Image registration method, computer equipment and storage device

Through the method of adaptively filtering the threshold value and projection error relationship, the accuracy of image registration is improved, the problem of low image registration accuracy in the prior art is solved, and higher matching accuracy is achieved.

CN114463393BActive Publication Date: 2024-11-08ZHEJIANG DAHUA TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111670473.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-31
Publication Date
2024-11-08
Estimated Expiration
2041-12-31

AI Technical Summary

Technical Problem

The existing image registration technology has low image registration accuracy due to scene differences, brightness changes, viewing angle changes and other factors, resulting in a large number of wrong matches.

Method used

By determining multiple sets of candidate matching point pairs in the image pair to be registered, filtering them based on the filter threshold of the candidate matching point pair, adjusting the filter threshold adaptively using the projection error relationship to improve the accuracy of feature point matching, and passing the target Matching points perform image registration for determining image transformation parameters.

Benefits of technology

Improves the accuracy of image registration, reduces error matching, and enhances the stability and accuracy of image registration.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114463393B_ABST
    Figure CN114463393B_ABST
Patent Text Reader

Abstract

The present application discloses an image registration method, a computer device and a storage device. The method includes: determining multiple groups of candidate matching point pairs in the image pair to be registered, wherein the image pair to be registered includes a first image and a second image, and the candidate matching point pairs include a first feature point of the first image and a second feature point of the second image; filtering each candidate matching point pair based on a first filtering threshold of each candidate matching point pair to obtain a number of first target matching point pairs, wherein the first filtering threshold of the candidate matching point pair is related to a first error relationship of the candidate matching point pair, and the first error relationship of the candidate matching point pair is a magnitude relationship between the candidate matching point pair and other candidate matching point pairs regarding a first projection error, and the first projection error is determined using the first transformation parameter of the image pair to be registered; using a number of first target matching point pairs, determining the second transformation parameter of the image pair to be registered. The above scheme can improve the accuracy of image registration.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] Technology Neighborhood

[0002] The present application relates to the field of image processing technology, and in particular to an image registration method, a computer device and a storage device. Background Art

[0003] Image registration is the process of matching and superimposing two or more images acquired at different times, with different sensors (imaging equipment) or under different conditions (weather, illumination, camera position and angle, etc.). Image registration technology has been widely used in remote sensing data analysis, computer vision, image processing, medical and other fields.

[0004] At present, in the matching process of image registration, a large number of false matches are often generated due to factors such as scene differences, brightness changes, perspective changes, and low overlap of images. These false matches will lead to low accuracy of image registration. Summary of the invention

[0005] The main technical problem solved by the present application is to provide an image registration method, a computer device and a storage device, which can improve the accuracy of image registration.

[0006] In order to solve the above problems, the first aspect of the present application provides an image registration method, which includes: determining multiple groups of candidate matching point pairs in an image pair to be registered, wherein the image pair to be registered includes a first image and a second image, and the candidate matching point pairs include a first feature point of the first image and a second feature point of the second image; filtering each candidate matching point pair based on a first filtering threshold of each candidate matching point pair to obtain a number of first target matching point pairs, wherein the first filtering threshold of the candidate matching point pair is related to a first error relationship of the candidate matching point pair, and the first error relationship of the candidate matching point pair is a size relationship between the candidate matching point pair and other candidate matching point pairs regarding a first projection error, and the first projection error is determined using a first transformation parameter of the image pair to be registered; using the number of first target matching point pairs, determining the second transformation parameter of the image pair to be registered.

[0007] In order to solve the above problem, the second aspect of the present application provides a computer device, which includes a memory and a processor coupled to each other, the memory stores program data, and the processor is used to execute the program data to implement any step of the above image registration method.

[0008] In order to solve the above problem, the third aspect of the present application provides a storage device, which stores program data that can be executed by a processor, and the program data is used to implement any step of the above image registration method.

[0009] The above scheme determines multiple groups of candidate matching point pairs in the image pair to be registered, wherein the image pair to be registered includes a first image and a second image, and the candidate matching point pairs include a first feature point of the first image and a second feature point of the second image; based on the first filtering threshold of each candidate matching point pair, each candidate matching point pair is filtered to obtain a plurality of first target matching point pairs; wherein; since the first filtering threshold of the candidate matching point pair is related to the first error relationship of the candidate matching point pair, the first error relationship of the candidate matching point pair is the magnitude relationship between the candidate matching point pair and other candidate matching point pairs regarding the first projection error, and the first projection error is determined by using the first transformation parameter of the image pair to be registered, so that the first filtering threshold can be adapted to the changes of each candidate matching point pair, so that for different regions of the image pair to be registered, the first filtering threshold of the candidate matching point pair can be adaptively determined to filter each candidate matching point pair, so as to improve the accuracy of matching the first feature point and the first feature point in the image to be registered; thus, the second transformation parameter of the image pair to be registered is determined by using the plurality of first target matching point pairs obtained after filtering, and the image pair to be registered is registered, so as to improve the accuracy of image registration. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in this application, the following briefly introduces the drawings required in the description of the embodiments. Obviously, the drawings described below are only some embodiments of this application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0011] Figure 1 It is a flowchart of the first embodiment of the image registration method of the present application;

[0012] Figure 2 This application Figure 1 A schematic diagram of a flow chart of an embodiment of step S11;

[0013] Figure 3 This application Figure 1 A schematic diagram of a flow chart of an embodiment of step S12;

[0014] Figure 4 This application Figure 1 A schematic diagram of a flow chart of an embodiment of step S13;

[0015] Figure 5 This is a schematic diagram of an example of matching results of using the RANSAC algorithm to perform feature matching on the image pair to be registered;

[0016] Figure 6 It is a flowchart of the second embodiment of the image registration method of the present application;

[0017] Figure 7This is a schematic diagram showing an example of matching results of performing feature matching on a pair of images to be registered according to the second embodiment of the image registration method of the present application;

[0018] Figure 8 It is a structural schematic diagram of an embodiment of the image registration device of the present application;

[0019] Fig. 9 It is a structural schematic diagram of an embodiment of a computer device of the present application;

[0020] Fig.10 It is a structural diagram of an embodiment of a storage device of the present application. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0022] The terms "first" and "second" in this application are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Thus, the features defined as "first" and "second" can explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally also includes steps or units that are not listed, or optionally also includes other steps or units inherent to these processes, methods, products or devices.

[0023] Reference to "embodiments" in this application means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0024] The present application provides the following embodiments, and each embodiment is described in detail below.

[0025] See also Figure 1 , Figure 1 1 is a flow chart of the first embodiment of the image registration method of the present application. The method may include the following steps:

[0026] S11: Determine multiple groups of candidate matching point pairs in the image pair to be registered, wherein the image pair to be registered includes a first image and a second image, and the candidate matching point pairs include a first feature point of the first image and a second feature point of the second image.

[0027] The image pair to be registered includes a first image and a second image, wherein the first image and the second image may be images acquired at different times, using different sensors (imaging devices) or under different conditions (weather, illumination, camera position and angle, etc.).

[0028] The first image and the second image may be registered, for example, image B may be registered to image A, image A may be used as a reference image and image B may be used as a floating image, thereby obtaining a deformation field of image B registered to image A.

[0029] In some embodiments, the first image and the second image are remote sensing images. A remote sensing image refers to an image with ground target features obtained by an imaging system on an aircraft or satellite. The first image and the second image may be multi-view remote sensing images, which refer to two images (referred to as a reference image and an image to be registered) collected by the same sensor, but with a horizontal or (and) vertical change in the collection angle. Multi-view remote sensing image registration can achieve image-assisted navigation.

[0030] In some embodiments, the first image and the second image are medical images. Through the registration of medical images, various information can be accurately integrated into the same image, allowing doctors to observe lesions and structures from various angles more conveniently and accurately. At the same time, through the registration of dynamic images collected at different times, the changes in lesions and organs can be quantitatively analyzed, making medical diagnosis, surgical planning, and radiotherapy planning more accurate and reliable.

[0031] In some embodiments, the first image and the second image are video frame images, which can be used to perform video analysis and pattern recognition and automatically track the motion changes of the target object by registering the video frame images.

[0032] The first image and the second image of the present application may also be other images that need to be registered, and the present application does not impose any limitation on this.

[0033] When performing image registration on a first image and a second image, for example, registering the second image to the first image, feature points of the first image and the second image can be extracted respectively, and feature descriptions of the feature points can be determined, so as to match the feature points of the first image and the second image, and determine multiple groups of candidate matching point pairs in the image pair to be registered, wherein the candidate matching point pairs include a first feature point of the first image and a second feature point of the second image, and the first feature point and the second feature point are feature points of initial matching.

[0034] S12: Filter each candidate matching point pair based on a first filtering threshold of each candidate matching point pair to obtain a plurality of first target matching point pairs.

[0035] At least two groups of candidate matching point pairs may be selected to determine first transformation parameters of the image pair to be registered, thereby determining a first projection error of each candidate matching point pair using the first transformation parameters.

[0036] A magnitude relationship between the candidate matching point pair and other candidate matching point pairs regarding the first projection error is used as the first error relationship between the candidate matching point pairs.

[0037] In some embodiments, the first filtering threshold of the candidate matching point pairs is related to the first error relationship of the candidate matching point pairs, and therefore, the first filtering threshold of the candidate matching point pairs can be determined based on the first error relationship of each candidate matching point pair. Each group of candidate matching point pairs can correspond to one first filtering threshold.

[0038] Each candidate matching point pair is filtered based on the first filtering threshold of each candidate matching point pair, that is, the wrong matches in the candidate matching point pairs can be filtered to obtain a plurality of first target matching point pairs.

[0039] S13: Determine second transformation parameters of the image pair to be registered using a plurality of first target matching point pairs.

[0040] In some embodiments, the second transformation parameter is a homography transformation parameter, wherein the homography transformation parameter can be represented by a matrix, and the homography transformation parameter matrix describes the relationship between two images of the same object taken at different viewing angles, that is, the homography transformation parameter can represent the position mapping relationship between the first image and the second image.

[0041] By using a plurality of first target matching point pairs, second transformation parameters for registering the second image to the first image are determined, so that the second image can be registered to the first image using the second transformation parameters.

[0042] In this embodiment, by determining multiple groups of candidate matching point pairs in the image pair to be registered, wherein the image pair to be registered includes a first image and a second image, and the candidate matching point pairs include a first feature point of the first image and a second feature point of the second image; based on the first filtering threshold of each candidate matching point pair, each candidate matching point pair is filtered to obtain a plurality of first target matching point pairs; wherein; since the first filtering threshold of the candidate matching point pair is related to the first error relationship of the candidate matching point pair, the first error relationship of the candidate matching point pair is the size relationship between the candidate matching point pair and other candidate matching point pairs regarding the first projection error, and the first projection error is determined by using the first transformation parameter of the image pair to be registered, so that the first filtering threshold can adapt to the changes of each candidate matching point pair, so that for different areas of the image pair to be registered, the first filtering threshold of the candidate matching point pair can be adaptively determined to filter each candidate matching point pair, so as to improve the accuracy of matching the first feature point and the first feature point in the image to be registered; thus, by using the plurality of first target matching point pairs obtained after filtering, the second transformation parameter of the image pair to be registered is determined, and the image pair to be registered is registered, so as to improve the accuracy of image registration.

[0043] In some embodiments, see Figure 2 The above step S11, determining multiple groups of candidate matching point pairs in the image pair to be registered, may include the following steps:

[0044] S111: Acquire multiple groups of initial matching point pairs in the image pair to be registered.

[0045] A preset feature detection algorithm can be used to respectively determine the first feature points and the second feature points of the first image and the second image in the image pair to be registered, as well as the descriptors of the feature points. Then, the first feature points of the first image and the second feature points on the second image are matched using a preset feature matching algorithm to obtain multiple groups of initial matching point pairs of the image pair to be registered, the initial matching point pairs including the first feature points and the second feature points of the initial matching.

[0046] Among them, the preset feature detection algorithms include DFM algorithm (Dynamic Feature Match), FAST algorithm (Features from Accelerated Segment Test), etc., and this application does not impose any restrictions on this.

[0047] The preset feature matching algorithms include DNNS algorithm (Dense Nearest Neighbor Search), SIFT algorithm (Scale Invariant Feature Transform), and other algorithms, and this application does not impose any restrictions on this.

[0048] In some implementations, taking the preset feature matching algorithm including the DNNS algorithm as an example, the DNNS algorithm may be used to determine initial matching point pairs in the first image and the second image using a mutual nearest neighbor search and a preset threshold ratio.

[0049] The Euclidean distance between each first feature point in the first image and each second feature point in the second image can be obtained, and the initialization matching point pair is determined based on the Euclidean distance. The smaller the value of the Euclidean distance, the better the matching quality of the first feature point and the second feature point.

[0050] As an example, image A includes a first feature point Pa and a first feature point Pc, and image B includes a second feature point Pb and a second feature point Pd. If the Euclidean distance between the first feature point Pa and the second feature point Pb is the smallest, the second feature point Pb is the closest point to the first feature point Pa. If the Euclidean distance between the first feature point Pa and the second feature point Pd is the second smallest, the second feature point Pd is the second closest point to the first feature point Pa.

[0051] Similarly, if the second feature point Pb of image B is closest to the first feature point Pa in image A, the second closest point is the first feature point Pc.

[0052] If the ratio of the Euclidean distance of the first feature point Pa from the nearest point Pb to the second nearest point Pd in ​​image B (hereinafter referred to as the nearest neighbor distance ratio) is less than the preset threshold ratio, it can be said that the nearest point Pb matches the first feature point Pa. At the same time, if the ratio of the Euclidean distance of the second feature point Pb from the nearest point Pa to the second nearest point Pc in image A (hereinafter referred to as the nearest neighbor distance ratio) is less than the preset threshold ratio, it can be said that the nearest point Pa matches the second feature point Pb. Therefore, the first feature point Pa and the second feature point Pb can be called a pair of initial matching points.

[0053] In the above manner, multiple groups of initial matching point pairs in the image pairs to be registered can be obtained.

[0054] S112: Select at least one set of seed point pairs from the multiple sets of initial matching point pairs.

[0055] Since there are some erroneous matching point pairs in the initial matching point pairs, these erroneous matching point pairs are usually inconsistent with the surrounding correct matching point pairs.

[0056] In order to solve the above problem, the present application selects at least one set of seed point pairs from multiple sets of initial matching point pairs to divide the entire image into multiple local areas, thereby filtering out erroneous matching point pairs based on the local areas of the seed point pairs.

[0057] In some embodiments, the matching confidence of the feature point can be determined based on the nearest neighbor distance ratio of the feature point (first feature point or second feature point) in the initial matching point pair, and at least one set of seed point pairs can be selected based on the matching confidence of the feature point. Among them, the smaller the nearest neighbor distance ratio, the higher the matching confidence, which can also indicate the matching credibility of the initial matching point pair. The value range of the nearest neighbor distance ratio and / or the matching confidence can be a number between 0 and 1. Of course, other value ranges can also be used, and the present application is not limited to this.

[0058] In some embodiments, for each first feature point belonging to an initial matching point pair in the first image, if the first feature point has the highest matching confidence within its second distance range, for example, there are other first feature points of multiple initial matching point pairs within the second distance range R of the first feature point, but the matching confidence of the first feature point is the highest, then the first feature point can be used as the first seed point of the first image. At the same time, the initial matching point pair corresponding to the first seed point can also be used as a seed point pair, that is, the second feature point initially matched by the first feature point can be used as the second seed point of the second image.

[0059] By selecting seed point pairs in the image to be registered in a local area (second distance range) of the image, the selected seed point pairs can have good feature representation capabilities, and the selected seed point pairs are evenly distributed in the image.

[0060] S113: For each group of seed point pairs, find at least one group of neighboring point pairs within the first distance range of the seed point pairs.

[0061] The seed point pair includes a first seed point in the first image and a second seed point in the second image.

[0062] Based on each group of seed point pairs, at least one group of neighboring point pairs can be found within a first distance range of the seed point pairs.

[0063] Specifically, if the initial matching point pair is within the first distance range of the seed point pair, the initial matching point pair is used as the neighborhood point pair, wherein the initial matching point pair is within the first distance range of the seed point pair when: the first feature point of the initial matching point is within the first distance range of the first seed point of the seed point pair, and the second feature point of the initial matching point is within the first distance range of the second seed point of the seed point pair. The first distance range is greater than the second distance range.

[0064] In some embodiments, if the first feature point of the initial matching point pair is within the third distance range of the first seed point of the seed point pair, and the second feature point of the initial matching point pair is within the fourth distance range of the second seed point of the seed point pair, the initial matching point pair is used as a neighborhood point pair, wherein the third distance range and the fourth distance range are greater than the second distance range.

[0065] As an example, the seed point pair includes a first seed point P1 and a second seed point P2. The initial matching point pair includes a first feature point Pa and a second feature point Pb. Image A includes a first seed point P1. Within the third distance range R of the first seed point P1, a = βR includes any first feature point Pa (initial matching point pair), where R can be expressed as a second distance range, and β is a parameter greater than 1. In the image B, a seed point P2 is included. If the second feature point Pb is within the fourth preset range R of the seed point P2 b =γR, where R can be expressed as the second distance range and γ is a parameter greater than 1. Then the first feature point Pa can be determined as a neighborhood point of the first seed point P1. The first feature point Pa and the second feature point Pb can be regarded as a set of neighborhood point pairs.

[0066] Among them, the third distance range R a and the fourth distance range R b They can be the neighborhood space sizes of image A and image B respectively, and parameters β and γ can control the overlapping area between different neighborhoods.

[0067] By the above method, for each group of seed point pairs, all neighboring point pairs within the first distance range of the seed point pairs can be selected.

[0068] S114: taking the seed point pair and the neighborhood point pair as candidate matching point pairs.

[0069] The seed point pairs and the neighborhood point pairs may be used as candidate matching point pairs, and all the neighborhood point pairs of each group of seed point pairs may be added to the neighborhood point pair sets of various sub-point pairs. The neighborhood point pair sets may include multiple groups of candidate matching point pairs.

[0070] In some implementations, each group of neighborhood point pairs may be added to the neighborhood point pair set of its seed point pair as candidate matching point pairs.

[0071] In this embodiment, by using a group of seed point pairs with high matching confidence and uniform distribution, and selecting neighborhood point pairs within the first range of the seed point pairs, the entire image (image pair to be registered) can be divided into multiple local areas (neighborhood point pair sets). The selected seed point pairs and neighborhood point pairs have good feature characterization capabilities in the local areas. The selected seed point pairs and neighborhood point pairs are used to filter out abnormal initial matching point pairs, which can improve the matching accuracy.

[0072] In some embodiments, see Figure 3 The above step S12, filtering each candidate matching point pair based on the first filtering threshold of each candidate matching point pair to obtain a plurality of first target matching point pairs, may include the following steps:

[0073] S121: Selecting a second number of groups of candidate matching point pairs from the plurality of groups of candidate matching point pairs to form a sample point pair set.

[0074] In this embodiment, each seed point pair and a neighborhood point pair set consisting of all its neighborhood points are taken as units, and steps S121 to S125 in this embodiment are performed on the neighborhood point pair set of each seed point pair in turn for multiple rounds of iterations. A neighborhood point pair set may include multiple groups of candidate matching point pairs, that is, seed point pairs and multiple groups of neighborhood point pairs.

[0075] In some implementations, the number of iterations may be preset to a first number, that is, the first number K rounds of steps S121 to S125 in this embodiment are performed.

[0076] In some implementations, during each round of iteration, a second number n of candidate matching point pairs may be selected from multiple groups of candidate matching point pairs in the neighborhood point pair set to form a sample point pair set.

[0077] In some implementations, during each iteration, the second number of candidate matching point pairs with the highest matching confidence are selected from the multiple groups of candidate matching point pairs in the neighborhood point pair set, that is, the first second number of candidate matching point pairs ranked by matching confidence are selected to form the sample point pair set.

[0078] In some embodiments, there is a preset relationship between the first number and the second number. In each round of iteration, at least two groups of candidate matching point pairs need to be randomly selected from the second number n groups of candidate matching point pairs (see the following step S122 for details), wherein the number of at least two groups may be a third number m. The pre-set relationship may include: the first number K of iterations is less than or equal to the number of combinations of the second number n and the third number m.

[0079] Combination means: taking m different elements (0≤m≤n) from n different elements at a time, and combining them into a group regardless of their order, is called a combination of m elements selected from n elements without repetition. The total number of all such combinations is called the number of combinations.

[0080] That is, the preset relationship among the first quantity K, the second quantity n and the third quantity m can be expressed as:

[0081]

[0082] In the above formula (1), K represents the first number (number of iterations), n represents the second number (number of sample point pairs), and m represents the third number (the composition of candidate matching point pairs selected in each round of iteration). Indicates the number of combinations.

[0083] In some embodiments, taking two sets of matching point pairs as an example, the preset relationship includes: the product of the difference between the second number and the preset value and the second number is greater than or equal to twice the first number. The preset relationship between the first number and the second number can be expressed as:

[0084]

[0085] In the above formula (2), K represents the first number, n represents the second number, and a represents a preset value. For example, if the preset value is 1, the preset relationship between the first number and the second number can be expressed as:

[0086]

[0087] By transforming the above formula (3), it can be obtained that the preset relationship between the first quantity and the second quantity can be expressed as:

[0088]

[0089] In the above formula (4), K is a first number, which can represent the number of iterations, and n is a second number, which can represent the number of sample point pairs.

[0090] S122: Determine a first transformation parameter using at least two groups of candidate matching point pairs.

[0091] In some implementations, during each round of iteration, at least two groups of candidate matching point pairs may be randomly selected from the sample point pair set.

[0092] In some embodiments, the first transformation parameter is an affine transformation parameter. Affine transformation means that in geometry, a vector space undergoes a linear transformation and is followed by a translation to transform into another vector space. Affine transformation can represent a transformation from an affine plane to itself, which can maintain the collinearity of points and the parallelism of lines. Collinearity means that the relative position relationship between two-dimensional lines remains unchanged, parallel lines remain parallel, and the intersection angle of intersecting lines remains unchanged. Parallelism means that an arc remains an arc after transformation. A straight line remains a straight line.

[0093] In some application scenarios, the first transformation parameter may be a central affine transformation, wherein the central affine transformation refers to an affine transformation containing an invariant point, and the invariant point is called the center of the central affine transformation.

[0094] In some embodiments, the seed point pair is used as the center of the affine transformation, and the first transformation parameter can be expressed as:

[0095] P′=AP

[0096] That is:

[0097]

[0098] In the above formula (5), P represents the pixel coordinates before transformation, P' represents the coordinates after transformation, and A represents the affine transformation matrix, which contains four parameters and can be solved by any two sets of candidate matching point pairs, so as to determine the affine transformation matrix, that is, determine the first transformation parameter A.

[0099] Taking the registration of the second image B to the first image A as an example, after selecting two sets of candidate matching point pairs, P may represent the second feature point in the candidate matching point pair, and P' may represent the first feature point in the candidate matching point pair.

[0100] S123: Determine a first projection error of each candidate matching point pair based on the first transformation parameter.

[0101] In each round of iteration, after selecting at least two groups of candidate matching point pairs from the sample point pair set of the neighborhood point pair set and determining the first transformation parameter A, the first projection error of each candidate matching point pair in the neighborhood point pair set can be obtained.

[0102] Taking the jth iteration as an example, the set of neighborhood point pairs is N i , the first transformation parameter (affine transformation parameter) obtained is For the candidate matching point pair k, the first projection error can be obtained as:

[0103]

[0104] In the above formula (6), represents the first projection error of candidate matching point pair k, It can represent the second feature point in the candidate matching point pair k, It can represent the first feature point in the candidate matching point pair k. Represents the second feature point of the second image, through the first transformation parameter Mapped to the location in the first image.

[0105] This step can obtain the first projection error of each candidate matching point pair.

[0106] In some embodiments, in the process of registering the images to be registered, factors such as atmospheric or lens distortion may introduce nonlinearity in the two images due to the following reasons. The feature points found in the two images do not represent the same spatial point in space. Due to factors such as the neighborhood points in the image are not on the same plane in the three-dimensional space, there is no clear boundary between the first projection residual of the first transformation parameter and the true transformation parameter. A fixed filtering threshold is set for the first projection residual, and it is not possible to accurately determine whether the candidate matching point pairs are correctly matched. In order to solve this problem, the present application adopts an adaptive threshold for each candidate matching point pair, that is, a first filtering threshold for filtering, as follows.

[0107] S124: Based on the first error relationship of each candidate matching point pair, determine a first filtering threshold for each candidate matching point pair.

[0108] The first error relationship of the candidate matching point pairs is represented by the point pair statistics of the candidate matching point pairs.

[0109] Each candidate matching point pair can be used as a reference matching point pair, wherein the point pair statistics of the reference matching point pair are the number of candidate matching point pairs whose first projection error is less than or equal to the reference matching point pair. For example, when obtaining the first filtering threshold of the candidate matching point pair, the candidate matching point pair can be used as a reference matching point pair, and the first projection error of the reference matching point pair is Rc. If the first projection error of other candidate matching point pairs is Rh, the number of candidate matching point pairs whose Rh is less than or equal to Rc can be counted, which is the point pair statistics.

[0110] The first filtering threshold of the reference matching point pairs is determined by using the point pair statistics of the reference matching point pairs, wherein the ratio between the point pair statistics of the reference matching point pairs and the total point pair quantity of the candidate matching point pairs can be used as the first filtering threshold of the reference matching point pairs.

[0111] In some implementations, the ratio of the number of reference matching point pairs to the total number of candidate matching point pairs and the threshold coefficient may be used as the first filtering threshold of the reference matching point pairs. The first filtering threshold may be expressed as:

[0112]

[0113] In the above formula (7), t a represents the first filtering threshold of candidate matching point pairs (reference matching point pairs), P(r k ) represents the point pair statistics of candidate matching point pairs (reference matching point pairs), M represents the total number of candidate matching point pairs, and t represents the threshold coefficient.

[0114] In some embodiments, if the first projection error of the reference matching point pair (candidate matching point pair) is less than or equal to the first filtering threshold, the reference matching point pair (candidate matching point pair) can be determined to be an inlier pair, and the inlier can be represented as a correct match. Otherwise, the reference matching point pair (candidate matching point pair) is determined to be an outlier pair.

[0115] In some implementations, each group of candidate matching point pairs may correspond to a first filtering threshold.

[0116] As an example, the total number of point pairs including candidate matching point pairs in the neighborhood point pair set is M, and the process of determining the first filtering threshold of the candidate matching point pair k1 is as follows: the candidate matching point pair k1 is taken as the reference matching point pair, and the first projection error of the reference matching point pair k1 is R1. It can be assumed that the first projection error is the first filtering threshold t1. Then, the first filtering threshold t1 can be used to determine whether the first projection errors of other candidate matching point pairs are less than or equal to the first filtering threshold t1, and the number Pk of candidate matching point pairs whose first projection errors are less than or equal to the first filtering threshold t1 can be counted. Thus, the first filtering threshold of the candidate matching point pair k1 can be determined as:

[0117] S125: retaining the candidate matching point pairs for which the magnitude relationship between the first projection error and the first filtering threshold meets the preset requirement as the first target matching point pairs.

[0118] If the first projection error of the candidate matching point pair is less than or equal to the first filtering threshold, the candidate matching point pair can be judged as an inner point pair, that is, the candidate matching point pair can be added to the inner point pair set of the neighborhood point pair set. The candidate matching point pair as an inner point pair can represent the candidate matching point pair that meets the preset requirements, and can be retained as the first target matching point pair, that is, the inner point pair set.

[0119] If the first projection error of the candidate matching point pair is greater than the first filtering threshold, the candidate matching point pair can be determined to be an external point pair and is not used as the first target matching point pair.

[0120] Through the above method, all first target matching point pairs of the neighborhood point pair set, that is, the inner point pair set, can be determined.

[0121] In this embodiment, the method of constructing a sample point pair set can increase the probability of feature points with higher matching confidence being sampled, so that the best first transformation parameter can be found in the smallest possible number of iterations, thereby reducing the amount of calculation.

[0122] In addition, by determining whether the match is correct through the first filtering threshold of each candidate matching point pair, correct matches can be fully retained and incorrect matches can be filtered out.

[0123] In some embodiments, the above-mentioned step of filtering each candidate matching point pair based on the first filtering threshold of each candidate matching point pair to obtain a plurality of first target matching point pairs is performed for a first number of rounds, and the first transformation parameters used in each round are determined by using at least two groups of candidate matching point pairs selected in each round.

[0124] In some embodiments, see Figure 4 The above step S13, using a plurality of first target matching point pairs to determine the second transformation parameters of the image pair to be registered, may include the following steps:

[0125] S131: Selecting a first target matching point pair from one of the first number of rounds as a second target matching point pair.

[0126] In the first number of iterations, for each neighborhood point pair set, a first target matching point pair set can be selected in each iteration, that is, the inner point pair set, and the iteration with the largest number of first target matching point pairs can be selected as the target round, that is, the round with the largest number of inner point pairs in the inner point pair set as the target round. And the first target matching point pair of the target round is used as the second matching point pair set, that is, as the optimal inner point pair set.

[0127] Thus, each group of first target matching point pairs in the second matching point pair set is used as a second target matching point pair.

[0128] S132: Determine second transformation parameters of the image pair to be registered based on the second target matching point pair.

[0129] For each neighborhood point pair set, at least two sets of second target matching point pairs can be used to determine new first transformation parameters of the image pair to be registered, that is, at least two sets of candidate matching point pairs selected by the target round can be used to determine new first transformation parameters of the image pair to be registered.

[0130] Based on the second filtering threshold of each candidate matching point pair, each candidate matching point pair is filtered respectively to obtain a plurality of third target matching point pairs, wherein the second filtering threshold of the candidate matching point pair is related to the second error relationship of the candidate matching point pair, the second error relationship of the candidate matching point pair is the size relationship between the candidate matching point pair and other candidate matching point pairs regarding the second projection error, and the second projection error is determined using the new first transformation parameter.

[0131] The specific implementation process of this process can refer to the above-mentioned step S12, and this application will not go into details here.

[0132] Through the above method, several third target matching point pairs of each neighborhood point pair set can be obtained, that is, the final inner point pair set. Several third target matching points can be used as correctly matched neighborhood point pairs, so that incorrectly matched neighborhood point pairs in each neighborhood point pair set can be filtered out.

[0133] Thus, several third target matching point pairs of the entire pair of images to be registered can be obtained, and based on the position information of the several third target matching point pairs, the second transformation parameters are calculated and obtained, wherein the second transformation parameters are homography transformation parameters.

[0134] The projection of a point on the same plane in three-dimensional space on two different pixel planes can be described by a homography relationship. Since remote sensing images can be approximately considered as images of objects on the same plane, the relationship between the image pairs to be registered can be described by homography transformation parameters. In other words, the second image can be registered to the first image using homography transformation parameters.

[0135] In some embodiments, the homography transformation parameters may be expressed in the form of a matrix, for example, the homography transformation parameter matrix H is a 3x3 matrix. The relationship between the homography transformation parameters for registering the first image and the second image may be expressed as:

[0136] P 1 =HP 2

[0137] That is:

[0138]

[0139] In the above formula (8), P1 represents the first feature point of the third target matching point pair of the first image, P2 represents the second feature point of the third target matching point pair of the second image, and H represents the homography transformation parameter matrix. The homography transformation parameter matrix H has 8 unknown parameters and 1 known parameter, and can be solved using at least four pairs of non-collinear third target matching point pairs.

[0140] In this embodiment, by using the first filtering threshold to filter abnormal matches of the neighborhood pair set (local neighborhood), correctly matched neighborhood point pairs, that is, a plurality of third target matching points, can be obtained. Thus, the second parameter is determined based on the plurality of correctly matched third target matching points, and the floating image (second image) is mapped to the reference image (first image) through the homography transformation parameter, thereby improving the registration accuracy between the image pairs to be registered.

[0141] In the prior art, the first image and the second image are remote sensing images. The RANSAC algorithm (RANdomSAmple Consensus) can be used to match feature points of the first image and the second image. The RANSAC algorithm can estimate the parameters of the mathematical model from a set of observed data containing outliers in an iterative manner. The RANSAC algorithm assumes that the data contains correct data and abnormal data (or noise). The correct data is recorded as inliers and the abnormal data is recorded as outliers.

[0142] See also Figure 5 In the process of matching the feature points of the first image and the second image using the RANSAC algorithm, since the textures of the image pair to be registered are relatively similar, there are a large number of wrong matches in the feature matching results.

[0143] In order to solve the above problems, the present application improves the RANSAC algorithm by using a new sampling method (selecting a set of sample point pairs) and an adaptive threshold (a first filtering threshold) to improve the RANSAC algorithm, and uses the improved RANSAC algorithm to match feature points of the first image and the second image. The process has reference to the following embodiments.

[0144] See also Figure 6 , Figure 6 1 is a flow chart of the second embodiment of the image registration method of the present application. In this embodiment, an improved RANSAC algorithm can be used to obtain the first transformation parameters and a plurality of third target matching points.

[0145] That is, the neighborhood point set N i As a unit, the RANSAC algorithm can be used to calculate the first transformation parameter for each neighborhood point pair set, and determine whether the neighborhood point pairs of each neighborhood point pair set are correctly matched. The correctly matched neighborhood point pairs in the neighborhood point pair set are called inner point pairs, and the abnormal or wrongly matched neighborhood point pairs are called outer point pairs. The domain point pairs in this embodiment can be represented as the above-mentioned candidate matching point pairs.

[0146] Among them, the improved RANSAC algorithm steps are as follows:

[0147] S21: Build a second number of sample point pair sets before matching confidence.

[0148] The set of domain point pairs N i The candidate matching point pairs are sorted by matching confidence, and the second number (first n) of candidate matching point pairs with the highest matching confidence are selected to construct a sample point pair set.

[0149] S22: Randomly select two groups of candidate matching point pairs from the sample point pair set and determine the first transformation parameter.

[0150] From the sample point pair set, two sets of candidate matching point pairs (sample point pairs) are randomly selected, and the first transformation parameters are determined using the two sets of candidate matching point pairs. The first transformation parameters are the central affine transformation matrix A, which is recorded as the affine transformation model A.

[0151] S23: Select inlier pairs based on the first projection error and the first filtering threshold, and determine an inlier pair set.

[0152] The first transformation parameter can be used to calculate the first projection errors of all candidate matching point pairs in the neighborhood point pair set, and based on the first error relationship of each candidate matching point pair, the first filtering threshold of each candidate matching point pair can be determined accordingly. If the first projection error of the candidate matching point pair is less than or equal to the first filtering threshold, the candidate matching point pair can be used as an inlier pair and added to the inlier pair set I. By performing step S23 on each group of candidate matching point pairs, the neighborhood point pair set N can be obtained. i The corresponding set of interior point pairs is I.

[0153] S24: Determine whether the number of interior point pairs in the interior point pair set I is greater than the optimal interior point pair set I m The number of inliers.

[0154] In the first round of iteration, the optimal interior point set I m is the first round of neighborhood point pair set N i The corresponding set of interior point pairs is I.

[0155] If the number of inliers in the current inlier set I is greater than that in the optimal inlier set I m , then execute the following step S25, and the interior point pair set I can be used as the optimal interior point pair set I m .

[0156] If the number of inliers in the current inlier pair set I is not greater than that in the optimal inlier pair set I m , then execute the following step S26.

[0157] S25: Take the current interior point pair set I as the optimal interior point pair set I m .

[0158] S26: Determine whether the number of iterations j is greater than a first number K (iteration number threshold).

[0159] If it is determined to be greater than the first number K, the iteration is exited and the following step S29 is executed. Otherwise, the number of iterations j is increased by 1, and the above steps S22 to S25 are repeated.

[0160] S27: Based on the optimal interior point pair set I m , determine new first transformation parameters.

[0161] That is, we can use the optimal interior point pair set I m The inner point pair (the second target matching point pair) determines the new first transformation parameters.

[0162] S28: Selecting an inner point pair as a third target matching point pair based on the second projection error and the second filtering threshold.

[0163] Based on the new first transformation parameter, the optimal interior point pair set I can be obtained m The second projection error and the second filtering threshold of each inner point pair in the , so that the inner point pair can be selected again as the third target matching point pair, which is expressed as the neighborhood point pair set N i The correctly matched neighborhood point pairs in are also the correctly matched candidate matching point pairs.

[0164] By repeating the above steps S21 to S28, each neighborhood point pair set N can be obtained. i The correctly matched neighborhood point pairs in the image are used to obtain a plurality of third target matching point pairs, thereby using the plurality of third target matching point pairs to obtain second transformation parameters, and using the second transformation parameters to register the second image to the first image.

[0165] See also Figure 7 Among the matching point pairs in the first image and the second image, the abnormal matching or incorrect matching point pairs are filtered out, so that the feature matching accuracy of the images to be registered is high. Even in the matching of remote sensing images with relatively consistent textures, the above method of the present application can also perform feature matching on remote sensing images well, which can reduce the incorrect matching between feature points and has a high feature matching accuracy.

[0166] The specific implementation of this embodiment can refer to the implementation process of the above embodiment, which will not be repeated here.

[0167] For the above embodiment, the present application provides an image registration device. Figure 8 , Figure 8 3 is a schematic diagram of the structure of an embodiment of an image registration device of the present application. The image registration device 30 comprises a matching module 31 , a filtering module 32 and a registration module 33 .

[0168] The matching module 31 is used to determine multiple groups of candidate matching point pairs in the image pair to be registered, wherein the image pair to be registered includes a first image and a second image, and the candidate matching point pairs include a first feature point of the first image and a second feature point of the second image.

[0169] The filtering module 32 is used to filter each candidate matching point pair based on the first filtering threshold of each candidate matching point pair to obtain a plurality of first target matching point pairs, wherein the first filtering threshold of the candidate matching point pair is related to the first error relationship of the candidate matching point pair, and the first error relationship of the candidate matching point pair is the size relationship between the candidate matching point pair and other candidate matching point pairs regarding the first projection error, and the first projection error is determined using the first transformation parameter of the image pair to be registered.

[0170] The registration module 33 is used to determine the second transformation parameters of the image pair to be registered by using a plurality of first target matching point pairs.

[0171] The specific implementation of this embodiment can refer to the implementation process of the above embodiment, which will not be repeated here.

[0172] For the above embodiment, the present application provides a computer device, see Fig. 9 , Fig. 9 4 is a schematic diagram of the structure of an embodiment of a computer device of the present application. The computer device 40 includes a memory 41 and a processor 42, wherein the memory 41 and the processor 42 are coupled to each other, the memory 41 stores program data, and the processor 42 is used to execute the program data to implement the steps in any embodiment of the above-mentioned image registration method.

[0173] In this embodiment, the processor 42 may also be referred to as a CPU (Central Processing Unit). The processor 42 may be an integrated circuit chip having signal processing capabilities. The processor 42 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. A general-purpose processor may be a microprocessor or the processor 42 may also be any conventional processor, etc.

[0174] The specific implementation of this embodiment can refer to the implementation process of the above embodiment, which will not be repeated here.

[0175] The method of the above embodiment can be implemented in the form of a computer program, so the present application proposes a storage device, see Fig.10 , Fig.10 The storage device 50 stores program data 51 that can be run by a processor, and the program data 51 can be executed by the processor to implement the steps of any embodiment of the above-mentioned image registration method.

[0176] The specific implementation of this embodiment can refer to the implementation process of the above embodiment, which will not be repeated here.

[0177] The storage device 50 of this embodiment can be a medium that can store program data 51, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or it can also be a server that stores the program data 51. The server can send the stored program data 51 to other devices for execution, or it can also execute the stored program data 51 by itself.

[0178] In the several embodiments provided in the present application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device implementation described above is only schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0179] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0180] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0181] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a storage device, which is a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for an electronic device (which can be a personal computer, a server, or a network device, etc.) or a processor to perform all or part of the steps of each implementation method of the present application.

[0182] Obviously, those skilled in the art should understand that the above modules or steps of the present application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0183] The above descriptions are merely embodiments of the present application and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. An image registration method, It is characterized in that The method comprises: Determine a plurality of groups of candidate matching point pairs in the pair of images to be registered, wherein the pair of images to be registered includes a first image and a second image, and the candidate matching point pairs include a first feature point of the first image and a second feature point of the second image; Based on the first filtering threshold of each candidate matching point pair, each candidate matching point pair is filtered respectively to obtain a plurality of first target matching point pairs, wherein the first filtering threshold of the candidate matching point pair is related to the first error relationship of the candidate matching point pair, the first error relationship of the candidate matching point pair is the magnitude relationship between the candidate matching point pair and other candidate matching point pairs with respect to the first projection error, the first projection error is determined by using the first transformation parameter of the image pair to be registered; the first transformation parameter is an affine transformation parameter; Determining second transformation parameters of the to-be-registered image pair using the plurality of first target matching point pairs, wherein the second transformation parameters are homography transformation parameters; The filtering of each candidate matching point pair based on the first filtering threshold of each candidate matching point pair to obtain a plurality of first target matching point pairs includes: Determining the first transformation parameter using at least two groups of the candidate matching point pairs; Determining a first projection error of each of the candidate matching point pairs based on the first transformation parameter; Based on the first error relationship of each of the candidate matching point pairs, a first filtering threshold of each of the candidate matching point pairs is correspondingly determined, wherein the first error relationship of the candidate matching point pairs is characterized by the number of point pair statistics of the candidate matching point pairs, including: taking each of the candidate matching point pairs as a reference matching point pair, and using the number of point pair statistics of the reference matching point pairs to determine the first filtering threshold of the reference matching point pairs, wherein the number of point pair statistics of the reference matching point pairs is the number of the candidate matching point pairs whose first projection error is less than or equal to the reference matching point pairs; The candidate matching point pairs for which the magnitude relationship between the first projection error and the first filtering threshold meets a preset requirement are retained as the first target matching point pairs.

2. The method according to claim 1, It is characterized in that The step of determining a first filtering threshold of the reference matching point pairs by using the point pair statistics of the reference matching point pairs includes: The ratio between the statistical number of point pairs of the reference matching point pairs and the total number of point pairs of the candidate matching point pairs is used as a first filtering threshold of the reference matching point pairs.

3. The method according to claim 1, It is characterized in that The step of filtering each of the candidate matching point pairs based on the first filtering threshold of each of the candidate matching point pairs to obtain a plurality of first target matching point pairs is performed for a first number of rounds, and the first transformation parameters used in each round are determined by using at least two groups of the candidate matching point pairs selected in each round; The determining the second transformation parameters of the to-be-registered image pair by using the plurality of first target matching point pairs comprises: Selecting the first target matching point pair in one round from the first number of rounds as the second target matching point pair; Based on the second target matching point pair, second transformation parameters of the to-be-registered image pair are determined.

4. The method according to claim 3, It is characterized in that Before filtering each of the candidate matching point pairs based on the first filtering threshold of each of the candidate matching point pairs to obtain a plurality of first target matching point pairs, the method further includes: A second number of groups of candidate matching point pairs are selected from the multiple groups of candidate matching point pairs to form a sample point pair set, wherein at least two groups of candidate matching point pairs selected in each round are selected from the sample point pair set, and there is a preset relationship between the first number and the second number.

5. The method according to claim 4, It is characterized in that The at least two groups of candidate matching point pairs selected in each round are randomly selected; and / or, the product of the difference between the second number and a preset value and the second number is greater than or equal to twice the first number; And / or, selecting a second number of groups of candidate matching point pairs from the multiple groups of candidate matching point pairs comprises: The second number of groups of candidate matching point pairs with the highest matching confidence level are selected from the multiple groups of candidate matching point pairs.

6. The method according to claim 3, It is characterized in that The step of selecting the first target matching point pair in one round from the first number of rounds as the second target matching point pair includes: Among the first number of rounds, selecting a round with the largest number of the first target matching point pairs as a target round, and using the first target matching point pairs of the target round as a second matching point pair set; And / or, determining the second transformation parameter of the image pair to be registered based on the second target matching point pair includes: Determining new first transformation parameters of the pair of images to be registered using at least two groups of the second target matching point pairs; Based on the second filtering threshold of each candidate matching point pair, each candidate matching point pair is filtered to obtain a plurality of third target matching point pairs, wherein the second filtering threshold of the candidate matching point pair is related to the second error relationship of the candidate matching point pair, the second error relationship of the candidate matching point pair is the magnitude relationship between the candidate matching point pair and other candidate matching point pairs with respect to the second projection error, and the second projection error is determined by using the new first transformation parameter; The second transformation parameters are calculated based on the position information of the plurality of third target matching point pairs.

7. The method according to claim 1, It is characterized in that The step of determining a plurality of groups of candidate matching point pairs in the image pair to be registered comprises: Acquire multiple groups of initial matching point pairs in the image pair to be registered; Selecting at least one set of seed point pairs from the multiple sets of initial matching point pairs; For each group of seed point pairs, find at least one group of neighboring point pairs within a first distance range of the seed point pairs; The seed point pair and the neighborhood point pair are used as the candidate matching point pair.

8. The method according to claim 7, It is characterized in that The step of selecting at least one set of seed point pairs from the multiple sets of initial matching point pairs comprises: For each of the first feature points in the first image that belongs to the initial matching point pair, if the first feature point has the highest matching confidence within its second distance range, taking the first feature point as a first seed point, and taking the initial matching point pair corresponding to the first seed point as the seed point pair, wherein the first distance range is greater than the second distance range; The seed point pair includes a first seed point in the first image and a second seed point in the second image, and searching for at least one set of neighboring point pairs within a first distance range of the seed point pair includes: If the initial matching point pair is within the first distance range of the seed point pair, the initial matching point pair is used as the neighborhood point pair, wherein the initial matching point pair is within the first distance range of the seed point pair if: the first feature point of the initial matching point is within the first distance range of the first seed point of the seed point pair, and the second feature point of the initial matching point is within the first distance range of the second seed point of the seed point pair.

9. The method according to claim 1, It is characterized in that The first image and the second image are remote sensing images.

10. A computer device, It is characterized in that The invention comprises a memory and a processor coupled to each other, wherein the memory stores program data, and the processor is used to execute the program data to implement the steps of the method according to any one of claims 1 to 9.

11. A storage device, It is characterized in that Program data that can be executed by a processor are stored, and the program data is used to implement the steps of the method according to any one of claims 1 to 9.