image pair detection and geometric positioning gross error image identification again. And (4) carrying out iterative processing until geometric positioning gross errors do not exist in the whole adjustment net. The invention effectively controls the transmission and accumulation of linear errors in the stereo adjustment network, improves the reliability and precision of the large-scale uncontrolled stereo adjustment network, and realizes high-precision geometric positioning under global uncontrolled.
Description
Technical Field
The invention belongs to the technical field of space photogrammetry and satellite data processing, and particularly relates to a large-scale uncontrolled stereo adjustment network geometric positioning gross error detection and processing method.
Background
The block adjustment is a key step of global uncontrolled mapping, and the quality of the adjustment result directly determines the geometric positioning precision of a final product. In the traditional mapping project, when large-scale regional network adjustment is carried out, a processing mode of dividing the adjustment into cells and then connecting edges in the regions is adopted, and a certain number of ground control points are required to be uniformly distributed in each region to ensure the geometric positioning precision of the adjustment result. However, high-precision ground control points are difficult to arrange in an overseas area, the traditional partition adjustment operation mode using the ground control points is not applicable any more, the requirement of overseas high-precision mapping cannot be met, and large-scale high-precision three-dimensional adjustment is required under the condition of no control.
However, in a large-scale uncontrolled stereo adjustment net, the number of images participating in adjustment is large, images containing geometric positioning gross errors inevitably exist, and in the adjustment processing process, the geometric positioning gross errors can generate linear error transmission and accumulation, so that the overall geometric positioning accuracy of the adjustment net is reduced.
Disclosure of Invention
The technical problem solved by the invention is as follows: the method overcomes the defects of the prior art, provides a large-scale uncontrolled three-dimensional adjustment network geometric positioning gross error detection and processing method, effectively controls the transmission and accumulation of linear errors in the three-dimensional adjustment network, improves the reliability and precision of the large-scale uncontrolled three-dimensional adjustment network, and realizes high-precision geometric positioning under global uncontrolled.
The purpose of the invention is realized by the following technical scheme: a large-scale uncontrolled stereo adjustment net geometric positioning gross error detecting and processing method comprises the following steps: the method comprises the following steps: eliminating mismatching points of the images to be matched by adopting a scale-invariant feature matching and random sampling consistency algorithm to obtain matched connection points; step two: according to the matched connection points, representing the geometric relationship between the stereopair by using connection point residual errors, and further counting errors in the geometric positioning of the adjustment network; step three: taking 3 times of errors in the geometric positioning of the adjustment net as a geometric positioning gross error threshold, and performing rough positioning image pair detection and geometric positioning gross error image identification based on the mutual geometric relationship of the stereopair; step four: and updating the geometric positioning gross error image, reconstructing a connection relation, performing geometric positioning error analysis and gross error identification again by using 2 times of errors in the geometric positioning of the adjustment net as a gross error threshold, and performing iterative processing until no geometric positioning gross error exists in the adjustment net.
In the method for detecting and processing the geometric positioning gross error of the large-scale uncontrolled stereo adjustment network, in the first step, the step of eliminating mismatching points from the images to be matched by adopting a scale invariant feature matching and random sampling consistency algorithm to obtain matched connection points comprises the following steps: dividing regular grids of an image to be matched according to preset intervals, extracting scale-invariant feature points in each grid, and describing the feature points by using feature vectors; then, according to the geographic information of the images to be matched, finding the corresponding area range of each grid on the adjacent images, and extracting the scale-invariant feature points and describing the feature vectors in the area range; further, the Euclidean distance between two characteristic point vectors is used as a similarity judgment criterion of the characteristic points between two adjacent overlapped images to carry out characteristic point matching; and finally, eliminating the mismatching points by adopting a random sampling consistency algorithm to obtain matched connection points.
In the method for detecting and processing the geometric positioning gross error of the large-scale uncontrolled stereo adjustment network, in the second step, the residual error of the connection point refers to the difference between the coordinate of the original image after the back calculation from the reference image matching point and the coordinate of the image point with the same name of the original image.
In the method for detecting and processing the geometric positioning gross error of the large-scale uncontrolled stereo adjustment network, in the second step, the residual error of the connecting point is obtained through the following steps: performing coordinate forward calculation according to the coordinates of the connecting point image points on the reference image, the positioning parameters and the reference DEM to obtain the coordinates of the object space points; performing coordinate back calculation according to the object space point coordinates and the positioning parameters of the original image to obtain the coordinates of the image points of the original image; and taking the poor value of the original image point coordinate and the original image connection point image point coordinate as the connection point residual error.
In the method for detecting and processing the geometric positioning gross error of the large-scale uncontrolled three-dimensional block network, in the third step, 3 times of the error in the geometric positioning of the whole block network is adopted as a threshold value of the geometric positioning gross error, image pair detection is carried out in a rough positioning mode, the error in the geometric positioning between each image pair is counted in sequence, and if the error in the geometric positioning of a certain image pair is larger than a preset geometric positioning gross error threshold value, the image pair is considered to have the geometric positioning gross error.
In the method for detecting and processing geometric positioning gross error of the large-scale uncontrolled stereo adjustment net, in the third step, if an image pair formed by the image 1 and the image 2 is provided, and if errors in geometric positioning between the image 1 and other adjacent images are larger than a preset error value, the image 1 is considered to have geometric positioning gross error; the errors in the geometric positioning between the image 2 and the other adjacent images are smaller than the preset error value, and the image 2 is considered to have no geometric positioning gross error.
A large-scale uncontrolled stereo adjustment net geometric positioning gross error detecting and processing system comprises: the device comprises a first module, a second module and a third module, wherein the first module is used for eliminating mismatching points of an image to be matched by adopting a scale-invariant feature matching and random sampling consistency algorithm to obtain matched connection points; the second module is used for representing the geometric relationship between the stereopair by adopting the residual error of the connecting points according to the matched connecting points, and further counting the error in the geometric positioning of the adjustment network; the third module is used for performing rough positioning image pair detection and geometric positioning rough difference image identification based on the mutual geometric relationship of the stereopair by taking 3 times of errors in the geometric positioning of the adjustment net as a geometric positioning rough difference threshold; and the fourth module is used for updating the geometric positioning gross error image, reconstructing a connection relation, performing geometric positioning error analysis and gross error identification again by using 2 times of errors in the geometric positioning of the adjustment net as a gross error threshold value, and performing iterative processing until no geometric positioning gross error exists in the adjustment net.
In the system for detecting and processing geometric positioning gross errors of the large-scale uncontrolled stereo adjustment network, the step of eliminating mismatching points of the images to be matched by adopting a scale invariant feature matching and random sampling consistency algorithm to obtain matched connection points comprises the following steps: dividing regular grids of an image to be matched according to preset intervals, extracting scale-invariant feature points in each grid, and describing the feature points by using feature vectors; then, according to the geographic information of the images to be matched, finding the corresponding area range of each grid on the adjacent images, and extracting the scale-invariant feature points and describing the feature vectors in the area range; further, the Euclidean distance between two feature point vectors is used as a similarity judgment criterion of feature points between two adjacent overlapped images to carry out feature point matching; and finally, eliminating the mismatching points by adopting a random sampling consistency algorithm to obtain matched connection points.
In the system for detecting and processing geometric positioning gross errors of the large-scale uncontrolled stereo adjustment network, the residual errors of the connection points refer to the differences between coordinates after the reference image matching points are inversely calculated to the original image and the coordinates of the same-name image points of the original image.
In the system for detecting and processing the geometric positioning gross error of the large-scale uncontrolled stereo adjustment network, the residual error of the connecting point is obtained through the following steps: performing coordinate forward calculation according to the coordinates of the connecting point image points on the reference image, the positioning parameters and the reference DEM to obtain the coordinates of the object space points; performing coordinate back calculation according to the object space point coordinates and the positioning parameters of the original image to obtain the coordinates of the image points of the original image; and taking the difference between the original image point coordinates and the original image connection point coordinates as a connection point residual error.
Compared with the prior art, the invention has the following beneficial effects:
in the large-scale uncontrolled stereo adjustment network, the positioning precision of the stereo image pair is analyzed by constructing the connection relation between the stereo image pairs, the stereo image pair with larger positioning deviation with each adjacent stereo image pair is detected based on the mutual geometric relation of the stereo image pairs, the stereo image pair with higher positioning precision is adopted for replacement, the detection and the processing of the geometric rough positioning rough error of the stereo adjustment network are realized by iterative processing, the transmission and the accumulation of linear errors in the stereo adjustment network are effectively controlled, the reliability and the precision of the large-scale uncontrolled stereo adjustment network are improved, and the high-precision geometric positioning under global uncontrolled is realized.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
FIG. 1 is a flow chart of a method for detecting and processing geometric positioning gross errors of a large-scale uncontrolled stereo adjustment network according to an embodiment of the present invention;
FIG. 2 is a flow chart of geometric positioning error analysis provided by an embodiment of the present invention;
fig. 3 is a schematic diagram of geometric positioning gross error detection and identification provided by the embodiment of the invention.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art. It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict. The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings.
Fig. 1 is a flowchart of a geometric positioning gross error detection and processing method for a large-scale uncontrolled stereo adjustment network according to an embodiment of the present invention. As shown in fig. 1, the method comprises the steps of:
And mismatching point rejection is performed by adopting a scale-invariant feature matching and random sampling consistency algorithm, so that a large number of reliable and uniformly distributed connection point results in a large-scale uncontrolled stereo adjustment network are obtained, and a foundation is provided for geometric positioning precision analysis.
Dividing regular grids of an image to be matched according to a certain interval, extracting scale-invariant feature points in each grid, and describing the feature points by using feature vectors; then, finding out the corresponding approximate area range of each grid on the adjacent images according to the geographic information of the images, and extracting the scale-invariant feature points and describing the feature vectors in the area; further, the Euclidean distance between two feature point vectors is used as a similarity judgment criterion of feature points between two adjacent overlapped images to carry out feature point matching; and finally, rejecting mismatching points by adopting a random sampling consistency algorithm.
And 2, analyzing the geometric positioning error of the stereopair.
And acquiring a connection point residual error through positive and negative coordinate calculations according to the connection point and the geometric image positioning parameters obtained through matching, representing the geometric positioning precision consistency between the stereo image pairs by using the connection point residual error, and counting errors in the geometric positioning of the whole adjustment network.
And according to the connection points obtained by matching, representing the geometric relationship between the stereo pairs by using connection point residual errors, and further counting errors in the geometric positioning of the whole adjustment network. The connection point residual error is the difference between the coordinates of the original image and the coordinates of the same-name image points of the original image after the reference image matching point is inversely calculated, and the coordinates are firstly calculated according to the coordinates (X, Y) of the connection point image points on the reference image, the positioning parameters and the reference DEM to obtain the corresponding coordinates (X, Y, Z) of the object point. Then according to the object space point coordinate and the positioning parameter of the original image, carrying out coordinate back calculation to obtain the corresponding original image point coordinate (c) 1 ,r 1 ). Finally, the coordinate of the original image point and the coordinate (x) of the original image connecting point are obtained by calculation 1 ,y 1 ) Is used as the connection point residual (vx, vy).
Wherein, (X, Y) is the coordinate of the connecting point image point on the reference image, (X, Y, Z) is the coordinate of the object point, (c) 1 ,r 1 ) The coordinates of the image points of the original image connection points are (x) 1 ,y 1 ) And performing inverse calculation on the object space point to obtain the image point coordinate of the original image.
And 3, detecting and identifying geometric positioning gross errors of the stereopair.
And taking 3 times of errors in geometric positioning of the adjustment network as a geometric positioning gross error threshold, and performing rough positioning image pair detection and geometric positioning gross error image identification based on the mutual geometric relationship of the stereopair.
Adopting 3 times of the error in the geometric positioning of the whole adjustment net as the threshold value of the geometric positioning gross error, carrying out rough geometric positioning image pair detection, sequentially counting the error in the geometric positioning between each image pair, and considering that the geometric positioning gross error exists in the image pair if the error in the geometric positioning of a certain image pair is greater than the threshold value of the geometric positioning gross error. If the image pair formed by the image 1/the image 2 is subjected to geometric positioning gross error image identification, errors in geometric positioning between each scene image and an adjacent image are respectively counted, errors in geometric positioning between the image 1 and the other adjacent images are large, and the image is considered to have geometric positioning gross error; the geometric positioning error between the image 2 and the other adjacent images is small, and it is considered that the image has no geometric positioning gross error.
And 4, eliminating geometric positioning gross error of the adjustment network by iterative processing.
And updating the image with the geometric positioning gross error, reconstructing a connection relation, performing geometric positioning error analysis and gross error identification again by using 2 times of the error in the geometric positioning of the adjustment net as a gross error threshold, and performing iterative processing until the geometric positioning gross error does not exist in the adjustment net.
Specifically, 1, matching is carried out by adopting scale-invariant feature points to obtain a large number of uniformly distributed connection points, and a RANSAC algorithm is adopted to remove mismatching points.
(1) Scale invariant feature point matching
The matching algorithm of the feature points with unchanged scale has strong matching capability, can extract stable features, can process the matching problem of two images under the conditions of translation, rotation, affine transformation, visual angle transformation, illumination transformation and the like, and even has more stable matching capability for images shot at any angle. The basic idea of the connection point matching method based on the scale invariant feature is as follows: firstly, dividing a regular grid according to a certain interval on an image to be matched, extracting a scale-invariant feature point in each grid, and describing the feature point by using a feature vector; then, finding out the corresponding approximate area range of each grid on the adjacent images according to the geographic information of the images, and extracting the scale-invariant feature points and describing the feature vectors in the area; further, the Euclidean distance between two feature point vectors is used as a similarity judgment criterion of feature points between two adjacent overlapped images to carry out feature point matching; and finally, according to the priori knowledge that the feature points of the image need to be matched in a one-to-one correspondence mode, gross error elimination is carried out on the feature points obtained by utilizing the matching of the scale-invariant feature points, and many-to-one matching feature points are deleted.
(2) RANSAC mismatch point rejection
And eliminating a small number of error matching points possibly occurring in the connection point matching by adopting a RANSAC gross error elimination algorithm. The RANSAC algorithm iteratively estimates the parameters of a mathematical model from a set of observed data that includes outliers.
The RANSAC algorithm proposes the following basic assumptions:
1) "Incluster" data may describe several sets of model parameters, while "outlier" data is data that is not suitable for modeling;
2) The data may be affected by noise, which refers to outliers, such as from extreme noise or from incorrect assumptions or measurements that misinterpret the data;
3) RANSAC assumes that, given a set of (usually small) inliers, there is a program that can estimate the parameters that best explain or are best suited for this data model.
The RANSAC algorithm principle is as follows, according to basic assumptions:
11 Randomly select n points from the dataset as inlier points;
12 Computing model parameters appropriate for the inlier points;
13 Point-by-point judgment is carried out on the remaining points in the data set to find more inner set points if the remaining points are within the error range of the model;
14 Iteration 12), 13) until the set point does not change within two iterations.
2. And (5) carrying out geometric positioning error analysis. As shown in fig. 2, according to the connection points obtained by matching, the geometric relationship between the stereopair is represented by the connection point residual error, and further the error in the geometric positioning of the whole adjustment network is counted. The connection point residual error is the difference between the coordinates of the original image and the coordinates of the same-name image points of the original image after the reference image matching points are inversely calculated, and the calculation process of the single connection point residual error is as follows:
(1) Selecting the same name image points of the original image and the reference image;
(2) The coordinates P (X, Y, Z) of the ground points are inversely calculated from the homonymous image points of the reference image and the DEM data;
(3) And (3) performing inverse calculation on the ground points to the original image through an original image RPC model to obtain image point coordinates (x, y):
wherein, (X, Y) is the coordinates of the connecting point image point on the reference image, and (X, Y, Z) is the coordinates of the object point.
(4) Computing connection point residual error (vx, vy)
Wherein (c) 1 ,r 1 ) The coordinates of the image points of the original image connection points are (x) 1 ,y 1 ) And performing inverse calculation on the object space point to obtain the image point coordinate of the original image.
3. And detecting a positioning gross error image and identifying a geometric positioning gross error image. As shown in fig. 3, 3 times of the error in the geometric positioning of the whole adjustment net is used as a threshold value of the geometric positioning gross error, image pairs of rough geometric positioning are detected, the error in the geometric positioning between each image pair is counted in sequence, if the error in the geometric positioning of a certain image pair is greater than the threshold value of the geometric positioning gross error, the image pair is considered to have the geometric positioning gross error, such as the image pair formed by the image 1/the image 2 in fig. 3. Performing geometric positioning gross error image identification on the image pair, respectively counting errors in geometric positioning between each scene image and an adjacent image, wherein the errors in geometric positioning between the image 1 and the other adjacent images are large, and considering that the image has geometric positioning gross error; the geometric positioning error between the image 2 and the other adjacent images is small, and it is considered that the image has no geometric positioning gross error.
4. And eliminating geometric positioning gross error in the stereo adjustment network through iterative processing. After geometric positioning error analysis and error identification are finished, replacing the image with geometric positioning gross error by the image with higher geometric positioning precision, reconstructing a connection relation, performing geometric positioning error analysis and gross error identification again, and taking 2 times of the error in the geometric positioning of the whole adjustment net as a geometric positioning gross error threshold. Through iterative processing, until no geometric positioning gross error exists in the adjustment net, the reliability and the adjustment precision of the adjustment net are ensured, and the geometric positioning precision of the large-scale uncontrolled three-dimensional adjustment net is improved.
This embodiment also provides a large-scale uncontrolled stereoscopic adjustment net geometric orientation gross error detection and processing system, and this system includes: the device comprises a first module, a second module and a third module, wherein the first module is used for eliminating mismatching points of an image to be matched by adopting a scale-invariant feature matching and random sampling consistency algorithm to obtain matched connection points; the second module is used for representing the geometric relationship between the stereopair by adopting the residual error of the connecting points according to the matched connecting points, and further counting the error in the geometric positioning of the adjustment network; the third module is used for performing rough positioning image pair detection and geometric positioning rough difference image identification based on the mutual geometric relationship of the stereopair by taking 3 times of errors in the geometric positioning of the adjustment net as a geometric positioning rough difference threshold; and the fourth module is used for updating the geometric positioning gross error image, reconstructing a connection relation, performing geometric positioning error analysis and gross error identification again by using 2 times of errors in the geometric positioning of the adjustment net as a gross error threshold value, and performing iterative processing until no geometric positioning gross error exists in the adjustment net.
The method is suitable for the input images of the large-scale uncontrolled stereo adjustment net without systematic deviation, the geometric positioning error meets the requirement of unbiased normal state distribution, the images have no high-order error, and stereo images have no upper and lower parallax.
In the large-scale uncontrolled stereo adjustment network, the positioning precision of the stereo image pair is analyzed by constructing the connection relation between the stereo image pairs, the stereo image pair with larger positioning deviation with each adjacent stereo image pair is detected based on the mutual geometric relation of the stereo image pairs, the stereo image pair with higher positioning precision is adopted for replacement, the detection and the processing of the geometric rough positioning rough error of the stereo adjustment network are realized by iterative processing, the transmission and the accumulation of linear errors in the stereo adjustment network are effectively controlled, the reliability and the precision of the large-scale uncontrolled stereo adjustment network are improved, and the high-precision geometric positioning under global uncontrolled is realized.
Although the present invention has been described with reference to the preferred embodiments, it is not intended to limit the present invention, and those skilled in the art can make variations and modifications of the present invention without departing from the spirit and scope of the present invention by using the methods and technical contents disclosed above.
Claims (10)
1. A large-scale uncontrolled stereo adjustment net geometric positioning gross error detecting and processing method is characterized by comprising the following steps:
the method comprises the following steps: eliminating mismatching points of the image to be matched by adopting a scale-invariant feature matching and random sampling consistency algorithm to obtain matched connection points;
step two: according to the matched connection points, representing the geometric relationship between the stereo image pairs by using connection point residual errors, and further counting errors in the geometric positioning of the adjustment net;
step three: taking 3 times of errors in the geometric positioning of the adjustment net as a geometric positioning gross error threshold, and performing rough positioning image pair detection and geometric positioning gross error image identification based on the mutual geometric relationship of the stereopair;
step four: and updating the geometric positioning gross error image, reconstructing a connection relation, performing geometric positioning error analysis and gross error identification again by using 2 times of errors in the geometric positioning of the adjustment net as a gross error threshold, and performing iterative processing until no geometric positioning gross error exists in the adjustment net.
2. The large-scale uncontrolled stereo adjustment net geometric positioning gross error detecting and processing method according to claim 1, characterized in that: in the first step, the method for eliminating the mismatching points to obtain the matched connection points by adopting the scale-invariant feature matching and random sampling consistent algorithm for the images to be matched comprises the following steps:
dividing regular grids of an image to be matched according to preset intervals, extracting scale-invariant feature points in each grid, and describing the feature points by using feature vectors; then, according to the geographic information of the images to be matched, finding the corresponding area range of each grid on the adjacent images, and extracting the scale-invariant feature points and describing the feature vectors in the area range; further, the Euclidean distance between two feature point vectors is used as a similarity judgment criterion of feature points between two adjacent overlapped images to carry out feature point matching; and finally, eliminating the mismatching points by adopting a random sampling consistency algorithm to obtain matched connection points.
3. The large-scale uncontrolled stereo adjustment net geometric positioning gross error detecting and processing method according to claim 1, characterized in that: in the second step, the residual error of the connection point refers to the difference between the coordinates of the original image and the coordinates of the same-name image point of the original image after the reference image matching point is inversely calculated.
4. The large-scale uncontrolled stereo adjustment net geometric positioning gross error detecting and processing method according to claim 1, characterized in that: in the second step, the connection point residual error is obtained through the following steps:
performing coordinate forward calculation according to the coordinates of the connecting point image points on the reference image, the positioning parameters and the reference DEM to obtain the coordinates of the object space points;
performing coordinate back calculation according to the object space point coordinates and the positioning parameters of the original image to obtain the coordinates of the image points of the original image;
and taking the difference between the original image point coordinates and the original image connection point coordinates as a connection point residual error.
5. The large-scale uncontrolled stereo adjustment net geometric positioning gross error detecting and processing method according to claim 1, characterized in that: in the third step, 3 times of the error in the geometric positioning of the whole adjustment net is adopted as the threshold value of the geometric positioning gross error, image pair detection of rough positioning is carried out, the error in the geometric positioning between each image pair is counted in sequence, and if the error in the geometric positioning of a certain image pair is larger than the preset geometric positioning gross error threshold value, the image pair is considered to have the geometric positioning gross error.
6. The large-scale uncontrolled stereo adjustment net geometric positioning gross error detecting and processing method according to claim 5, characterized in that: in the third step, if the image pair formed by the image 1 and the image 2 is a pair, and if the errors in the geometric positioning between the image 1 and the other adjacent images are all larger than the preset error value, the image 1 is considered to have geometric positioning gross error; the errors in the geometric positioning between the image 2 and the other adjacent images are smaller than the preset error value, and the image 2 is considered to have no geometric positioning gross error.
7. A large-scale uncontrolled solid adjustment net geometric positioning gross error detecting and processing system is characterized by comprising:
the first module is used for eliminating mismatching points of the images to be matched by adopting a scale-invariant feature matching and random sampling consistency algorithm to obtain matched connection points;
the second module is used for representing the geometric relationship between the stereopair by adopting the residual error of the connecting points according to the matched connecting points, and further counting the error in the geometric positioning of the adjustment network;
the third module is used for performing rough positioning image pair detection and geometric positioning rough difference image identification based on the mutual geometric relationship of the stereopair by taking 3 times of errors in the geometric positioning of the adjustment net as a geometric positioning rough difference threshold;
and the fourth module is used for updating the geometric positioning gross error image, reconstructing a connection relation, performing geometric positioning error analysis and gross error identification again by using 2 times of errors in the geometric positioning of the adjustment net as a gross error threshold value, and performing iterative processing until no geometric positioning gross error exists in the adjustment net.
8. The large scale uncontrolled stereo adjustment net geometric positioning gross error detection and processing system according to claim 7, characterized in that: the method for eliminating the mismatching points of the images to be matched by adopting a scale-invariant feature matching and random sampling consistency algorithm to obtain the matched connection points comprises the following steps:
dividing regular grids of an image to be matched according to preset intervals, extracting scale-invariant feature points in each grid, and describing the feature points by using feature vectors; then, according to the geographic information of the images to be matched, finding the corresponding area range of each grid on the adjacent images, and extracting the scale-invariant feature points and describing the feature vectors in the area range; further, the Euclidean distance between two feature point vectors is used as a similarity judgment criterion of feature points between two adjacent overlapped images to carry out feature point matching; and finally, eliminating the mismatching points by adopting a random sampling consistency algorithm to obtain matched connection points.
9. The large scale uncontrolled stereo adjustment net geometric positioning gross error detection and processing system according to claim 7, characterized in that: the residual error of the connection point refers to the difference between the coordinate of the original image after the coordinate is inversely calculated from the matching point of the reference image and the coordinate of the image point with the same name of the original image.
10. The large scale uncontrolled stereo adjustment net geometric positioning gross error detection and processing system according to claim 7, characterized in that: the connection point residual error is obtained by the following steps:
performing coordinate forward calculation according to the coordinates of the connecting point image points on the reference image, the positioning parameters and the reference DEM to obtain the coordinates of the object space points;
performing coordinate back calculation according to the object space point coordinates and the positioning parameters of the original image to obtain the coordinates of the image points of the original image;
and taking the difference between the original image point coordinates and the original image connection point coordinates as a connection point residual error.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111592527.4A CN114494429B (en) | 2021-12-23 | 2021-12-23 | Large-scale uncontrolled three-dimensional adjustment net geometric positioning gross error detection and processing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111592527.4A CN114494429B (en) | 2021-12-23 | 2021-12-23 | Large-scale uncontrolled three-dimensional adjustment net geometric positioning gross error detection and processing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114494429A CN114494429A (en) | 2022-05-13 |
| CN114494429B true CN114494429B (en) | 2023-01-31 |
Family
ID=81493928
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111592527.4A Active CN114494429B (en) | 2021-12-23 | 2021-12-23 | Large-scale uncontrolled three-dimensional adjustment net geometric positioning gross error detection and processing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114494429B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9245201B1 (en) * | 2013-03-15 | 2016-01-26 | Excelis Inc. | Method and system for automatic registration of images |
| CN105761247A (en) * | 2016-01-30 | 2016-07-13 | 武汉大学 | Super-large scale optical satellite image regional network reliability analysis method and system |
| CN107945218A (en) * | 2017-11-22 | 2018-04-20 | 中国资源卫星应用中心 | The big distorted image matching process in edge |
-
2021
- 2021-12-23 CN CN202111592527.4A patent/CN114494429B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9245201B1 (en) * | 2013-03-15 | 2016-01-26 | Excelis Inc. | Method and system for automatic registration of images |
| CN105761247A (en) * | 2016-01-30 | 2016-07-13 | 武汉大学 | Super-large scale optical satellite image regional network reliability analysis method and system |
| CN107945218A (en) * | 2017-11-22 | 2018-04-20 | 中国资源卫星应用中心 | The big distorted image matching process in edge |
Non-Patent Citations (2)
| Title |
|---|
| 一种动态分区的边境狭长区域网平差方法;薛磊等;《遥感信息》;20191220(第06期);全文 * |
| 资源三号全国无控制整体区域网平差关键技术及应用;王密等;《武汉大学学报(信息科学版)》;20170405(第04期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114494429A (en) | 2022-05-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111815757B (en) | Large member three-dimensional reconstruction method based on image sequence | |
| CN110363858B (en) | Three-dimensional face reconstruction method and system | |
| Moulon et al. | Adaptive structure from motion with a contrario model estimation | |
| Melo et al. | Unsupervised intrinsic calibration from a single frame using a" plumb-line" approach | |
| CN108010123B (en) | Three-dimensional point cloud obtaining method capable of retaining topology information | |
| CN110349215B (en) | Camera pose estimation method and device | |
| CN111524168A (en) | Point cloud data registration method, system and device and computer storage medium | |
| CN110070610B (en) | Feature point matching method, and feature point matching method and device in three-dimensional reconstruction process | |
| CN101826206A (en) | Camera self-calibration method | |
| CN111612731B (en) | Measuring method, device, system and medium based on binocular microscopic vision | |
| El Hazzat et al. | 3D reconstruction system based on incremental structure from motion using a camera with varying parameters | |
| CN112652020B (en) | Visual SLAM method based on AdaLAM algorithm | |
| CN110942102B (en) | Probability relaxation epipolar matching method and system | |
| CN114202632A (en) | Grid linear structure recovery method and device, electronic equipment and storage medium | |
| CN117132737B (en) | Three-dimensional building model construction method, system and equipment | |
| CN111739071A (en) | Rapid iterative registration method, medium, terminal and device based on initial value | |
| CN113658194B (en) | Point cloud splicing method and device based on reference object and storage medium | |
| CN114119987A (en) | Feature extraction and descriptor generation method and system based on convolutional neural network | |
| CN117030620A (en) | Method and device for adjusting regional network based on multisource optical remote sensing satellite image | |
| CN117726747A (en) | Three-dimensional reconstruction method, device, storage medium and equipment for complementing weak texture scene | |
| CN114494429B (en) | Large-scale uncontrolled three-dimensional adjustment net geometric positioning gross error detection and processing method | |
| CN116402867A (en) | Three-dimensional reconstruction image alignment method for fusing SIFT and RANSAC | |
| CN114399547B (en) | Monocular SLAM robust initialization method based on multiframe | |
| CN109919998B (en) | Satellite attitude determination method and device and terminal equipment | |
| Tola | Multiview 3D Reconstruction of a scene containing independently moving objects |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |