Method for Identifying Surface Damage in Disaster Areas Based on the Fractal Dimension Change of SAR Image Texture Information

Abstract

The present invention relates to the field of remote sensing geoscience, and particularly relates to a method for identifying damaged areas on the surface of a disaster area based on the fractal dimension change of SAR image texture information. The method for identifying damaged areas on the surface of a disaster area includes the following steps: S1, preprocessing the pre-disaster and post-disaster SAR images; S2, calculating the multi-scale fractal dimension of the SAR image texture information based on a multi-order neighborhood; S3, detecting the multi-scale image fractal dimension change; S4, judging the consistency of the multi-scale fractal dimension change detection results; S5, mapping the spatial distribution and grading of the damaged areas on the surface. The present invention makes full use of the scale invariance of the SAR image structure texture information characteristics of ground objects, and identifies the damaged areas on the surface of the disaster area based on the calculation of the texture fractal dimension change of the two-phase pre-disaster and post-disaster SAR images by STFCM. The STFCM method can quickly and accurately identify the damaged areas on the surface of the disaster area and local key areas, providing technical support for remote sensing mapping of the affected area, rapid assessment of the disaster situation, and decision-making for emergency disaster relief response.

Description

Technical Field

[0001] The present invention relates to a method for identifying surface damage in disaster areas based on the fractal dimension change of SAR image texture information, belonging to the field of remote sensing geoscience. Background Art

[0002] Natural disasters such as earthquakes, floods, landslides, debris flows, typhoons, forest fires, etc. and wars are common challenges faced by human society. Major disasters and wars have caused damage to ground objects and building collapses; rapid and reliable remote sensing identification of disaster area damage and building collapses is extremely important for emergency response and disaster reduction. However, disaster areas are often affected by clouds, fog, smoke and dust, etc., making it difficult to obtain clear high-resolution surface images in the first time after the disaster. To win the golden 72 hours of post-disaster rescue, rapid and efficient assessment of ground object damage and building collapses has become an important topic of common concern in the current scientific and technological community and society.

[0003] In previous studies, domestic and foreign researchers have proposed many surface change detection methods based on remote sensing technology, which are divided into three categories according to data sources: change detection based on new and old remote sensing images, change monitoring based on new remote sensing images and old non-remote sensing images, and three-dimensional change detection based on stereo image pairs; divided into three scales of pixel level, object level, and scene level according to the scale of the processing object; divided into two categories of direct comparison and post-classification comparison according to whether they have been classified. Among them, the automatic disaster identification method based on the texture change of high-resolution images is widely used in automatic disaster remote sensing identification. Such automatic disaster identification methods based on optical images are becoming more and more popular and have achieved good results.

[0004] However, the existing automatic disaster identification methods based on optical images are extremely vulnerable to cloud contamination, which greatly limits their detection efficiency and time. In addition, there are often clouds, fog, smoke and dust after the disaster, which have a huge impact on the image quality, and this also limits the practical application of such methods. With the rise of microwave remote sensing technology, microwave remote sensing images can effectively overcome the influence of interference factors such as clouds, bringing the possibility of rapid and efficient surface change detection. Due to the huge differences between microwave SAR images and optical images, the existing methods cannot overcome the defects of SAR images and are difficult to give full play to the advantages of SAR images, making rapid and efficient disaster identification stagnate again. The main difficulties are as follows: 1) Affected by the imaging characteristics of SAR images, the resolution, gray level, etc. of pre-disaster and post-disaster SAR images are quite different, making it difficult to use them synergistically; 2) The texture information of SAR images is weak and the interference noise is serious, making it difficult to accurately obtain the true information of the surface scene; 3) After the disaster, the surface structure is chaotic and there are many interference factors, and it is impossible to give a consistent and reliable surface damage identification result.

[0005] Therefore, it is necessary to develop a method for identifying surface damage in disaster areas based on the fractal dimension change of SAR image texture information to solve the above difficult problems. Summary of the Invention

[0006] The object of the present invention is to provide a method for identifying surface damage in disaster areas based on the fractal dimension change of SAR image texture information for surface damage detection, so as to solve the problems proposed in the background technology. Based on the scale invariance of the structural texture information characteristics of ground object SAR images, combined with a multi-scale and multi-order neighborhood sliding window, the present invention realizes the rapid assessment of the surface damage situation after the disaster. First, to solve the problem that it is difficult to synergistically use pre-disaster and post-disaster images caused by the difference in SAR imaging methods, the SAR image preprocessing method proposed by the present invention effectively overcomes the problem that it is difficult to synergistically use due to the large difference between pre-disaster and post-disaster SAR images. Secondly, aiming at the problems of weak SAR image texture information and serious noise interference, the present invention introduces the concept of neighborhood and proposes a multi-order neighborhood SAR texture fractal dimension calculation method based on nested windows, which effectively enhances the texture information of buildings in SAR images and weakens the interference of noise on SAR image information extraction. Finally, to solve the problem of uncertain results caused by the chaotic surface structure after the disaster, the present invention proposes a method for judging the monotonicity of fractal dimension change based on multiple scales. The result consistency analysis based on the monotonicity of multi-scale fractal dimension change can make the results more accurate and reliable.

[0007] To achieve the above object, the present invention provides a method for identifying surface damage in disaster areas based on the fractal dimension change of SAR image texture information, including the following steps:

[0008] Step S1. After obtaining the pre-disaster and post-disaster SAR images of the study area, preprocess the pre-disaster SAR image and the post-disaster SAR image. The steps of preprocessing include:

[0009] Step S1.1. Perform precise registration on the pre-disaster SAR image and the post-disaster SAR image. The precise registration includes picking up homologous image points and image correction;

[0010] Step S1.2. Based on the SAR image with higher resolution among the pre-disaster SAR image and the post-disaster SAR image, resample the SAR image with lower resolution;

[0011] Step S1.3. Normalize the image gray values of the pre-disaster SAR image and the post-disaster SAR image;

[0012] Step S1.4. Determine the study area according to the spatial ranges of the pre-disaster SAR image and the post-disaster SAR image, and crop the pre-disaster SAR image and the post-disaster SAR image according to the study area range to obtain the SAR image data block of the study area;

[0013] Step S2. Based on the SAR image data block obtained by preprocessing in Step S1, calculate the multi-scale fractal dimension information of the SAR image texture information of the multi-order neighborhood before and after the disaster respectively. The specific steps are as follows:

[0014] Step S2.1: Calculate the single-scale multi-order neighborhood SAR image ruggedness for the SAR image data block;

[0015] Step S2.2: Fit the single-scale multi-order neighborhood ruggedness and the order size calculated in Step S2.1 to obtain the fractal dimension at this scale;

[0016] Step S2.3: Repeat Step S2.1 and Step S2.2 to calculate and obtain the multi-scale multi-order neighborhood fractal dimensions;

[0017] Step S3: Detect the multi-scale fractal dimension changes in the pre-disaster and post-disaster SAR images based on the multi-scale multi-order neighborhood fractal dimension results of the pre-disaster and post-disaster SAR images obtained in Step S2;

[0018] Step S4: Conduct a consistency study on the fractal dimension changes in the SAR images for the multi-scale fractal dimension change detection results obtained in Step S3. The specific steps are as follows:

[0019] Step S4.1: Determine the scale for calculating the monotonicity of the fractal dimension change in the SAR image, and calculate the monotonicity of the fractal dimension change in the SAR image according to the determined scale;

[0020] Step S4.2: Based on the results of the monotonicity of the fractal dimension change in the SAR image, conduct a consistency study on the results of the multi-scale fractal dimension changes in the SAR image;

[0021] Step S5: Obtain the spatial distribution and classification mapping of the surface damage.

[0022] Furthermore, in Step S2.1, taking any pixel c in the study area as the initial pixel, the formula for calculating the single-scale multi-order neighborhood SAR image ruggedness of pixel c is:

[0023] D c(i) = f[p c , p c(i) , i = 1, 2, 3,..., n

[0024] where p c is the gray value of pixel c, p c(i) is the set of gray values of the pixels in the i-th order neighborhood centered on pixel c, f is the function for calculating the ruggedness between pixel c and its i-th order neighborhood pixels, D c(i) is the ruggedness of pixel c under the i-th order neighborhood, and n is the total number of orders of the multi-order neighborhood.

[0025] Furthermore, the formula for fitting the single-scale multi-order neighborhood ruggedness and the order size in Step S2.2 is:

[0026]

[0027] Among them, R c(i) is the size of the square area surrounded by the i-th order neighborhood of pixel c, and k is the slope of the linear relationship between the multi-order neighborhood ruggedness and the order size obtained by least squares fitting. is the mean value of the ruggedness set of pixel c in the multi-order neighborhood, is the mean value of the size set of the multi-order neighborhood of pixel c.

[0028] Furthermore, the calculation formula for the multi-scale multi-order neighborhood fractal dimension in step S2.3 is:

[0029]

[0030] Among them, R c(i) is the size of the square area surrounded by the i-th order neighborhood of pixel c, is the mean value of the ruggedness set of pixel c in the multi-order neighborhood, is the mean value of the size set of the multi-order neighborhood of pixel c, m < n, j is the scale, m represents the multi-scale dimension, and F c(j) is the multi-order neighborhood fractal dimension of pixel c at scale j; calculate all values of j, and then obtain the multi-scale multi-order neighborhood fractal dimension F c .

[0031] Furthermore, in step S3, the multi-scale fractal change between the pre-disaster SAR image and the post-disaster SAR image is detected according to the following formula:

[0032] ΔF c(j) = F c(j)，2 - F c(j)，1

[0033] Among them, F c(j)，1 is the multi-order neighborhood fractal dimension of pixel c in the pre-disaster SAR image at scale j, F c(j)，2 is the multi-order neighborhood fractal dimension of pixel c in the post-disaster SAR image at scale j, and ΔF c(j) is the change value of the fractal dimension before and after the disaster of pixel c at scale j.

[0034] Furthermore, in step S4.1, according to the characteristics of the ground objects in the study area, the value range of j is determined to be [1, m], and m < n. The monotonicity of the fractal change is calculated according to the following formula:

[0035]

[0036] Among them, ΔF c is the set of multi-scale multi-order neighborhood fractal changes of pixel c, and T (c) is the logical value calculated for the monotonicity of the multi-scale multi-order neighborhood fractal change of pixel c. If the value is 1, it is monotonic; if the value is 0, it is non-monotonic.

[0037] Furthermore, in step S4.2, the consistency judgment of the multi-scale fractal change results of the pre-disaster SAR image and the post-disaster SAR image is carried out according to the following formula:

[0038] B (c) = T (c) * ΔF c (a)

[0039] where T (c) is the calculation logic value of the monotonicity of the multi-scale multi-order neighborhood fractal change of pixel c, a is the relatively optimal scale obtained according to requirements, and ΔF c(a) is the multi-order neighborhood fractal change value of pixel c at scale a, and B (c) is the ground object damage judgment value of pixel c.

[0040] Furthermore, the specific steps of step S5 are as follows: Slide pixel c to the next adjacent new pixel, repeat steps S2 to S4, and calculate the ground object damage judgment value at the new pixel position; traverse the entire map to obtain the final map of the spatial distribution and classification of surface damage.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] (1) The method for identifying surface damage in the disaster area of the present invention integrates the characteristics of active SAR images, effectively overcomes the obstacles to disaster monitoring caused by clouds, rain, smoke and dust, and successfully realizes the efficient and rapid assessment of the surface damage situation after the disaster, providing technical support for remote sensing detection and emergency response of the disaster area and disaster intensity. The STFCM method does not require manual intervention, independently processes and automatically calculates, and is expected to realize the automatic identification and range extraction of the surface damage location of the on-orbit operating satellite.

[0043] (2) The method for identifying surface damage in the disaster area of the present invention proposes a SAR image preprocessing mode, which not only solves the problem of numerical drift caused by differences in homologous SAR images, but also effectively solves the problem of difficult collaborative use caused by dimensional differences in heterogeneous active SAR images, providing a new idea for the joint processing of multi-source SAR images in the future.

[0044] (3) The method for identifying surface damage in the disaster area of the present invention proposes a SAR fractal calculation method based on multi-order neighborhoods, which effectively enhances the SAR texture features of buildings and weakens the influence of noise. On this basis, by constructing a multi-scale fractal change monotonicity judgment method, the reliability of the results is enhanced. This method provides a new calculation idea for SAR texture fractal calculation, making it possible to identify surface damage based on the change of SAR texture fractal.

[0045] In addition to the purposes, features and advantages described above, the present invention has other purposes, features and advantages. The following will refer to the drawings to further elaborate on the present invention in detail. Description of the Drawings

[0046] The drawings are used to provide a further understanding of the embodiments of the present invention and form a part of the specification, which, together with the following specific implementation manners, is used to explain the embodiments of the present invention, but does not constitute a limitation to the embodiments of the present invention. In the drawings:

[0047] Figure 1 is a flowchart of the method for identifying damaged areas on the ground surface based on the fractal dimension change of SAR image texture information provided by the embodiments of the present invention;

[0048] Figure 2 is the Sentinel-1 GRD-level VV-polarized SAR image of the Tilkeoglu area in Turkey on February 4, 2023;

[0049] Figure 3 is the HISEA-1 ORG-level VV-polarized SAR image of the Tilkeoglu area in Turkey on February 6, 2023;

[0050] Figure 4 is a flowchart of the SAR image preprocessing in the method for identifying damaged areas on the ground surface based on the fractal dimension change of SAR image texture information provided by the embodiments of the present invention;

[0051] Figure 5 is a flowchart of the multi-scale fractal dimension calculation of the multi-order neighborhood SAR image texture in the overall process of the method for identifying damaged areas on the ground surface based on the fractal dimension change of SAR image texture information of the present invention;

[0052] Figure 6 is a flowchart of the consistency judgment of the multi-scale fractal dimension change detection results in the overall process of the method for identifying damaged areas on the ground surface based on the fractal dimension change of SAR image texture information of the present invention;

[0053] Figure 7 is a spatial distribution map of the damaged situation on the ground surface in the Tilkeoglu area after the M7.8 earthquake in Turkey on February 6, 2023;

[0054] Figure 8 is an optical image of the ground surface in the Tilkeoglu area after the M7.8 earthquake in Turkey on February 6, 2023. Specific Embodiment Manner

[0055] The following will elaborate on the embodiments of the present invention in detail with reference to the drawings, but the present invention can be implemented in many different ways defined and covered by the claims.

[0056] As Figure 1As shown in the figure, a method for identifying damage to the surface of the disaster area (including slopes, ground, vegetation, roads, buildings, etc.) based on the fractal change of SAR image texture information (abbreviated as STFCM, SAR Texture Fractal Change-based Model) in this embodiment includes the following steps:

[0057] Step S1: Obtain the pre-disaster SAR image and the post-disaster SAR image with the closest time before and after the disaster in the study area. In this embodiment, the pre-disaster SAR image selects the Sentinel-1 GRD-level VV-polarized SAR image of the Tilkeoglu region in Turkey on February 4, 2023 (as Figure 2 shown), and the post-disaster SAR image selects the HISEA-1 ORG-level VV-polarized SAR image of the Tilkeoglu region in Turkey after the earthquake on February 6, 2023 (as Figure 3 shown), and preprocess the pre-disaster and post-disaster SAR images to provide a data basis for the input of subsequent algorithms. The specific process of SAR image preprocessing is as Figure 4 shown; the SAR image preprocessing specifically includes the following steps:

[0058] Step S1.1: Fine registration of SAR images: Fine registration of the pre-disaster SAR image and the post-disaster SAR image. This fine registration includes picking corresponding image points and image correction. Since the damage detection of buildings has high requirements for the correspondence of ground objects, the fine registration of SAR images is necessary. Use ENVI5.3 software to effectively select corresponding image points and complete the fine registration of heterogeneous SAR images through the polynomial image correction method.

[0059] Step S1.2: Resampling of SAR images: Based on the SAR image with the higher resolution among the pre-disaster SAR image and the post-disaster SAR image, resample the SAR image with the lower resolution to ensure that the spatial resolutions of the pre-disaster and post-disaster SAR images are the same. In order to effectively mine the fractal dimension information of SAR image texture, in this embodiment, the SAR image with the lower resolution is resampled to 3m×3m based on the SAR image with the higher resolution.

[0060] Step S1.3: Normalization of image gray values: Normalize the gray values of the pre-disaster SAR image and the post-disaster SAR image; normalization is to eliminate the data deviation and dimensional difference caused by different sensor imaging methods of SAR images from different satellite platforms. To eliminate the dimensional difference between the pre-disaster SAR image and the post-disaster SAR image and effectively suppress the interference of ground object features caused by sensor imaging differences, the maximum value of the image normalization range should be less than or equal to 50% of the maximum order neighborhood side length of the multi-order neighborhood, that is, the two images are jointly normalized to [0, 10].

[0061] Step S1.4, Image Mosaic and Cropping: Determine the study area based on the spatial ranges of the pre-disaster SAR image and the post-disaster SAR image, and crop the pre-disaster and post-disaster SAR images according to the range of the study area to obtain the final SAR image data block of the study area.

[0062] Step S2. Based on the SAR image data blocks obtained by preprocessing in Step S1, calculate the multi-scale fractal dimension information of the pre-disaster and post-disaster multi-order neighborhood SAR image textures respectively. The calculation process is as Figure 5 shown, and specifically includes the following steps:

[0063] Step S2.1, Calculate the ruggedness of the single-scale multi-order neighborhood SAR image for the SAR image data block.

[0064] Step S2.2, Fit the single-scale multi-order neighborhood ruggedness calculated in Step S2.1 with the order size to obtain the fractal dimension at this scale.

[0065] Step S2.3, Repeat Step S2.1 and Step S2.2 to calculate the multi-scale multi-order neighborhood fractal dimension.

[0066] Step S3. Detect the multi-scale fractal dimension changes of the pre-disaster and post-disaster SAR images based on the multi-scale multi-order neighborhood fractal dimension results of the pre-disaster and post-disaster SAR images obtained in Step S2.

[0067] Step S4. Conduct a consistency study on the fractal dimension changes of the SAR images for the multi-scale fractal dimension change detection results obtained in Step S3, as Figure 6 shown, and specifically includes the following steps:

[0068] Step S4.1, Determine the scale for calculating the monotonicity of the fractal dimension change of the SAR image, and calculate the monotonicity of the fractal dimension change of the SAR image according to the determined scale.

[0069] Step S4.2, Based on the monotonicity result of the fractal dimension change of the SAR image, conduct a consistency study on the multi-scale fractal dimension change results of the SAR image.

[0070] Step S5. Obtain the final spatial distribution and classification mapping of the surface damage.

[0071] In this embodiment, in Step S2.1, taking any pixel c in the study area as the initial pixel, the formula for calculating the ruggedness of the single-scale multi-order neighborhood SAR image of the pixel c is:

[0072] D c(i) =f[p c ,p c(i) ,i=1,2,3,...,n

[0073] where, p cis the gray value of pixel c, p c(i) is the set of gray values of the i-th order neighborhood pixels centered on pixel c, f is the ruggedness function for calculating the ruggedness between pixel c and its i-th order neighborhood pixels, D c(i) is the ruggedness under the i-th order neighborhood of pixel c, and n is the total order number of the multi-order neighborhood.

[0074] Specifically, step S2.1 is as follows: The single-scale multi-order neighborhood SAR image ruggedness D is calculated by the prism surface area method, and the multi-order neighborhood prism surface area within the sliding window is calculated by the splitting method. This part is implemented through Matlab programming. The specific calculation process of the multi-order neighborhood prism surface area is as follows:

[0075] Take the gray values of the four corner points (A, B, C, D) and the center point of a certain order neighborhood square centered on the center point (O) within the sliding window as the heights of each point of the rugged prism at the top, and calculate the surface area of the top of the prism. Connect the center point and the corner points, divide the top into four sloping triangles, and calculate the sum of the areas of the four sloping triangles, which is the surface area of the top of the sliding window of this size.

[0076] Step S2.1.1: Calculate the surface area of a single triangle. Taking triangle OAB as an example, its specific calculation formula is:

[0077]

[0078] where q is the perimeter of triangle OAB, l OA is the side length of OA, l OB is the side length of OB, l AB is the side length of AB.

[0079]

[0080] where S OAB is the area of triangle OAB.

[0081] Step S2.1.2: Calculate the surface area of the top of the prism. Its specific calculation formula is:

[0082] S i = S OAB + S OBC + S OCD + S ODA

[0083] where S i is the surface area of the top of the prism formed by the grid of the i-th order neighborhood, S OAB is the area of triangle OAB, S OBC is the area of triangle OBC, S OCD is the area of triangle OCD, S ODA is the area of triangle ODA.

[0084] Step S2.1.3: Calculate the surface area of the tops of other prisms in different - order neighborhoods.

[0085] In this embodiment, in step S2.2, the single - scale multi - order neighborhood ruggedness is fitted with the order size, and the fitting calculation formula is:

[0086]

[0087] where D c(i) is the ruggedness of pixel c in the i - th order neighborhood, which is represented by the surface area of the top of the prism in the i - th order neighborhood in this embodiment; R c(i) is the size of the square area surrounded by the i - th order neighborhood of pixel c, which is represented by the square of the side length of the i - th order neighborhood in this embodiment; n is the total number of orders of the multi - order neighborhood; n = 9 in this embodiment; k is the slope of the linear relationship between the multi - order neighborhood ruggedness and the order size obtained by least - squares fitting; is the mean value of the ruggedness set of pixel c in the multi - order neighborhood, which is represented by the mean value of the surface area set of the tops of the prisms in the multi - order neighborhood in this embodiment; is the mean value of the size set of the multi - order neighborhood of pixel c, which is represented by the mean value of the square of the side - length set of the multi - order neighborhood in this embodiment.

[0088] In this embodiment, the calculation formula for the multi - scale multi - order neighborhood fractal dimension in step S2.3 is:

[0089]

[0090] where D c(i) is the ruggedness of pixel c in the i - th order neighborhood, which is represented by the surface area of the top of the prism in the i - th order neighborhood in this embodiment; R c(i) is the size of the square area surrounded by the i - th order neighborhood of pixel c, which is represented by the square of the side length of the i - th order neighborhood in this embodiment; n is the total number of orders of the multi - order neighborhood; n = 9 in this embodiment, j is the scale, and j = 1, 2, 3 in this embodiment; is the mean value of the ruggedness set of pixel c in the multi - order neighborhood, which is represented by the mean value of the surface area set of the tops of the prisms in the multi - order neighborhood in this embodiment; is the mean value of the size set of the multi - order neighborhood of pixel c, which is represented by the mean value of the square of the side - length set of the multi - order neighborhood in this embodiment; F c(j) is the multi - order neighborhood fractal dimension of pixel c at scale j.

[0091] In step S3 of this embodiment, the multi - scale fractal change between the pre - disaster SAR image and the post - disaster SAR image is detected according to the following formula

[0092] ΔF c(j) =Fc(j),2 -F c(j),1

[0093] Among them, F c(j),1 is the multi-order neighborhood fractal dimension of the pre-disaster Sentinel-1 image pixel c at scale j, and F c(j)，2 is the multi-order neighborhood fractal dimension of the post-disaster TY image pixel c at scale j, and ΔF c(j) is the change value of the fractal dimension before and after the disaster for pixel c at scale j.

[0094] In this embodiment, in step S4.1, to determine the scale for calculating the monotonicity of the fractal dimension change, calculate the monotonicity of the fractal dimension change. According to the characteristics of the ground objects in the study area, the value range of j is determined to be [1, m], and m < n. Calculate the monotonicity of the fractal dimension change according to the following formula:

[0095]

[0096] Among them, ΔF c is the multi-scale multi-order neighborhood fractal dimension change set of pixel c, and T (c) is the calculation logic value of the monotonicity of the multi-scale multi-order neighborhood fractal dimension change of pixel c. If the value is 1, it is monotonic; if the value is 0, it is non-monotonic.

[0097] In step S4.2 of this embodiment, conduct a consistency study on the multi-scale fractal dimension change results of the pre-disaster SAR image and the post-disaster SAR image according to the following formula:

[0098] B (c) = T (c) * ΔF c(a)

[0099] Among them, T (c) is the calculation logic value of the monotonicity of the multi-scale multi-order neighborhood fractal dimension change of pixel c, a is the relatively optimal scale obtained according to requirements, and ΔF c(a) is the multi-order neighborhood fractal dimension change of pixel c at scale a, and B (c) is the ground object damage judgment value of pixel c.

[0100] In step S5 of this embodiment, the specific process of obtaining the spatial distribution and classification mapping of surface damage is as follows:

[0101] Slide pixel c to the next adjacent new pixel, repeat steps S2 to S4, calculate the ground object damage judgment value at the new pixel position, traverse the entire map (726 rows, 458 columns), and obtain the final spatial distribution and classification mapping of surface damage, as Figure 7 shown. By comparing with the post-earthquake surface optical image of the Tirk Oglu region in Figure 8 , it can be seen that Figure 7The spatial distribution of the surface damage situation in the Tilkoğlu area after the earthquake obtained is consistent with the actual situation.

[0102] Through the proposed SAR image preprocessing process, the method of the present invention effectively solves the problem of collaborative use of heterogeneous SAR images; by proposing a multi-order neighborhood SAR image texture fractal dimension calculation method, it effectively enhances the weak SAR texture information of buildings and can effectively overcome the interference brought by SAR noise to the calculation of texture information; by constructing a multi-scale fractal dimension change monotonicity test method, the rationality of the fractal dimension change result can be evaluated, providing more reliable disaster damage data support for post-disaster rescue and emergency decision-making.

[0103] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for identifying surface damage in disaster areas based on the fractal dimension change of SAR image texture information, characterized in that, It includes the following steps: Step S1: After obtaining the pre-disaster and post-disaster SAR images of the study area, preprocess the pre-disaster SAR image and the post-disaster SAR image. The steps of preprocessing include: Step S1.1: Perform precise registration on the pre-disaster SAR image and the post-disaster SAR image. The precise registration includes picking corresponding image points and image correction; Step S1.2: Taking the SAR image with the higher resolution among the pre-disaster SAR image and the post-disaster SAR image as the benchmark, resample the SAR image with the lower resolution; Step S1.3: Normalize the image gray values of the pre-disaster SAR image and the post-disaster SAR image; Step S1.4: Determine the study area according to the spatial ranges of the pre-disaster SAR image and the post-disaster SAR image, and crop the pre-disaster SAR image and the post-disaster SAR image according to the study area range to obtain the SAR image data blocks of the study area; Step S2: Based on the SAR image data blocks obtained by preprocessing in Step S1, calculate the multi-scale fractal dimension information of the SAR image texture information of the multi-order neighborhood before and after the disaster respectively. The specific steps are as follows: Step S2.1: Calculate the roughness of the single-scale multi-order neighborhood SAR image for the SAR image data blocks; Step S2.2: Fit the single-scale multi-order neighborhood roughness calculated in Step S2.1 with the order size to obtain the fractal dimension at this scale; Step S2.3: Repeat Step S2.1 and Step S2.2 to calculate and obtain the multi-scale multi-order neighborhood fractal dimension; Step S3: Detect the multi-scale fractal dimension changes of the pre-disaster and post-disaster SAR images based on the results of the multi-scale multi-order neighborhood fractal dimensions of the pre-disaster and post-disaster SAR images obtained in Step S2; Step S4: Conduct a consistency study on the fractal dimension change detection results of the multi-scale fractal dimension obtained in Step S3. The specific steps are as follows: Step S4.1: Determine the scale for calculating the monotonicity of the fractal dimension change of the SAR image, and calculate the monotonicity of the fractal dimension change of the SAR image according to the determined scale; Step S4.2: Based on the results of the monotonicity of the fractal dimension change of the SAR image, conduct a consistency study on the results of the multi-scale fractal dimension change of the SAR image; Step S5: Obtain the spatial distribution and classification mapping of the surface damage.

2. The method for identifying surface damage in disaster areas according to claim 1, wherein, In Step S2.1, taking any pixel c in the study area as the initial pixel, the calculation formula for the roughness of the single-scale multi-order neighborhood SAR image of the pixel c is: D c(i) = f[p c , p c(i) , i = 1, 2, 3, ..., n Among them, p c is the gray value of pixel c, p c(i) is the set of gray values of the i-th order neighborhood pixels centered on pixel c, f is the ruggedness function for calculating the ruggedness between pixel c and its i-th order neighborhood pixels, D c(i) is the ruggedness under the i-th order neighborhood of pixel c, and n is the total order number of the multi-order neighborhood.

3. The method for identifying surface damage in disaster areas according to claim 2, wherein The calculation formula for fitting the single-scale multi-order neighborhood roughness with the order size in Step S2.2 is: Among them, R c(i) is the size of the square area surrounded by the i-th order neighborhood of pixel c, and k is the slope of the linear relationship between the multi-order neighborhood ruggedness and the order size obtained by least squares fitting. is the mean value of the ruggedness set of pixel c under the multi-order neighborhood. is the mean value of the size set of the multi-order neighborhood of pixel c.

4. The method for identifying surface damage in disaster areas according to claim 2, characterized in that, The calculation formula for the multi-scale multi-order neighborhood fractal dimension in Step S2.3 is: Among them, R c(i) is the size of the square area surrounded by the i-th order neighborhood of pixel c, is the mean value of the roughness set of pixel c under the multi-order neighborhood, is the mean value of the size set of the multi-order neighborhood of pixel c, m < n, j is the scale, m represents the multi-scale dimension, F c(j) is the multi-order neighborhood fractal dimension of pixel c at scale j; calculate all values of j, and then obtain the multi-scale multi-order neighborhood fractal dimension F c .

5. The method for identifying surface damage in disaster areas according to claim 4, wherein In Step S3, the multi-scale fractal dimension changes of the pre-disaster SAR image and the post-disaster SAR image are detected according to the following formula: ΔF c(j) =F c(j),2 -F c(j),1 Among them, F c(j),1 is the multi-order neighborhood fractal dimension of pixel c in the pre-disaster SAR image at scale j, and F c(j)2 is the multi-order neighborhood fractal dimension of pixel c in the post-disaster SAR image at scale j. ΔF c(j) is the change value of the fractal dimension before and after the disaster for pixel c at scale j.

6. The method for identifying surface damage in the disaster area according to claim 5, wherein, In Step S4.1, according to the characteristics of the ground objects in the study area, determine that the value range of j is [1, m], and m < n, and calculate the monotonicity of the fractal dimension change according to the following formula: Among them, ΔF c is the multi-scale multi-order neighborhood fractal dimension change set of pixel c, and T (c) is the logical value calculated for the monotonicity of the multi-scale multi-order neighborhood fractal dimension change of pixel c. If the value is 1, it is monotonic; if the value is 0, it is non-monotonic.

7. The method for identifying surface damage in the disaster area according to claim 6, wherein, In Step S4.2, the consistency of the multi-scale fractal dimension change results of the pre-disaster SAR image and the post-disaster SAR image is studied according to the following formula: B (c) = T (c) * ΔF c(a) Among them, T (c) is the calculation logic value of the monotonicity of the multi-scale and multi-order neighborhood fractal change of pixel c, a is the relatively optimal scale obtained according to requirements, and ΔF c(a) is the multi-order neighborhood fractal change value of pixel c at scale a, and B (c) is the judgment value of the damage of the ground object of pixel c.

8. The method for identifying surface damage in the disaster area according to claim 2, wherein The specific steps of step S5 are as follows: Slide pixel c to the next adjacent new pixel, repeat steps S2 to S4, and calculate the ground object damage judgment value at the new pixel position; Traverse the entire map to obtain the final map of the spatial distribution and classification of surface damage.

Images