A method for extracting seismic surface rupture based on color segmentation and object orientation

Abstract

The present application relates to a kind of based on color segmentation and object-oriented seismic surface rupture extraction method, comprising: cutting to seismic surface rupture image, obtain the image of region of interest;The image of region of interest is converted from RGB color space to HSV color space, obtain conversion area image;Based on the optimal principle of extracting surface rupture and removing background noise, using sliding bar color segmentation is carried out to conversion area image, obtain rough segmentation image;Based on the optimal principle of extracting surface rupture and removing background noise, using sliding bar is carried out to several sub-regions in rough segmentation image screening, obtain fine extraction result.The present application can quickly, accurately extract surface rupture.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a method for extracting seismic surface rupture based on color segmentation and object orientation. BACKGROUND

[0002] Earthquakes with a magnitude of 6.5 or above generally form a seismic surface rupture zone of several kilometers to several hundred kilometers on the surface. The geometric structure of the surface rupture contains the kinematic characteristics of the seismogenic structure. It not only provides important quantitative data for analyzing the spatial distribution of coseismic displacement of active faults and developing active fault avoidance, but also has important scientific significance for in-depth understanding of the rupture propagation law, deformation kinematics and dynamic mechanism of the shallow crust of the seismogenic fault. Therefore, quickly obtaining the seismic surface rupture after an earthquake and conducting fine research on the surface rupture are important basis for mastering the earthquake disaster and studying the properties of the seismogenic fault.

[0003] Currently, the seismic surface rupture is mainly obtained by field geological survey method and image visual interpretation method. However, both methods have certain limitations. The field geological survey method is time-consuming and labor-intensive, and many places cannot be directly reached due to complex topography, resulting in incomplete survey data. Although the image visual interpretation method can compensate for the defect of incomplete field geological survey data to some extent, it is still a labor-intensive and time-consuming work, and the extracted surface rupture is linear, which cannot obtain the fine features such as the width of the surface rupture.

[0004] With the development of unmanned aerial vehicle technology and aerial remote sensing technology, high-resolution images unaffected by clouds and fog can be obtained, providing conditions for the extraction of fine structures of coseismic surface rupture of large earthquakes. Due to the complex scene of high-resolution images, there are rich ground feature information and a large amount of noise. When using traditional pixel-based machine learning methods to extract information from high-resolution images, there are limitations such as insufficient use of image information, poor classification effect, and slow speed. In addition, deep learning still needs a large amount of labeled data set as the basis, and the labeled data is still a labor-intensive work. SUMMARY

[0005] Therefore, the present application provides a method for extracting seismic surface rupture based on color segmentation and object orientation, which can quickly and accurately extract the surface rupture.

[0006] To achieve the above purpose, the present application provides the following solutions:

[0007] A method for extracting seismic surface rupture based on color segmentation and object orientation, comprising:

[0008] cropping the seismic surface rupture image to obtain a region of interest image;

[0009] convert the region of interest image from an RGB color space to an HSV color space to obtain a converted region image;

[0010] based on an optimal principle of extracting a ground surface rupture and removing background noise, perform color segmentation on the converted region image using a sliding bar to obtain a coarse segmentation image;

[0011] based on the optimal principle of extracting the ground surface rupture and removing the background noise, perform screening on a plurality of sub-regions in the coarse segmentation image using the sliding bar to obtain a fine extraction result.

[0012] Preferably, the cropping of the seismic ground surface rupture image to obtain the region of interest image comprises:

[0013] drawing a polygon on the seismic ground surface rupture image displayed in the front end and generating a mask in the background;

[0014] calculating the polygon to obtain a four-to-range, and cropping to obtain a to-be-masked image based on the four-to-range;

[0015] performing convolution operation on the mask and the to-be-masked image to obtain the region of interest image.

[0016] Preferably, based on the optimal principle of extracting the ground surface rupture and removing the background noise, the color segmentation on the converted region image using the sliding bar to obtain the coarse segmentation image specifically comprises:

[0017] based on the optimal principle of extracting the ground surface rupture and removing the background noise, dynamically adjusting the maximum value of H, the minimum value of H, the maximum value of S, the minimum value of S, the maximum value of V, and the minimum value of V of the converted region image through the sliding bar, comparing before and after the adjustment in real time, and selecting the optimal result to obtain the coarse segmentation image.

[0018] Preferably, based on the optimal principle of extracting the ground surface rupture and removing the background noise, the screening on the plurality of sub-regions in the coarse segmentation image using the sliding bar to obtain the fine extraction result comprises:

[0019] performing outer contour extraction on each sub-region in the coarse segmentation image to obtain a sub-region set;

[0020] calculating constraint parameters of each sub-region in the sub-region set to obtain a constraint parameter set; the constraint parameter set comprises a spectrum standard deviation set, a spectrum mean value set, and a minimum circumscribed rectangle aspect ratio set;

[0021] The constraint parameter set is screened by dynamically adjusting the maximum value of spectral standard deviation, the minimum value of spectral standard deviation, the maximum value of spectral mean value and the minimum value of the length-width ratio of the minimum outer rectangle based on the principle of optimal extraction of surface rupture and removal of background noise, and the optimal result is selected by comparing before and after screening in real time, so that the fine extraction result is obtained.

[0022] Preferably, the coarse segmentation image is divided based on a contour tracking algorithm to obtain the sub-region set.

[0023] Preferably, the coarse segmentation image is divided based on a contour tracking algorithm to obtain the sub-region set, and specifically:

[0024] Based on the chain code principle, the contour points and their directions of each connected domain on the coarse segmentation image are searched, and are expressed by chain codes, and the inner and outer contours of each connected domain are marked, and the outer contour is taken to obtain the sub-region set.

[0025] According to the specific embodiments provided by the present application, the following technical effects are disclosed:

[0026] The present application relates to a kind of based on color segmentation and object-oriented seismic surface rupture extraction method, comprising: cutting to seismic surface rupture image, obtain the region of interest image;The region of interest image is converted from RGB color space to HSV color space, obtain conversion area image;Based on the principle of optimal extraction of surface rupture and removal of background noise, using sliding bar to the conversion area image is color segmented, obtain coarse segmentation image;Based on the principle of optimal extraction of surface rupture and removal of background noise, using sliding bar to the several sub-regions in the coarse segmentation image is screened, obtain fine extraction result.The present application can quickly, accurately extract surface rupture. BRIEF DESCRIPTION OF DRAWINGS

[0027] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0028] Figure 1 Flow chart of the present application based on color segmentation and object-oriented seismic surface rupture extraction method;

[0029] Figure 2 Schematic diagram of the present application region of interest image;

[0030] Figure 3 Schematic diagram of the present application color segmentation interface;

[0031] Figure 4 a coarse segmentation image of the present application;

[0032] Figure 5 a coarse segmentation image screening interface of the present application;

[0033] Figure 6 a fine extraction result of the present application. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the present application will be apparently and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the protection scope of the present application.

[0035] The purpose of the present application is to provide a color segmentation and object-oriented seismic surface rupture extraction method to quickly and accurately extract the surface rupture.

[0036] In order to make the above-mentioned purposes, characteristics and advantages of the present application more apparent and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0037] Figure 1 A flow chart of the color segmentation and object-oriented seismic surface rupture extraction method of the present application. As shown in the figure, Figure 1 the present application provides a color segmentation and object-oriented seismic surface rupture extraction method, which comprises:

[0038] Step S1, cutting the seismic surface rupture image to obtain a region of interest image.

[0039] Specifically, the step S1 comprises:

[0040] Step S11, drawing a polygon on the seismic surface rupture image displayed in the front end to generate a mask in the background. The polygon is any one of a self-defined regular polygon and a self-defined irregular polygon. The pixel value in the region of the polygon mask is 255, and the pixel value outside the region of the polygon mask is 0.

[0041] Step S12, calculating the four-to-range of the polygon to cut the to-be-masked image based on the four-to-range.

[0042] Step S13, performing convolution operation on the mask and the to-be-masked image to obtain the region of interest image. As shown in the figure, Figure 2 (a) is a seismic surface rupture image, Figure 2 (a) is a seismic surface rupture image, Figure 2(b) is a region of interest image.

[0043] The formula of the convolution operation is as follows:

[0044] ROI=image*mask / 255

[0045] In the formula, mask represents a mask, image represents an image to be masked, and ROI is a region of interest image.

[0046] Step S2, converting the region of interest image from an RGB color space to an HSV color space to obtain a converted region image. The conversion formula is as follows:

[0047] R, G, B ∈ [0, 1];

[0048] max=max(R, G, B);

[0049] min=min(R, G, B);

[0050] m=max-min;

[0051] g_mins_b=G-B;

[0052] S=0 if max=0;

[0053] S=m / max otherwise;

[0054] H=0 if max=min;

[0055] H=60°*g_mins_b / m if max=R and G>=B;

[0056] H=60°*g_mins_b / m+360° if max=R and G<B;

[0057] H=60°*g_mins_b / m+120° if max=G;

[0058] H=60°*g_mins_b / m+240° if max=B;

[0059] V=max.

[0060] Step S3, performing color segmentation on the converted region image based on the principle of extracting ground rupture and removing background noise optimally to obtain a coarse segmentation image.

[0061] Specifically, as Figure 3As shown, based on the optimal principles of extracting surface fractures and removing background noise, the maximum and minimum values ​​of H, S, V, and V of the transformed region image are dynamically adjusted using a slider. The results before and after adjustment are compared in real time, and the optimal result is selected to obtain the coarse segmentation image. Figure 3 In the diagram: Hue_Min is the minimum H value, Hue_Max is the maximum H value, Sat_Min is the minimum S value, Sat_Max is the maximum S value, Val_Min is the minimum V value, and Val_Max is the maximum V value. The white elements represent the real-time segmentation results. The coarse segmentation image is as follows: Figure 4 As shown. Figure 4 (a) is an image of surface rupture after an earthquake. Figure 4 (b) is a coarsely segmented image.

[0062] Step S4: Based on the optimal principles of extracting surface fractures and removing background noise, a slider is used to filter several sub-regions within the coarsely segmented image to obtain fine extraction results.

[0063] Further, step S4 includes:

[0064] Step S41: Extract the outer contour of each sub-region in the coarsely segmented image to obtain a sub-region set.

[0065] The coarsely segmented image is divided based on a contour tracking algorithm to obtain the sub-region set.

[0066] Based on the chain code principle, the contour points and their directions of each connected region are searched on the coarse segmentation image and represented by chain code. The inner and outer contours of each connected region are marked from the chain code, and the outer contour is extracted to obtain the sub-region set.

[0067] Step S42: Calculate the constraint parameters of each sub-region of the sub-region set to obtain the constraint parameter set; the constraint parameter set includes the spectral standard deviation set, the spectral mean set, and the minimum bounding rectangle aspect ratio set.

[0068] Step S43, as follows Figure 5 As shown, based on the optimal principles of extracting surface fractures and removing background noise, the maximum and minimum values ​​of the spectral standard deviation, the maximum value of the spectral mean, and the minimum aspect ratio of the minimum bounding rectangle are dynamically adjusted using a slider to filter the set of constraint parameters. The results before and after filtering are compared in real time to select the optimal result, thus obtaining the fine extraction result. The fine extraction result can be output as a shapefile (SHP) format vector surface and a fine extraction image. Figure 5 In the Chinese dictionary: std_min is the minimum standard deviation of the spectrum, std_max is the minimum standard deviation of the spectrum, h / w_min is the minimum aspect ratio of the minimum bounding rectangle, and mean_max is the maximum mean of the spectrum.Figure 5 The middle white element is the real-time screening result.

[0069] The present application verifies the unmanned aerial vehicle image with a resolution of about 2cm, the surface rupture is very small, some of which has a width of only 2cm, and even some of which has a length of only a few centimeters, and the morphological characteristics are very complex, based on the existing traditional machine learning and deep learning method, the fine structure of the surface rupture plane cannot be effectively extracted, and the fine extraction result of the seismic surface rupture of the present application is as shown in Figure 6 Figure 6 (a) is a seismic surface rupture image, Figure 6 (b) is the fine extraction result of the surface rupture, that is, the shp format vector plane; Figure 6 Figure 6 (k) respectively Figure 6 The fine structure of the surface rupture contour line of the different regions corresponding to the rectangular frame in (b) can be seen from the figure, the contour basically coincides with the rupture boundary on the image, and the requirement of automatically extracting the fine structure of the surface rupture plane can be met.

[0070] At present, the deep learning method needs a large amount of training labeled data, although some automatic labeling tools can realize automatic labeling, but it is still limited by the use scene, and the samples after labeling need to be processed by artificial again, which has high cost, while the method of the present application can be used as a generation tool of the surface rupture sample data set, and the labeled result can be adaptively obtained through artificial intervention, so that different labeling requirements can be met in real time and accurately.

[0071] The present application does not use filtering processing image to remove noise, which will not damage the true value of the image, so that the high resolution characteristics of the image can be fully utilized, and the fine structure of the surface rupture can be effectively extracted.

[0072] The present application removes part of the background noise by designing a self-defined irregular polygon grid clipping function, and improves the speed and accuracy.

[0073] The present application is based on the object-oriented idea, and uses the spectral characteristics and shape characteristics of the surface rupture object to improve the surface rupture extraction accuracy.

[0074] The present application sets a sliding bar dynamic parameter instead of manually selecting a threshold value to segment the image, so as to realize the coarse segmentation and fine extraction process in real time, and thus the surface rupture can be quickly extracted.

[0075] The present application can automatically extract the surface rupture vector plane, so that the quantitative parameters such as width can be further calculated, which lays a foundation for understanding the focal mechanism.

[0076] The embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments, and the same and similar parts of each embodiment can be referred to each other.​​

[0077] The principles and implementations of the present application are described in the specific examples herein, and the above examples are only used to help understand the method of the present application and its core idea; at the same time, for those skilled in the art, according to the idea of the present application, there will be changes in specific implementation and application range. In view of the above, the content of the specification should not be understood as a limitation of the present application.

Claims

1. A method for extracting seismic surface rupture based on color segmentation and object orientation, characterized in that, include: Cropping the earthquake surface rupture image yields the region of interest image, including: Polygons are drawn from the earthquake surface rupture image displayed on the front end, and a mask is generated in the background; The polygon is calculated to obtain its boundaries, and the image to be masked is obtained by cropping based on the boundaries. A convolution operation is performed on the mask and the image to be masked to obtain the region of interest image. The formula for the convolution operation is as follows: ROI=miage*mask / 255; In the formula: mask represents the mask, image represents the image to be masked, and ROI is the region of interest image; The region of interest image is converted from the RGB color space to the HSV color space to obtain the converted region image. The conversion formula used is as follows: H=60°*g_mins_b / m if max=R and G>=B; H=60°*g_mins_b / m+360°if max=R and G <B; H=60°*g_mins_b / m+120°if max=G; H=60°*g_mins_b / m+240°if max=B; Based on the optimal principles of extracting surface fractures and removing background noise, a slider is used to perform color segmentation on the converted region image to obtain a coarse segmentation image. The outer contour of the coarsely segmented image is extracted based on a contour tracking algorithm to obtain a set of sub-regions, specifically: Based on the chain code principle, the contour points and their directions of each connected region are searched on the coarse segmentation image and represented by chain code. The inner and outer contours of each connected region are marked from the chain code, and the outer contour is extracted to obtain the sub-region set. Based on the optimal principles of extracting surface fractures and removing background noise, a slider is used to filter several sub-regions within the coarsely segmented image to obtain fine extraction results. Specifically, this includes: dynamically adjusting the maximum and minimum values ​​of the spectral standard deviation, the maximum value of the spectral mean, and the minimum aspect ratio of the minimum bounding rectangle using the slider to filter the constraint parameter set; comparing the results before and after filtering in real time; selecting the optimal result; and obtaining the fine extraction result. The fine extraction result can be output as a shapefile vector surface and a fine extraction image.

2. The color segmentation and object-oriented based seismic surface rupture extraction method according to claim 1, characterized in that, Based on the optimal principles of extracting surface fractures and removing background noise, a slider is used to perform color segmentation on the converted region image to obtain a coarse segmentation image, specifically: Based on the optimal principles of extracting surface fractures and removing background noise, the maximum and minimum values ​​of H, S, V, and V of the transformed region image are dynamically adjusted using a slider. The results before and after adjustment are compared in real time, and the optimal result is selected to obtain the coarse segmentation image.

3. The color segmentation and object-oriented based seismic surface rupture extraction method according to claim 1, characterized in that, Based on the optimal principles of extracting surface fractures and removing background noise, a slider is used to filter several sub-regions within the coarsely segmented image to obtain fine extraction results, including: The outer contours of each sub-region in the coarsely segmented image are extracted to obtain a set of sub-regions; Calculate constraint parameters of each sub-region of the sub-region set to obtain a constraint parameter set; the constraint parameter set comprises a spectrum standard deviation set, a spectrum mean value set and a minimum circumscribed rectangle aspect ratio set.

Images