Optical and SAR image fusion method facing visual navigation requirement

Abstract

The invention relates to the field of image fusion, in particular to an optical and SAR image fusion method for a visual navigation demand. Comprising the following steps: S1, preprocessing an optical image and an SAR image, converting the optical image into a gray level image by adopting an average value method, and obtaining an equalized SAR image through gray level equalization processing; s2, pixel-level navigation landmark detection of the SAR image is realized by using a CFAR detection algorithm; s3, processing the CFAR detection result of the SAR image by adopting morphological operation, and removing a false alarm target and interference pixels to obtain an SAR image navigation landmark segmentation mask; and S4, extracting to-be-fused pixels from the SAR image by using the segmentation mask, fusing the to-be-fused pixels with the optical image, and realizing transmission of spectral information through IHS transformation to obtain a fusion result of the optical image and the SAR image. The invention aims to solve the problem that an image fusion method for enhancing an optical image by using unique information of an SAR image while keeping main attributes of the optical image is lacked in the prior art.

Description

[Technical field]

[0001] The invention relates to the field of image fusion, and in particular to an optical and SAR image fusion method oriented to visual navigation requirements. [Background technology]

[0002] As the application requirements of remote sensing images continue to increase, the information that a single image can provide is limited, so it is necessary to integrate multimodal image data into one image to form a richer and more meaningful fused image. Recently, many researchers have focused on the problem of optical and SAR (Synthetic Aperture Radar) image fusion, which has been used to provide unique information for land classification, target recognition and object detection.

[0003] At present, the fusion of optical and SAR images uses a hybrid method to fuse the texture information extracted from the SAR image with the high-frequency detail information obtained by the wavelet transform of the optical image, or introduces a multi-scale morphological gradient to enhance the use of edge features, so that the fusion result has higher correlation and lower spectral distortion. In addition, in view of the serious noise damage problem in SAR images, the NSST (Non-Subsampled Shearlet Transform) method is used to suppress a large amount of SAR noise, making the fusion result clearer. In order to avoid spectral and spatial distortion in fusion, a gain injection method is proposed to reduce the information loss in the multi-scale space, or an improved method based on phase consistency information is proposed to reduce the spectral distortion caused by the nonlinear radiation difference between images. In contrast to the pursuit of injecting more information into the fusion result, some studies have begun to focus on the direct use value of the fused image. By using the saliency detection method to extract the SAR image pixel saliency map instead of the SAR image for fusion, the spectral and spatial information are better preserved. And based on the method of adaptive multi-scale Gaussian co-processing filter decomposition, the texture noise is smoothed and the edge information in the texture is retained, thereby generating a fusion result with low noise interference.

[0004] Although the above methods can obtain information-rich, low-spectral and spatial distortion optical and SAR image fusion results, thus providing richer external environment information for visual navigation. However, in the subsequent application of images in visual navigation, it is necessary to match the satellite reference image with the optical real-time image. Therefore, the purpose of visual navigation to fuse optical and SAR images is to enhance the optical image by using the unique information of the SAR image while retaining the main attributes of the optical image. However, the optical and SAR image fusion method for visual navigation needs is not yet visible. [Summary of the invention]

[0005] In order to solve the above problems, the present invention provides an optical and SAR image fusion method oriented to visual navigation requirements.

[0006] The present invention is implemented by the following technical scheme, and provides an optical and SAR image fusion method for visual navigation needs, comprising the following steps:

[0007] S1. Preprocessing the optical image and the SAR image, converting the optical image into a grayscale image by using an average value method, and obtaining a balanced SAR image by grayscale equalization processing, wherein the optical image is an RGB color three-channel image, and the SAR image is a single-channel grayscale image;

[0008] S2, using CFAR detection algorithm to realize pixel-level navigation landmark detection of SAR images;

[0009] S3, using morphological operations to process the SAR image CFAR detection results, remove false alarm targets and interference pixels, and obtain the SAR image navigation landmark segmentation mask;

[0010] S4. Use the segmentation mask to extract the pixels to be fused from the SAR image and fuse them with the optical image, and use the IHS transformation to realize the transmission of spectral information to obtain the fusion result of the optical and SAR images.

[0011] Specifically, the preprocessing in S1 includes the following steps:

[0012] S101, converting the optical image into a single-channel grayscale image by using the average value method, and taking the average value of the three channel components of the optical image as the grayscale value of the image, that is,

[0013]

[0014] Among them, R(m,n), G(m,n) and B(m,n) represent the pixel values ​​of the three components of the optical image R, G, and B at the position (m,n), respectively, and I(m,n) is the pixel value at the position (m,n) after the optical image is grayed;

[0015] S102, grayscale equalization is performed on the SAR image. The SAR image after grayscale equalization can be expressed as:

[0016]

[0017] Where μ() and σ() represent the mean and standard deviation of the image pixel values, respectively, S is the initial SAR image, S Eq is the SAR image after equalization, and I is the optical image processed by grayscale;

[0018] S103: Obtained pre-processed optical image I and SAR image SEq .

[0019] Specifically, S2 includes the following steps:

[0020] S201, design a CFAR detector to process the grayscale balanced SAR image S Eq Perform CFAR target detection;

[0021] The CFAR detector traverses all pixels in the entire SAR image through a sliding window, the sliding window is a hollow window, the sliding window includes a protection window and a background window, the background window is used to count background clutter to calculate the target detection threshold, and the protection window is used to prevent the target pixel from leaking into the background window, thereby affecting the correctness of the background clutter statistics;

[0022] The CFAR detector also includes a target area and an interference area. The size of the protection window and the background window is determined by the given target area size and interference area width. Assuming that the size of the target area is W×H, where W is the width of the target area and H is the height of the target area, then the size of the protection window is L protection ×L protection , where L protection It can be expressed as:

[0023] L protection =2max(M,N)+1 (3);

[0024] Assume that the width of the interference zone is L clutter , then the size of the background window is L background ×L background , where L background It can be expressed as:

[0025] L background =L protection +2L clutter (4);

[0026] S202, setting false alarm rate P fa , according to the set false alarm rate P fa Calculate the CFAR detection threshold ε th , the expression is:

[0027]

[0028] S203, assuming that the interference pixels satisfy the Rayleigh distribution, estimate the interference parameters in the interference area by combining the interference statistical model of the Rayleigh distribution, including the mean μ clutter and standard deviation σ clutter , the expressions are:

[0029] μ clutter =μ(SEq (A)) (6),

[0030] σ clutter =σ(S Eq (A)) (7),

[0031] Among them, A represents the interference area;

[0032] S204, traversing all positions of the equalized SAR image through the CFAR detector sliding window, calculating the interference parameters within the CFAR detector sliding window at any position, and obtaining the CFAR detection result of the equalized SAR image according to the following discriminant relationship,

[0033]

[0034] Among them, S cfar (m,n) and S Eq (m,n) represent the pixel values ​​of the detection result and the equalized SAR image at the position (m,n), μ clutter (m,n) and σ clutter (m,n) represent the mean and standard deviation of the interference within the sliding window of the CFAR detector at position (m,n), respectively.

[0035] Specifically, S3 includes the following steps:

[0036] S301, construct structural elements, and perform CFAR detection on the result S cfar Performing a morphological operation, wherein the morphological operation includes a dilation operation and an erosion operation;

[0037] S302, expansion operation process: construct a radius of r dil The circular structural element of the structure element has the origin as the center of the circle and the corresponding matrix is ​​S dil , whose size is (2r dil +1)×(2r dil +1), and

[0038]

[0039] The structural element is used to analyze the CFAR detection result S cfar A dilation operation is performed to fill the black holes caused by low-value speckle noise in the target area and the missing target pixels, which can be expressed as:

[0040]

[0041] in, is the result after expansion, S cfar +S dil Indicates the use of structural element corresponding matrix S dilExpanded CFAR test results S cfar , S dil The origin is translated to S cfar Pixel (m,n) position, if S dil At the pixel (m,n) position and S cfar The intersection (∩) of is not an empty set The corresponding pixel (m, n) in the output image is assigned a value of 1, otherwise it is assigned a value of 0;

[0042] S303, corrosion operation process: construct a radius r ero The circular structural element of the structure element has the origin as the center of the circle and the corresponding matrix is ​​S ero , whose size is (2r ero +1)×(2r ero +1), and

[0043]

[0044] Using this structural element to dilate the result Then perform the corrosion operation. It can be expressed as:

[0045]

[0046] in, is the result after corrosion, Indicates the use of structural element corresponding matrix S ero CFAR test results after corrosion expansion S ero The origin is translated to Pixel (m,n) position, if S ero At pixel (m,n), it is completely contained exist If there is an overlap, the corresponding pixel (m,n) in the output image is assigned a value of 1, otherwise it is assigned a value of 0;

[0047] S304: Process the CFAR detection result through morphological operations to generate a binary image It is the SAR image navigation landmark segmentation mask, denoted as

[0048] Specifically, S4 includes the following steps:

[0049] S401, use IHS transformation to process the optical image to obtain three components: intensity, hue and saturation. The transformation formula is:

[0050]

[0051] Among them, R 0 , G0 and B 0 Represents the three components of the optical image R, G, B, I 0 、v 1 and v 2 They are the three components of optical image intensity, hue and saturation;

[0052] S402, using segmentation mask M sar Extract the SAR image S to be fused from the SAR image F , the expression is:

[0053]

[0054] Among them, ⊙ represents S Eq and M sar Multiply the elements at corresponding positions, and S Eq and M sar The value of the pixel at position (m,n);

[0055] S403: The SAR image to be fused S F After superimposing with the grayscale optical image I, the intensity component I is replaced 0 , and then through the inverse IHS transformation, the fused image is finally obtained. The specific expression is:

[0056]

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] The present invention is oriented to the demand of visual navigation, fully utilizes the complementary information of optical and SAR images, pre-processes the optical image and the SAR image, uses the CFAR detection algorithm to detect the navigation landmarks at the pixel level of the SAR image, and then uses the morphological operation to process the detection result of the SAR image to obtain the SAR image navigation landmark segmentation mask, uses the segmentation mask to extract the pixels to be fused from the SAR image and fuses them with the optical image, and realizes the transmission of spectral information through the IHS transformation to obtain a fused image with richer content, so that suitable and reliable ground objects can be stably and efficiently extracted from the image as navigation landmarks, which is convenient for the subsequent processing and use of the image by visual navigation.

Brief Description of the Drawings

[0059] Figure 1 The overall framework diagram of the optical and SAR image fusion method for visual navigation requirements of the present invention;

[0060] Figure 2A schematic diagram of a sliding window of a CFAR detector of an optical and SAR image fusion method for visual navigation requirements of the present invention;

[0061] Figure 3 A schematic diagram of the morphological dilation operation process of the optical and SAR image fusion method for visual navigation requirements of the present invention;

[0062] Figure 4 A schematic diagram of the morphological corrosion operation process of the optical and SAR image fusion method for visual navigation requirements of the present invention;

[0063] Figure 5 This is a diagram showing the fusion effects of different fusion algorithms in an embodiment of the optical and SAR image fusion method for visual navigation requirements of the present invention. [Specific implementation method]

[0064] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings.

[0065] See also Figure 1-Figure 5 As shown, the present invention provides an optical and SAR image fusion method for visual navigation requirements, comprising the following steps:

[0066] S1. Preprocess the optical image and the SAR image, convert the optical image into a grayscale image by using the average value method, and obtain the equalized SAR image by grayscale equalization processing. The optical image is an RGB color three-channel image, and the SAR image is a single-channel grayscale image;

[0067] S101, convert the optical image into a single-channel grayscale image using the average value method, and take the average value of the three channel components of the optical image as the grayscale value of the image, that is,

[0068]

[0069] Among them, R(m,n), G(m,n) and B(m,n) represent the pixel values ​​of the three components of the optical image R, G, and B at the position (m,n), respectively, and I(m,n) is the pixel value at the position (m,n) after the optical image is grayed;

[0070] S102. Since the spectral grayscale of the SAR image and the optical image is quite different, grayscale equalization is required before fusion so that the input images have the same grayscale mean and standard deviation to avoid serious spectral distortion. Grayscale equalization is performed on the SAR image. The SAR image after grayscale equalization can be expressed as:

[0071]

[0072] Where μ() and σ() represent the mean and standard deviation of the image pixel values, respectively, S is the initial SAR image, S Eq is the SAR image after equalization, and I is the optical image processed by grayscale;

[0073] S103: Obtained pre-processed optical image I and SAR image S Eq .

[0074] S2. Using CFAR (Constant False Alarm Rate) detection algorithm to detect navigation landmarks at pixel level in SAR images.

[0075] S201, design a CFAR detector, determine the detector parameters for the grayscale balanced SAR image S Eq Perform CFAR target detection;

[0076] like Figure 2 As shown in the figure, the CFAR detector traverses all pixels in the entire SAR image through a sliding window. The sliding window is a hollow window. The sliding window includes a protection window and a background window. The background window is used to count background clutter to calculate the target detection threshold. The protection window is used to prevent the target pixels from leaking into the background window, which affects the correctness of the background clutter statistics.

[0077] The CFAR detector also includes a target area and an interference area. The size of the protection window and the background window are determined by the given target area size and interference area width. Assuming that the size of the target area is W×H, where W is the width of the target area and H is the height of the target area, then the size of the protection window is L protection ×L protection , where L protection It can be expressed as:

[0078] L protection =2max(M,N)+1 (3);

[0079] Assume that the width of the interference zone is L clutter , then the size of the background window is L background ×L background , where L background It can be expressed as:

[0080] L background =L protection +2L clutter (4);

[0081] S202, setting false alarm rate P fa The false alarm rate is pre-set according to the SAR image to be processed. fa Calculate the CFAR detection threshold εth , the expression is:

[0082]

[0083] S203, assuming that the interference pixels satisfy the Rayleigh distribution, estimate the interference parameters in the interference area by combining the interference statistical model of the Rayleigh distribution, including the mean μ clutter and standard deviation σ clutter , the expressions are:

[0084] μ clutter =μ(S Eq (A)) (6),

[0085] σ clutter =σ(S Eq (A)) (7),

[0086] Among them, A represents the interference-preserving area;

[0087] S204, traversing all positions of the equalized SAR image through the CFAR detector sliding window, calculating the interference parameters within the CFAR detector sliding window at any position, and obtaining the CFAR detection result of the equalized SAR image according to the following discriminant relationship,

[0088]

[0089] Among them, S cfar (m,n) and S Eq (m,n) represent the pixel values ​​of the detection result and the equalized SAR image at the position (m,n), μ clutter (m,n) and σ clutter (m,n) represent the mean and standard deviation of the interference within the sliding window of the CFAR detector at position (m,n), respectively.

[0090] S3, using morphological operations to process the SAR image CFAR detection results, remove false alarm targets and interference pixels, and obtain the SAR image navigation landmark segmentation mask;

[0091] S301, construct structural elements, and perform CFAR detection on the result S cfar Performing morphological operations, which in this embodiment include dilation operations and erosion operations, and in other embodiments may include opening operations and closing operations, wherein the opening operation is first erosion and then dilation, and the closing operation is first dilation and then erosion;

[0092] S302, expansion operation process: construct a radius of r dil The circular structural element of , the origin of the structural element is the center of the circle, such as Figure 3 The structural element shown is a circle with a radius r dil= 1, the corresponding matrix is ​​S dil , whose size is (2r dil +1)×(2r dil +1), and

[0093]

[0094] The structural element is used to analyze the CFAR detection result S cfar A dilation operation is performed to fill the black holes caused by low-value speckle noise in the target area and the missing target pixels, which can be expressed as:

[0095]

[0096] in, is the result after expansion, S cfar +S dil Indicates the use of structural element corresponding matrix S dil Expanded CFAR test results S cfar , S dil The origin is translated to S cfar Pixel (m,n) position, if S dil At the pixel (m,n) position and S cfar The intersection (∩) of is not an empty set The corresponding pixel (m, n) in the output image is assigned a value of 1, otherwise it is assigned a value of 0;

[0097] S303, corrosion operation process: construct a radius r ero The circular structural element of , the origin of the structural element is the center of the circle, such as Figure 4 The structural element shown is a circle with a radius r ero =2 Schematic diagram of the morphological corrosion operation process, the corresponding matrix is ​​S ero , whose size is (2r ero +1)×(2r ero +1), and

[0098]

[0099] Using this structural element to dilate the result Then perform the corrosion operation. It can be expressed as:

[0100]

[0101] in, is the result after corrosion, Indicates the use of structural element corresponding matrix S ero CFAR test results after corrosion expansion S eroThe origin is translated to Pixel (m,n) position, if S ero At pixel (m,n), it is completely contained exist If there is an overlap, the corresponding pixel (m,n) in the output image is assigned a value of 1, otherwise it is assigned a value of 0;

[0102] S304: Process the CFAR detection result through morphological operations to generate a binary image It is the SAR image navigation landmark segmentation mask, denoted as

[0103] S4. Use the segmentation mask to extract the pixels to be fused from the SAR image and fuse them with the optical image, and use the IHS (Intensity, Hue, Saturation, the RGB color system used in quantitative color processing on computers) transformation to realize the transmission of spectral information and obtain the fusion result of the optical and SAR images.

[0104] S401, use IHS transformation to process the optical image to obtain three components: intensity, hue and saturation. The transformation formula is:

[0105]

[0106] Among them, R 0 , G 0 and B 0 Represents the three components of the optical image R, G, B, I 0 、v 1 and v 2 They are the three components of optical image intensity, hue and saturation;

[0107] S402, using segmentation mask M sar Extract the SAR image S to be fused from the SAR image F , the expression is:

[0108]

[0109] Among them, ⊙ represents S Eq and M sar Multiply the elements at corresponding positions, and S Eq and M sar The value of the pixel at position (m,n);

[0110] S403: The SAR image to be fused S F After superimposing with the grayscale optical image I, the intensity component I is replaced 0, and then through the inverse IHS transformation, the fused image is finally obtained. The specific expression is:

[0111]

[0112] As a feasible embodiment of the present invention, drone aerial images of a certain area in Xi'an were selected as experimental data. The selected area contains targets such as water bodies, soil, vegetation, mountains, buildings and vehicles. The optical image is a true color image of three bands of R, G and B. The SAR image selects the HH polarization mode. The pixel size of the optical and SAR images is 512×512, and they have been registered. It is used as input for image fusion. The specific process is as follows:

[0113] Step 1: Preprocess the optical and SAR images, convert the optical image into a grayscale image using the average method, and obtain the equalized SAR image through grayscale equalization processing.

[0114] Step 2: Use the CFAR detection algorithm to detect navigation landmarks at the pixel level in the SAR image. In this embodiment, the target area size is set to 5×10, the false alarm rate is 0.04, and the CFAR detection result is obtained.

[0115] Step 3: Use morphological operations to process the SAR image CFAR detection results, remove false alarm targets and interference pixels, and obtain the SAR image navigation landmark segmentation mask. In this embodiment, a morphological erosion operation is used, and the radius r of the circular structure element is ero =1.

[0116] Step 4: Use the segmentation mask to extract the pixels to be fused from the SAR image and fuse them with the optical image, and use the IHS transformation to transfer the spectral information to obtain the fusion result of the optical and SAR images.

[0117] Finally, the fusion image quality results obtained in this embodiment are as follows. In order to further reflect the advantages of the present invention, the method of the present invention is compared with other traditional fusion algorithms. The quality of the fusion images obtained under the same experimental conditions is compared respectively. The results in Table 1 show that the method provided by the present invention is a DBF (Detection Based Fusion) algorithm, which is significantly better than other algorithms in terms of MI, PSNR, SSIM and NIQE.

[0118] Table 1 Evaluation results of image fusion using different algorithms

[0119] algorithm MI PSNR SSIM NIQE LP 1.826 -42.364 0.243 10.018 DTCWT 1.889 -42.382 0.241 9.981 NSCT 1.949 -42.367 0.248 10.360 HMSD 1.180 -37.126 0.415 8.543 WLS 2.852 -41.734 0.305 10.624 VSFF 2.606 -42.555 0.501 6.465 DBF 6.262 -18.385 0.546 3.964

[0120] also, Figure 5The fusion effects of different algorithms are given. In the area marked by the red box, the synthetic aperture radar image effectively highlights the target compared with the optical image. The fusion of optical and SAR images can effectively utilize the complementary advantages of the two images. Compared with other algorithms, the fusion results of the DBF algorithm proposed in the present invention are least affected by the SAR background noise. This is mainly because some other algorithms fail to fully suppress the background noise in SAR images, while others use suppression algorithms with limited effects. In contrast, the algorithm introduced in the present invention only focuses on the fusion detection results and discards all other data. This method greatly reduces the impact of SAR background noise on the fusion results. Figure 5 As shown in the yellow box in the middle, the main attributes of the optical image are better preserved. Therefore, the present invention can obtain better optical and SAR image fusion results for visual navigation needs.

Claims

1. An optical and SAR image fusion method for visual navigation needs, characterized in that: The following steps are involved: S1. Preprocessing the optical image and the SAR image, converting the optical image into a grayscale image by using an average value method, and obtaining a balanced SAR image by grayscale equalization processing, wherein the optical image is an RGB color three-channel image, and the SAR image is a single-channel grayscale image; S2, using CFAR detection algorithm to realize pixel-level navigation landmark detection of SAR images; S3, using morphological operations to process the SAR image CFAR detection results, remove false alarm targets and interference pixels, and obtain the SAR image navigation landmark segmentation mask; S4. Use the segmentation mask to extract the pixels to be fused from the SAR image and fuse them with the optical image, and use the IHS transformation to realize the transmission of spectral information to obtain the fusion result of the optical and SAR images.

2. The optical and SAR image fusion method for visual navigation requirements according to claim 1 is characterized in that: The pre-processing in S1 comprises the following steps: S101, converting the optical image into a single-channel grayscale image by using the average value method, and taking the average value of the three channel components of the optical image as the grayscale value of the image, that is, Among them, R(m,n), G(m,n) and B(m,n) represent the pixel values ​​of the three components of the optical image R, G, and B at the position (m,n), respectively, and I(m,n) is the pixel value at the position (m,n) after the optical image is grayed; S102, grayscale equalization is performed on the SAR image. The SAR image after grayscale equalization can be expressed as: Where μ() and σ() represent the mean and standard deviation of the image pixel values, respectively, S is the initial SAR image, S Eq is the SAR image after equalization, and I is the optical image processed by grayscale; S103: Obtained pre-processed optical image I and SAR image S Eq .

3. The optical and SAR image fusion method for visual navigation requirements according to claim 2 is characterized in that: The S2 comprises the following steps: S201, design CFAR detector, grayscale balanced SAR image S Eq Perform CFAR target detection; The CFAR detector traverses all pixels in the entire SAR image through a sliding window, the sliding window is a hollow window, the sliding window includes a protection window and a background window, the background window is used to count background clutter to calculate the target detection threshold, and the protection window is used to prevent the target pixel from leaking into the background window, thereby affecting the correctness of the background clutter statistics; The CFAR detector also includes a target area and an interference area. The size of the protection window and the background window is determined by the given target area size and interference area width. Assuming that the size of the target area is W×H, where W is the width of the target area and H is the height of the target area, then the size of the protection window is L protection ×L protection , where L protection It can be expressed as: L protection =2max(M,N)+1 (3); Assume that the width of the interference zone is L clutter , then the size of the background window is L background ×L background , where L background It can be expressed as: L background =L protection +2L clutter (4); S202, setting false alarm rate P fa , according to the false alarm rate P fa Calculate the CFAR detection threshold ε th , the expression is: S203, assuming that the interference pixels satisfy the Rayleigh distribution, estimate the interference parameters in the interference area by combining the interference statistical model of the Rayleigh distribution, including the mean μ clutter and standard deviation σ clutter , the expressions are: m clutter =μ(S Eq (A)) (6) s clutter =σ(S Eq (A)) (7); Where A represents the interference area, S Eq (A) represents the pixel value of the interference area image after equalization; S204, traversing all positions of the equalized SAR image through the CFAR detector sliding window, calculating the interference parameters within the CFAR detector sliding window at any position, and obtaining the CFAR detection result of the equalized SAR image according to the following discriminant relationship, Among them, S cfar (m,n) and S Eq (m,n) represent the pixel values ​​of the detection result and the equalized SAR image at the position (m,n), μ clutter (m,n) and σ clutter (m,n) represent the mean and standard deviation of the interference within the sliding window of the CFAR detector at position (m,n), respectively.

4. The optical and SAR image fusion method for visual navigation requirements according to claim 3 is characterized in that: The S3 comprises the following steps: S301, construct structural elements, and perform CFAR detection on the result S cfar Performing a morphological operation, wherein the morphological operation includes a dilation operation and an erosion operation; S302, expansion operation process: construct a radius of r dil The circular structural element of the structure element has the origin as the center of the circle and the corresponding matrix is ​​S dil , whose size is (2r dil +1)×(2r dil +1), and The structural element is used to analyze the CFAR detection result S cfar A dilation operation is performed to fill the black holes caused by low-value speckle noise in the target area and the missing target pixels, which can be expressed as: in, is the result after expansion, S cfar +S dil Indicates the use of structural element corresponding matrix S dil Expanded CFAR test results S cfar , S dil The origin is translated to S cfar Pixel (m,n) position, if S dil At the pixel (m,n) position and S cfar The intersection (∩) of is not an empty set The corresponding pixel (m, n) in the output image is assigned a value of 1, otherwise it is assigned a value of 0; S303, corrosion operation process: construct a radius r ero The circular structural element of the structure element has the origin as the center of the circle and the corresponding matrix is ​​S ero , whose size is (2r ero +1)×(2r ero +1), and Using this structural element to dilate the result Then the corrosion operation can be expressed as: in, is the result after corrosion, Indicates the use of structural element corresponding matrix S ero CFAR test results after corrosion expansion S ero The origin is translated to Pixel (m,n) position, if S ero At pixel (m,n), it is completely contained exist If there is an overlap, the corresponding pixel (m,n) in the output image is assigned a value of 1, otherwise it is assigned a value of 0; S304: Process the CFAR detection result through morphological operations to generate a binary image It is the SAR image navigation landmark segmentation mask, denoted as 5. The optical and SAR image fusion method for visual navigation requirements according to claim 4, characterized in that: The S4 comprises the following steps: S401, use IHS transformation to process the optical image to obtain three components: intensity, hue and saturation. The transformation formula is: Among them, R0, G0 and B0 represent the three components of the optical image R, G, and B respectively, and I0, v1 and v2 represent the three components of the optical image intensity, hue and saturation respectively; S402, using segmentation mask M sar Extract the SAR image S to be fused from the SAR image F , the expression is: Among them, ⊙ represents S Eq and M sar Multiply the elements at corresponding positions, and S Eq and M sar The value of the pixel at position (m,n); S403, the SAR image to be fused S F After superimposing with the grayscale optical image I to replace the intensity component I0, and then undergoing inverse IHS transformation, the fused image is finally obtained. The specific expression is:

Images