A method and system for refining matching points between optical and SAR satellite images

Abstract

This invention discloses a method and system for refining matching points between optical and SAR satellite images. The method includes extracting edge structure feature maps from optical and SAR satellite images using a multi-scale Log-Gabor filter, extracting feature points from the edge structure feature maps using the Harris algorithm, constructing radiation-independent feature descriptors containing local abstract structures for the feature points, and refining the matching points between optical and SAR satellite images in the study area using a coarse-to-fine enhancement matching strategy based on the radiation-independent feature descriptors. This invention constructs radiation-independent feature descriptors containing local abstract structures to describe and represent features, which reduces the complexity of the matching algorithm. The coarse-to-fine enhancement matching strategy eliminates mismatches in the initial matching, thus refining the matching points and significantly improving matching accuracy.

Description

Technical Field

[0001] This invention belongs to the field of image feature extraction technology, and in particular relates to a method and system for refining the matching points of optical and SAR satellite images. Background Technology

[0002] The rapid development of high-resolution remote sensing technology has provided crucial technical support for detailed monitoring and analysis of the Earth's surface. In particular, the successful launches of domestically produced high-resolution optical satellites such as Tianhui-1 and Ziyuan-3, as well as synthetic aperture radar (SAR) satellites such as Gaofen-3 and Tianhui-2, have made the acquisition and processing of large-scale, multi-source global remote sensing data more feasible and accurate. Combining domestically produced optical and SAR satellite imagery enables various tasks, including high-precision topographic mapping, target identification and localization, 3D reconstruction, and change detection. The successful implementation of these applications largely depends on the accuracy of matching optical and SAR imagery.

[0003] Optical and SAR satellite image matching is a crucial component of multimodal remote sensing image matching. Its goal is to identify identical or similar content in optical and SAR satellite images and achieve high-precision alignment. This process is particularly challenging for satellites imaged at different times, under different conditions, and with different sensors. Traditional image matching methods typically rely on image gradients or grayscale information for feature detection and description. However, due to significant nonlinear radiometric and geometric differences between multimodal images, traditional image matching methods often perform poorly in such tasks. Therefore, eliminating the impact of linear radiometric and geometric differences on optical and SAR image matching and improving the accuracy and robustness of matching algorithms are pressing issues that need to be addressed in current optical and SAR image matching technologies. Summary of the Invention

[0004] This invention addresses the shortcomings of existing technologies by providing a method for refining the matching points of optical and SAR satellite images, comprising the following steps:

[0005] Step 1: Acquire optical satellite imagery and SAR satellite imagery of the study area;

[0006] Step 2: Use a multi-scale Log-Gabor filter to extract edge structure feature maps of optical satellite images and SAR satellite images, and use the Harris algorithm to extract feature points from the edge structure feature maps of the two types of satellite images.

[0007] Step 3: For the feature points extracted in Step 2, construct a radiation-independent feature descriptor containing a local abstract structure;

[0008] Step 4: Based on the radiation-independent feature descriptors of feature points, a coarse-to-fine enhancement matching strategy is used to refine the matching points between optical satellite imagery and SAR satellite imagery of the study area.

[0009] Furthermore, the Log-Gabor filter in step 2 is defined as follows:

[0010] (1)

[0011] In the formula, L This indicates Log-Gabor filtering. Represents the polar radius in logarithmic polar coordinates. Represents the polar angle in logarithmic polar coordinates. Indicates the center frequency. The scale is represented as s , direction is o The center direction of the Log-Gabor filter, express bandwidth parameters, express The bandwidth parameters.

[0012] Optical satellite imagery and SAR satellite imagery are convolved with Log-Gabor filters to generate frequency domain response components at different scales and orientations, as shown in the following formula:

[0013] (2) In the formula, I represents the input image, which is an optical satellite image or a SAR satellite image; This represents an even-symmetric Log-Gabor filter at scale s and direction o. This represents an odd-symmetric Log-Gabor filter at scale s and direction o. express Frequency domain response components at scale s and direction o; express Frequency domain response components at scale s and direction o; This indicates a convolution operation.

[0014] The amplitude and phase response components of the input image are calculated based on the frequency domain response components at different scales and orientations. The calculation formulas are as follows:

[0015] (3)

[0016] (4)

[0017] In the formula, Indicates the scale of the input image s and direction oThe amplitude response component on, Indicates the scale of the input image s and direction o The phase response components on.

[0018] The phase consistency of the input image is calculated based on the amplitude response component and the phase response component, using the following formula:

[0019] (5)

[0020] In the formula, PC represents the phase consistency of the input image; It is a weight function; This represents the magnitude response components of the input image at scale s and direction o; It is the phase deviation function; T is the noise threshold; It is a local minimum value to prevent the denominator from being zero; sign This indicates that when its value is positive, the closed variable is equal to itself, otherwise it is zero.

[0021] The second moment of the input image is calculated using phase consistency, resulting in the edge structure feature map of the input image. The formula for calculating the second moment is as follows:

[0022] (6)

[0023] In the formula, The second moment of the input image, express o The angle of direction.

[0024] The Harris operator is used to extract feature points from the edge structure feature map of the input image. The formulas for calculating the covariance matrix and response function in the Harris operator are as follows:

[0025] (7)

[0026] (8)

[0027] In the formula, M It is the covariance matrix; w This is a Gaussian window function; Indicates the convolution operation; express Projection in the vertical direction; express Projection in the horizontal direction; R For response function; Represents the determinant value of a matrix; k This is an empirical constant, with a value range of 0.04-0.06; Represents the trace of the determinant of a matrix.

[0028] Points whose response function values ​​are greater than a set threshold are selected as feature points.

[0029] Furthermore, in step 3, the spatial domain convolution kernel is constructed using the results of the image filtered by the spatial domain even-symmetric filter and the image filtered by the spatial domain odd-symmetric filter using Log-Gabor filtering, and an inverse Fourier transform is performed in the spatial domain, as shown in the following formula:

[0030] (9)

[0031] In the formula, This represents the inverse Fourier transform. L This indicates Log-Gabor filtering. i Represents the imaginary unit. o Indicates the direction of the Log-Gabor filter. Let represent the real part of the spatial domain convolution kernel, corresponding to the result of filtering by an even-symmetric filter. The imaginary part of the spatial domain convolution kernel corresponds to the result of filtering by an odd-symmetric filter, as shown in the following formula:

[0032] (10)

[0033] In the formula, i Represents the imaginary unit. o Indicates the direction of the Log-Gabor filter. L (o) indicates o Directional Log-Gabor filtering.

[0034] Assuming feature points grayscale value Then feature points In the direction of Log-Gabor filter o Response value on The calculation formula is:

[0035] (11)

[0036] In the formula, This represents the convolution operation. This represents the imaginary part of the spatial domain convolution kernel.

[0037] For each feature point, calculate its response values ​​in each direction, and then combine them into a vector as the feature descriptor for that feature point, as shown in the following formula:

[0038] (12)

[0039] In the formula, For feature descriptors, For the direction of the Log-Gabor filter The response value on.

[0040] right Perform Gaussian convolution to obtain the radiation-independent feature descriptor for each feature point. The calculation formula is as follows:

[0041] (13)

[0042] In the formula, For radiation-independent feature descriptors, For feature descriptors, It is a two-dimensional Gaussian kernel. It is the standard deviation of the two-dimensional Gaussian kernel. This indicates a convolution operation.

[0043] Furthermore, in step 4, the radiation-independent feature descriptor is obtained. Then, coarse matching of feature points is performed using the minimum Euclidean distance as the metric. After coarse matching is completed, errors are eliminated using the FSC algorithm, and matching errors less than [a certain value] are selected. The matching points of each pixel are taken as the correct matching points, and the initial affine transformation model between the optical satellite image and the SAR satellite image is obtained from all the correct matching points. .

[0044] Based on the initial affine transformation model Assuming that the error between the predicted matching position and the true value follows a Gaussian distribution with a mean of 0, the Euclidean distance with position and scale constraints is calculated. and with The minimum measure is used for fine matching, and the specific calculation formula is as follows:

[0045] (14)

[0046] In the formula, Representing feature points in optical satellite imagery Location, Indicates feature points in the SAR satellite image to be matched Location, These are all constraint coefficients, and the calculation formula is as follows:

[0047] (15)

[0048] In the formula, Representing feature points in optical satellite imagery Feature points in the SAR satellite image to be matched The symbol for calculating position error. Indicates to Coordinates after affine transformation; The symbol for absolute value calculation is represented. and They represent and The scale.

[0049] After the fine matching is completed, the FSC algorithm is used again to remove mismatches, reducing the matching error to less than 1. The matching point of each pixel is the correct matching point, which ultimately achieves the refinement of the matching point position between optical and SAR satellite images.

[0050] The present invention also provides a system for refining the matching points of optical and SAR satellite images, for implementing the method for refining the matching points of optical and SAR satellite images as described above.

[0051] Furthermore, it includes a processor and a memory, the memory being used to store program instructions, and the processor being used to call the stored instructions in the memory to execute an optical and SAR satellite image matching point refinement method as described above.

[0052] Alternatively, it may include a readable storage medium storing a computer program that, when executed, implements the optical and SAR satellite image matching point refinement method described above.

[0053] Compared with the prior art, the present invention has the following advantages:

[0054] 1) This invention utilizes Log-Gabor filtering to extract edge structure feature maps of optical and SAR images, which can eliminate the influence of nonlinear radiation differences on the matching of optical and SAR images; 2) This invention constructs a radiation-independent feature descriptor containing local abstract structures to describe and represent features, which can reduce the complexity of the matching algorithm; 3) This invention proposes to use a coarse-to-fine enhanced matching strategy to eliminate mismatches in the initial matching, thereby refining the matching points and greatly improving the matching accuracy. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0056] Figure 1 This is a flowchart illustrating the refinement process of matching points between optical satellite imagery and SAR satellite imagery according to an embodiment of the present invention.

[0057] Figures 2(a) and 2(b) are pairs of optical satellite images and SAR satellite images according to an embodiment of the present invention, wherein Figure 2(a) is an optical satellite image and Figure 2(b) is a SAR satellite image.

[0058] Figure 3(a) is an edge structure feature map of an optical satellite image according to an embodiment of the present invention, and Figure 3(b) is an edge structure feature map of a SAR satellite image according to an embodiment of the present invention.

[0059] Figure 4(a) shows the feature points of the edge structure feature map of the optical satellite image extracted in the embodiment of the present invention, and Figure 4(b) shows the feature points of the edge structure feature map of the SAR satellite image extracted in the embodiment of the present invention.

[0060] Figure 5 These are radiation-independent feature maps of optical satellite images and SAR satellite images according to embodiments of the present invention. The upper image is the radiation-independent feature map of optical satellite images, and the lower image is the radiation-independent feature map of SAR satellite images.

[0061] Figure 6(a) shows the initial matching result of optical satellite imagery and SAR satellite imagery in an embodiment of the present invention, and Figure 6(b) shows the final matching result of optical satellite imagery and SAR satellite imagery in an embodiment of the present invention.

[0062] Figures 7(a)-7(f) show the local point matching results of optical satellite images and SAR satellite images in the embodiments of the present invention. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be further described below in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0064] Example 1

[0065] like Figure 1 As shown, this embodiment of the invention provides a method for refining the matching points of optical and SAR satellite images, including the following steps:

[0066] Step 1: Acquire optical satellite imagery and SAR satellite imagery of the study area.

[0067] Step 2: Use a multi-scale Log-Gabor filter to extract edge structure feature maps of optical satellite images and SAR satellite images, and use the Harris algorithm to extract feature points from the edge structure feature maps of the two types of satellite images.

[0068] The Log-Gabor filter is defined as follows:

[0069] (1)

[0070] In the formula, L This indicates Log-Gabor filtering. Represents the polar radius in logarithmic polar coordinates. Represents the polar angle in logarithmic polar coordinates. Indicates the center frequency. The scale is represented as s , direction is o The center direction of the Log-Gabor filter, express bandwidth parameters, express The bandwidth parameters.

[0071] Optical satellite imagery and SAR satellite imagery are convolved with Log-Gabor filters to generate frequency domain response components at different scales and orientations, as shown in the following formula:

[0072] (2) In the formula, I represents the input image, which is an optical satellite image or a SAR satellite image; This represents an even-symmetric Log-Gabor filter at scale s and direction o. This represents an odd-symmetric Log-Gabor filter at scale s and direction o. express Frequency domain response components at scale s and direction o; express Frequency domain response components at scale s and direction o; This indicates a convolution operation.

[0073] The amplitude and phase response components of the input image are calculated based on the frequency domain response components at different scales and orientations. The calculation formulas are as follows:

[0074] (3)

[0075] (4)

[0076] In the formula, Indicates the scale of the input image s and direction o The amplitude response component on, Indicates the scale of the input image s and direction o The phase response components on.

[0077] The phase consistency of the input image is calculated based on the amplitude response component and the phase response component, using the following formula:

[0078] (5)

[0079] In the formula, PC represents the phase consistency of the input image; It is a weight function; This represents the magnitude response components of the input image at scale s and direction o; It is the phase deviation function; T is the noise threshold; It is a local minimum value to prevent the denominator from being zero; sign This indicates that when its value is positive, the closed variable is equal to itself, otherwise it is zero.

[0080] The second moment of the input image is calculated using phase consistency, resulting in the edge structure feature map of the input image. The formula for calculating the second moment is as follows:

[0081] (6)

[0082] In the formula, The second moment of the input image, express o The angle of direction.

[0083] Considering that image moments have a strong response at edges and corners, the Harris operator is used to extract feature points from the edge structure feature map of the input image. The formulas for calculating the covariance matrix and response function in the Harris operator are as follows:

[0084] (7)

[0085] (8)

[0086] In the formula, M It is the covariance matrix; w This is a Gaussian window function; Indicates the convolution operation; express Projection in the vertical direction; express Projection in the horizontal direction; R For response function; Represents the determinant value of a matrix; k This is an empirical constant, with a value range of 0.04-0.06; Represents the trace of the determinant of a matrix.

[0087] Figure 2(a)-Figure 2(b) shows a pair of optical and SAR satellite images, where Figure 2(a) is the optical satellite image and Figure 2(b) is the SAR satellite image. Edge structure feature maps of the image pair are extracted, as shown in Figure 3(a)-Figure 3(b). Figure 3(a) shows the edge structure feature map of the optical satellite image, and Figure 3(b) shows the edge structure feature map of the SAR satellite image. It can be seen that the edge structure feature maps of the optical and SAR satellite images eliminate the nonlinear radiation differences between the images and can well represent the structural information of the images. Feature detection is performed on the edge structure feature maps using operators, resulting in the results shown in Figure 4(a)-Figure 4(b). Figure 4(a) shows the feature points of the extracted optical satellite image edge structure feature map, and Figure 4(b) shows the feature points of the extracted SAR satellite image edge structure feature map. It can be seen that the multi-scale Log-Gabor filter + Harris algorithm method proposed in this invention can extract a large number of uniformly distributed feature points on both the optical and SAR satellite images.

[0088] Step 3: For the feature points extracted in Step 2, construct a radiation-independent feature descriptor containing a local abstract structure.

[0089] Considering the advantages of the Log-Gabor filter in terms of frequency response characteristics and directional selectivity, it can accurately extract edge and texture information from images at multiple scales and in multiple directions. Therefore, the Log-Gabor filter is chosen for filtering remote sensing images. The Log-Gabor filter exhibits a log-normal distribution in the spatial frequency domain. This characteristic allows it to effectively suppress noise and prevent high-frequency attenuation while extracting edge information. Therefore, the filtered result can be regarded as the equivalent gradient of the image in various directions. However, the Log-Gabor filter consumes a large amount of memory when processing large-scale remote sensing data in the frequency domain. Using the results of the image filtered by the spatial domain even-symmetric filter and the image filtered by the spatial domain odd-symmetric filter, a spatial domain convolution kernel is constructed, and an inverse Fourier transform is performed in the spatial domain, as shown in the following formula:

[0090] (9)

[0091] In the formula, This represents the inverse Fourier transform. L This indicates Log-Gabor filtering. i Represents the imaginary unit. o Indicates the direction of the Log-Gabor filter. Let represent the real part of the spatial domain convolution kernel, corresponding to the result of filtering by an even-symmetric filter. The imaginary part of the spatial domain convolution kernel corresponds to the result of filtering by an odd-symmetric filter, as shown in the following formula:

[0092] (10)

[0093] In the formula, i Represents the imaginary unit. o Indicates the direction of the Log-Gabor filter. L (o) indicates o Directional Log-Gabor filtering.

[0094] Considering High sensitivity to changes in energy at image edges, selection The filtered result is used as an equivalent of the first-order gradient of the image. Assuming feature points... grayscale value Then feature points In the direction of Log-Gabor filter o Response value on The calculation formula is:

[0095] (11)

[0096] In the formula, This represents the convolution operation. This represents the imaginary part of the spatial domain convolution kernel.

[0097] For each feature point, calculate its response values ​​in each direction, and then combine them into a vector as the feature descriptor for that feature point, as shown in the following formula:

[0098] (12)

[0099] In the formula, For feature descriptors, For the direction of the Log-Gabor filter The response value on;

[0100] To improve the robustness of feature descriptors, Perform Gaussian convolution to obtain the radiation-independent feature descriptor for each feature point. The calculation formula is as follows:

[0101] (13)

[0102] In the formula, For radiation-independent feature descriptors, For feature descriptors, It is a two-dimensional Gaussian kernel. It is the standard deviation of the two-dimensional Gaussian kernel. This indicates a convolution operation.

[0103] according to At spatial locations within the image, radiation-independent feature descriptors are organized into a radiation-independent feature map. This feature map contains the local abstract feature structure of the image, and therefore can be used as a dense feature representation of the image. This embodiment yields a radiation-independent feature map containing local abstract structures, as shown below. Figure 5 As shown, the upper image is a radiation-independent feature map of optical satellite imagery, and the lower image is a radiation-independent feature map of SAR satellite imagery.

[0104] Step 4: Based on the radiation-independent feature descriptors of feature points, a coarse-to-fine enhancement matching strategy is used to refine the matching points between optical satellite imagery and SAR satellite imagery of the study area.

[0105] In obtaining radiation-independent feature descriptors Then, coarse matching of feature points is performed using the minimum Euclidean distance as the metric. After coarse matching, errors are eliminated using the FSC algorithm, and matching points with matching errors of less than 10 pixels are taken as correct matching points. An initial affine transformation model between optical satellite imagery and SAR satellite imagery is obtained from all correct matching points. .

[0106] Based on the initial affine transformation model Assuming that the error between the predicted matching position and the true value follows a Gaussian distribution with a mean of 0, the Euclidean distance with position and scale constraints is calculated. and with The minimum measure is used for fine matching, and the specific calculation formula is as follows:

[0107] (14)

[0108] In the formula, Representing feature points in optical satellite imagery Location, Indicates feature points in the SAR satellite image to be matched Location, These are all constraint coefficients, and the calculation formula is as follows:

[0109] (15)

[0110] In the formula, Representing feature points in optical satellite imagery Feature points in the SAR satellite image to be matched The symbol for calculating position error. Indicates to Coordinates after affine transformation; The symbol for absolute value calculation is represented. and They represent and The scale.

[0111] After the fine matching is completed, the FSC algorithm is used again to remove mismatched points, and the matching points with a matching error of less than 3 pixels are identified as correct matching points, thus achieving the refinement of the matching points between optical and SAR satellite images.

[0112] This embodiment uses the minimum Euclidean distance as the metric for coarse matching of feature points. Errors are eliminated using the FSC algorithm, and matching points with an error of less than 10 pixels are considered correct matches, resulting in the initial matching result shown in Figure 6(a). It can be observed that Figure 6(a) contains a large number of mismatches. The Euclidean distance with position and scale constraints is then calculated. and with Fine matching is performed using the minimum measure, and errors are eliminated using the FSC algorithm. Matching points with a matching error of less than 3 pixels are considered as correct matching points, resulting in the final matching result shown in Figure 6(b). It can be seen that high-quality feature correspondences are preserved in Figure 6(b). Figures 7(a), 7(b), 7(c), 7(d), 7(e), and 7(f) show the local point location results. It can be seen that the method proposed in this invention achieves high-precision point location results.

[0113] Example 2

[0114] Based on the same inventive concept, the present invention also provides an optical and SAR satellite image matching point refinement system, including a processor and a memory. The memory is used to store program instructions, and the processor is used to call the program instructions in the memory to execute the optical and SAR satellite image matching point refinement method as described above.

[0115] Example 3

[0116] Based on the same inventive concept, the present invention also provides a system for refining the matching points of optical and SAR satellite images, including a readable storage medium on which a computer program is stored. When the computer program is executed, it implements the optical and SAR satellite image matching point refinement method as described above.

[0117] In specific implementation, the method proposed in the technical solution of this invention can be automatically executed by those skilled in the art using computer software technology. System devices for implementing the method, such as computer-readable storage media storing the corresponding computer program of the technical solution of this invention and computer equipment including the computer program running the corresponding computer program, should also be within the protection scope of this invention.

[0118] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to replace them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

Claims

1. An optical and SAR satellite image matching point position refinement method, characterized in that, The method comprises the following steps: Step 1, obtaining optical satellite images and SAR satellite images of a study area; Step 2, extracting edge structure feature maps of the optical satellite images and the SAR satellite images by using a multi-scale Log-Gabor filter, and extracting feature points of the edge structure feature maps of the two kinds of satellite images by using a Harris algorithm; Step 3, constructing a radiation-independent feature descriptor containing a local abstract structure for the feature points extracted in Step 2; A spatial domain convolution kernel is constructed by using a result of filtering the images by a spatial domain even-symmetry filter and a result of filtering the images by a spatial domain odd-symmetry filter, and inverse Fourier transform is performed in a spatial domain, and a formula is as follows: （9） wherein denotes the inverse Fourier transform, L denotes the Log-Gabor filter, i denotes the unit imaginary number, o denotes the orientation of the Log-Gabor filter, denotes the real part of the spatial domain convolution kernel, corresponding to the result of the even symmetric filter filtering, denotes the imaginary part of the spatial domain convolution kernel, corresponding to the result of the odd symmetric filter filtering, according to the following formula: （10） wherein i denotes the unit imaginary number, o denotes the orientation of the Log-Gabor filter, L (o) denotes o directional Log-Gabor filtering; Assuming feature points grayscale value Then feature points In the direction of Log-Gabor filter o Response value on The calculation formula is: （11） wherein denotes a convolution operation, denotes the imaginary part of the spatial domain kernel; For each feature point, a response value in each direction is calculated, and then the response value is combined into a vector as a feature descriptor of the feature point, and a formula is as follows: （12） wherein is a feature descriptor, is a response value on a Log-Gabor filter direction on a Log-Gabor filter direction right Perform Gaussian convolution to obtain the radiation-independent feature descriptor for each feature point. The calculation formula is as follows: （13） wherein is a radiation independent feature descriptor, is a feature descriptor, is a two-dimensional Gaussian kernel, is a standard deviation of the two-dimensional Gaussian kernel, denotes a convolution operation; Step 4, based on the radiation-independent feature descriptor of the feature points, a coarse-to-fine enhanced matching strategy is used to realize refinement of matching point positions of the optical satellite images and the SAR satellite images of the study area; In obtaining the radiation-independent feature descriptor After that, the feature points are roughly matched by taking the minimum Euclidean distance as the measure, and after the rough matching is completed, the FSC algorithm is used to eliminate errors, and the matching points with matching errors less than pixels are taken as correct matching points, and the initial affine transformation model between the optical satellite image and the SAR satellite image is obtained from all the correct matching points .

2. The method for refining the matching points of optical and SAR satellite images as described in claim 1, characterized in that: A definition of the Log-Gabor filter in Step 2 is as follows: （1） wherein L denotes a Log-Gabor filter, denotes the polar radius of a log-polar coordinate, denotes the polar angle of a log-polar coordinate, denotes the center frequency, denotes a Log-Gabor filter with scale s and orientation o , denotes the bandwidth parameter , denotes the bandwidth parameter ; The optical satellite images and the SAR satellite images are respectively subjected to convolution operation with the Log-Gabor filter to generate frequency domain response components in different scales and directions, and a formula is as follows: (2) where I denotes the input image, here optical satellite imagery or SAR satellite imagery; represents even-symmetric Log-Gabor filter at scale s and orientation o; represents odd-symmetric Log-Gabor filter at scale s and orientation o; represents frequency domain response component at scale s and orientation o; represents frequency domain response component at scale s and orientation o; represents convolution operation.

3. The method of claim 2, wherein the method comprises: determining a first SAR image point corresponding to the optical image point; determining a second SAR image point corresponding to the optical image point; and determining the refined point based on the first SAR image point and the second SAR image point. Based on the frequency domain response components in different scales and directions generated in Step 2, amplitude response components and phase response components of the input images are calculated, and a calculation formula is as follows: （3） （4） wherein denotes the amplitude response component of the input image in scale s and direction o , denotes the phase response component of the input image in scale s and direction o . Based on the amplitude response components and the phase response components, phase consistency of the input images is calculated, and a calculation formula is as follows: （5） where PC denotes the phase coherency of the input image; is the weight function; denotes the amplitude response component of the input image at scale s and orientation o; is the phase deviation function; T is the noise threshold; is a minimum value to prevent the denominator from being zero; the symbol denotes that when its value is positive, the enclosed quantity equals itself, otherwise it is zero; Based on the phase consistency, a second moment of the input images is calculated to obtain an edge structure feature map of the input images, and a calculation formula of the second moment is as follows: （6） wherein the second moment of the input image, denotes o the angle of the direction.

4. The method of claim 3, wherein the method comprises: In Step 2, a Harris operator is used to extract feature points of the edge structure feature map of the input images, and a calculation formula of a covariance matrix and a response function in the Harris operator is as follows: （7） （8） wherein M is a covariance matrix; w is a Gaussian window function; denotes a convolution operation; denotes a projection in the vertical direction; denotes a projection in the horizontal direction; R is a response function; denotes a matrix determinant value; k is an empirical constant, with a value ranging from 0.04 to 0.06; denotes a trace of a matrix determinant; Points with a response function value greater than a set threshold value are selected as the feature points.

5. The method for refining the matching points of optical and SAR satellite images as described in claim 1, characterized in that: Step 4 is based on the initial affine transformation model Assuming that the error between the predicted matching position and the true value follows a Gaussian distribution with a mean of 0, the Euclidean distance with position and scale constraints is calculated. and with The minimum measure is used for fine matching, and the specific calculation formula is as follows: （14） wherein denotes the position of a feature point in the optical satellite image, denotes the position of a feature point in the SAR satellite image to be matched, are constraint coefficients, and the calculation formula is as follows:​​ （15） wherein, denotes the position error of a feature point in the optical satellite image to be matched in the SAR satellite image denotes the position error of a feature point in the optical satellite image denotes the coordinates of the feature point in the optical satellite image after an affine transformation; denotes the absolute value calculation symbol, and denote the scale of and respectively; After the fine matching, the false matching points are removed by FSC algorithm again. The matching points with matching error less than one pixel are correct matching points, and the fine matching of the optical and SAR satellite image matching points is realized finally.

6. An optical and SAR satellite image matching point position refining system, characterized in that, The device comprises a processor and a memory, the memory is used for storing program instructions, and the processor is used for calling the program instructions in the memory to execute the optical and SAR satellite image matching point position refinement method in any one of claims 1-5.

7. An optical and SAR satellite image matching point position refining system, characterized in that, The device comprises a readable storage medium, and the readable storage medium stores a computer program, and the computer program is executed to realize the optical and SAR satellite image matching point position refinement method in any one of claims 1-5.

Images