Image data set annotation segmentation method based on fault prediction deep learning

Abstract

The invention discloses an image data set labeling and segmenting method based on fault prediction deep learning, which belongs to a coal rock fault prediction method and comprises the steps of acquiring original three-dimensional seismic data in a work area to be measured, performing dimension reduction processing on the original three-dimensional seismic data, cutting an original two-dimensional matrix according to an Inline direction and the like. Three-dimensional seismic data are compressed into two-dimensional plane data through data dimension reduction, and then the three-dimensional seismic data are visualized into section two-dimensional image data of Time and CDP in each Inline direction through cutting in the Inline direction. Compared with the prior art, the three-dimensional seismic data processing method has the advantages that data calculation in traditional three-dimensional seismic data processing is skipped, image features are directly visualized, the image features of the edge part are reserved, edge information is not lost, the calculation amount is reduced, the calculation time is shortened, and the operation efficiency of the whole network is improved.

Description

Technical Field

[0001] The present invention relates to a coal rock fault prediction method, and more specifically, the present invention mainly relates to an image data set annotation and segmentation method based on fault prediction deep learning. Background Art

[0002] When the rock is subjected to force and the stress exceeds the strength limit of the rock, the rock breaks, forming a fault structure. Faults are one of the most common geological phenomena. In the process of coal mining, faults not only change the burial conditions of coal and rock layers, but also cause the coal and rock layers to break and displace significantly. On the one hand, this destroys the continuity and integrity of the coal seams, bringing resistance to coal mining; on the other hand, gas outbursts, water seepage, roof collapse and other accidents are prone to occur at the faults, which seriously affect the safe mining of the mining area. Therefore, predicting the distribution of faults in mining areas is an important part of safe mining of coal mines. At present, the geological structure of coal fields is usually detected by artificial seismic methods. Conventional fault interpretation methods determine the existence of faults and the size of the drop by visually identifying the reflected wave event axis misalignment and abnormal display. Traditional fault interpretation methods have long cycles, strong subjectivity, and difficulty in interpreting small faults. They are also highly dependent on the experience of interpreters and prior information about geological knowledge. It is difficult to meet the requirements of high precision, high efficiency and high speed for fault information in safe mining of coal mines. With the development and rise of deep learning discipline research represented by neural networks, automatic fault identification has become possible. Traditional machine learning detection methods extract the characteristics of earthquake attributes and combine classification algorithms to learn earthquake data. However, due to the problems of low contrast and noise interference in fault profiles, as well as data imbalance, small number of fault samples, and different fault heights, traditional machine learning methods have encountered challenges in fault detection. In recent years, deep learning models based on convolutional neural networks have played an important role in fault prediction. By autonomously learning fault features and learning non-fault features around the fault points, the model is continuously improved to identify faults in the entire area. Therefore, it is of great significance to study the construction and segmentation methods of fault prediction datasets based on deep learning. Summary of the invention

[0003] One of the purposes of the present invention is to address the above-mentioned shortcomings and provide an image data set annotation and segmentation method based on fault prediction deep learning, in the hope of solving the problems in the prior art that the traditional fault interpretation method is difficult to interpret small faults and relies on the experience of the interpreters and prior information about geological knowledge; the fault identification method of neural network deep learning also has technical problems such as low contrast, noise interference, small number of layer samples, and different fault drops.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0005] The present invention provides an image data set annotation and segmentation method based on fault prediction deep learning, the method comprising the following steps:

[0006] Step A: obtaining original three-dimensional seismic data in the work area to be measured, wherein the original three-dimensional seismic data records the amplitude information of each point in the work area to be measured in three dimensions, namely, Inline, CDP, and Time, and the data format is .segy.

[0007] Step B: Perform dimensionality reduction processing on the original three-dimensional seismic data to obtain the original two-dimensional matrix M by the following formula: tow-dim .

[0008] M tow-dim =Time×(Inline×CDP)

[0009] Step C: Inline the original two-dimensional matrix M tow-dim Cut and obtain multiple cutting two-dimensional matrices M Inline , cut the two-dimensional matrix M Inline The number is the number of Inlines.

[0010] Step D: for each of the two-dimensional cutting matrices M Inline Cut again to generate multiple two-dimensional small windows M of the current window patch Each of the two-dimensional small image windows M patch Time, Inline, and CDP information are recorded.

[0011] Step E: Through each two-dimensional small window M patch Recorded Inline, CDP, Time information, search for the corresponding Inline, CDP, Time information in the disclosed information of the work area to be tested, and determine the current two-dimensional small window M patch Is it the required data? If the judgment result is yes, the current two-dimensional small window M patch Add the counter parameter as an index and put it into a new three-dimensional matrix; otherwise, continue to judge the current cutting two-dimensional matrix M Inline The next two-dimensional small window M in patch .

[0012] Step F: According to the counter parameter, the fault and non-fault label matrices are simultaneously established, and the data set matrix and the label matrix are linked according to the counter parameter to complete the current cutting two-dimensional matrix M. Inline Repeat steps DF to obtain a three-dimensional fault data set, fault labels, a three-dimensional non-fault data set, and non-fault labels based on the revealed information of the work area to be tested.

[0013] Preferably, a further technical solution is that the original three-dimensional seismic data in the work area to be measured includes disclosed information of the work area to be measured, and the disclosed information of the work area to be measured includes disclosed fault information and disclosed non-fault information.

[0014] A further technical solution is: for each cutting two-dimensional matrix M Inline The cutting method is: use an n×n window and cut the two-dimensional matrix M Inline Move up, and after determining the center point of the window, extend n / 2-1 points to the left, n / 2 points to the right, n / 2 points upward, and n / 2 points downward.

[0015] A further technical solution is: the original three-dimensional seismic data is obtained from Geoframe software.

[0016] A further technical solution is: the newly created three-dimensional matrix is ​​a matrix of the two-dimensional small image window M patch After adding the counter parameter as an index, the counter×n×n three-dimensional data matrix is ​​rebuilt, and the two-dimensional matrix M is cut successively. Inlin After averaging, the counter parameters are incremented and put into the seismic data volume of the three-dimensional data matrix in sequence; the newly created three-dimensional matrix includes a fault three-dimensional matrix and a non-fault three-dimensional matrix.

[0017] A further technical solution is: the method also includes: according to the counter parameter, dividing the three-dimensional fault data set and the three-dimensional non-fault data set into a training set and a test set, and at the same time forming training labels and test labels, and putting the training set, test set, training labels and test labels into a convolutional neural network for deep learning of fault prediction.

[0018] Compared with the prior art, the beneficial effects of the present invention are at least one of the following:

[0019] Through data dimensionality reduction, the 3D seismic data is compressed into 2D plane data, and then cut along the Inline direction to materialize the 3D seismic data into 2D cross-sectional image data of Time and CDP in each Inline direction. This skips the traditional 3D seismic data processing that stays on data calculation and realizes the direct materialization of image features.

[0020] Through the n×n patch image segmentation method, each intercepted patch is indexed by its center point, which can establish an association between the image features and data, so that the image feature capture will not lose the actual data as the foundation. In addition, this segmentation method can better simulate the function of human vision to capture the information of two-dimensional images, making it easier for computers to obtain more information about the difference between fault points and non-fault points when establishing recognition models. It can also quickly capture the Inline, CDP, and Time information of the center point. While intercepting the amplitude information, using Inline, CDP, and Time as index information, the two-dimensional image data and the revealed breakpoint information can be quickly linked, thereby saving time and effort in making data set labels.

[0021] Through the patch image segmentation method of the center point, the image features of the edge part are retained, while the edge information is not lost, the amount of calculation and the calculation time are reduced, thus improving the operation efficiency of the overall network.

[0022] By re-entering the intercepted two-dimensional matrix into the three-dimensional matrix with the counter parameter, the label information and the two-dimensional matrix can be easily indexed, and in the established convolutional neural network, it is directly input as a three-dimensional matrix. The input of the three-dimensional matrix can process the two-dimensional image of each dimension at the same time, thereby realizing parallel computing in the network, reducing computing time and improving computing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 The present invention is a flowchart for describing an embodiment of the present invention.

[0024] Figure 2 It is used to illustrate the original two-dimensional matrix after dimensionality reduction processing in one embodiment of the present invention.

[0025] Figure 3 It is used to illustrate a two-dimensional cutting matrix after cutting along the Inline in one embodiment of the present invention.

[0026] Figure 4 25 random image patches of a single Inline are used to illustrate one embodiment of the present invention.

[0027] Figure 5 Partial cross-sectional images of tomographic and non-tomographic data sets of a single Inline are used to illustrate one embodiment of the present invention. DETAILED DESCRIPTION

[0028] As mentioned above, to meet the requirements of the training model for the data set, the image data set annotation and segmentation method based on fault prediction deep learning of the present invention is implemented as follows:

[0029] (1) Obtain 3D seismic data of the work area to be measured. The data format contains the amplitude information of each point and is stored in the form of Time, Inline, and CDP in the segy file S origin .

[0030] (2) Obtain the fault information of some of the work area to be tested that has been exposed. The data format is a label file L containing Time, Inline, and CDP. Fault .

[0031] (3) Obtain the non-fault information that has been exposed in the part of the work area to be tested. The data format is a label file L containing Time, Inline, and CDP. non-Fault .

[0032] (4) Perform dimensionality reduction processing on the original 3D seismic data to obtain M tow-dim =Time×(Inline×CDP) two-dimensional matrix M tow-dim .

[0033] (5) For the two-dimensional matrix M tow-dim By cutting in the inline direction, we can get I (the number of inline numbers) Time×CDP two-dimensional matrices M Inline .

[0034] (6) On each cut Time×CDP two-dimensional matrix, use an n×n window (patch) to move on the data points of this two-dimensional matrix, and generate an n×n two-dimensional small window M of the patch according to the center point of the patch (extending 7 points to the left, 8 points to the right, 7 points above, and 8 points below the patch). patch , the M patch The information stored is the amplitude information of n×n points centered at this point, and the Time, Inline, and CDP information are used as the matrix M patch Index label L patch .

[0035] (7) Given that in (5) Inline The Time, Inline, and CDP information of each point in the M Inline M patch The label L obtained in patch The index information is in L Fault and L non-Fault Traverse search.

[0036] (8) The search logic is label L patch The index information must be consistent with L Fault or L non-FaultThe information in L is exactly the same, that is, Time, Inline, and CDP are all the same. Fault or L non-Fault Find the same information in the matrix M patch Add another label L classification , store the M patch The center point of the matrix is ​​the fault point or the non-fault point, and the two-dimensional matrix M patch Stored in two fault and non-fault three-dimensional matrices M three-dim (counter×n×n), L classification Corresponding to M three-dim The counter in is recorded in the vector V lable and proceed to the next one; if label L patch The index information is in L Fault or L non-Fault If all the points are not found, then the point is discarded and the next one is moved to the next one. Inline All M patch All captured and searched completed.

[0037] (9) After the inner loop of (8) is completed, M Inline The next Inline section is intercepted to form an outer loop. In this way, the inner and outer loops are nested to capture all the breakpoint information and label information. Finally, the fault three-dimensional matrix M is generated based on the exposed fault information. three-dim-fault (counter×n×n) and its label information V lable-fault (faults are marked as 1, non-faults are marked as 0); and the non-fault three-dimensional matrix M three-dim-nonfault (counter×n×n) and its label information V lable-nonfault .

[0038] (10) Furthermore, in the three-dimensional matrix M three-dim-nonfault and M three-dim-fault The training set and the test set are divided into 7:3 ratios. Each two-dimensional image in this three-dimensional matrix is ​​controlled by a unique third-dimensional counter, so the counter can be used as an index to divide the label matrix into training set and test set. Thus, the training data set D is formed. train And the test dataset D test , and the accompanying training label vector L train and the test label vector L test .

[0039] (11) The initial work of making the data set has been completed. Now we can put the cut data set into the convolutional neural network, detect and identify the tomographic images, build a recognition model, and then predict the fault distribution of the entire area, so as to achieve the purpose of fault prediction.

[0040] The present invention will be further described below in conjunction with the accompanying drawings.

[0041] refer to Figure 1 As shown, one embodiment of the present invention is the above-mentioned image data set annotation and segmentation method based on fault prediction deep learning, and the method is performed according to the following steps:

[0042] S1. First, we need to obtain the original 3D seismic data in the work area. This data records the amplitude information of each point in the work area in three dimensions: Inline, CDP, and Time. The data format is .segy. As 3D seismic information, it is difficult for us to process and cut it directly, so we need to reduce its dimension, convert it into a 2D image, and retain its 3D seismic characteristics.

[0043] S2, such as Figure 2 As shown in the figure, after dimensionality reduction, the Inline and CDP in the 3D seismic data are integrated into the same dimension, but the stratigraphic characteristics and fault locations are still unclear.

[0044] S3. To obtain the stratigraphic characteristics and determine the fault location, we need to cut the seismic data along the Inline and generate a single Inline 2D seismic profile image on a certain Inline. This 2D seismic profile image has Time as the vertical axis and CDP as the horizontal axis.

[0045] like Figure 3 As shown, on this image, we can already perform traditional manual identification work, that is, we can judge the fault location based on experience and knowledge of interpretation work.

[0046] S4. In order to better capture the features and details near the breakpoints and put them into the neural network for training, it is necessary to segment the single Inline 2D seismic profile image.

[0047] Here, preferably, the above-mentioned cutting method for segmenting the single Inline two-dimensional seismic profile image is to use an n×n window (patch) to move on the target two-dimensional image to be cut. The logic of patch point selection is to take one point as the center point, extend n / 2-1 points to the left, extend n / 2 points to the right, extend n / 2-1 points upward, and extend n / 2 points downward (n is mostly an even number).

[0048] To prevent unnecessary features from interfering with the model's judgment, edge information points will not be used as center points. Instead, edge features will be captured together in the patch of their nearest neighbors.

[0049] In addition to the amplitude information of the n×n points, the intercepted patch also records the Inline, CDP, and Time information of the center point, such as Figure 4 Shown are 25 random patches of a single Inline.

[0050] After a single Inline two-dimensional profile image is segmented by patches, it is summarized in a four-dimensional matrix. Since the Inline is uniquely determined, the first two dimensions of this four-dimensional matrix are CDP and Time, and the last two dimensions store the patch data for call.

[0051] According to the disclosed information of the work area, through the Inline, CDP, and Time information corresponding to each small two-dimensional data matrix, search for the corresponding Inline, CDP, and Time information in the fault information and non-fault information to determine whether the two-dimensional data matrix is ​​the required data. If it is fault or non-fault data, add the counter parameter to the small two-dimensional data matrix as an index and put it into a new three-dimensional matrix.

[0052] According to the index parameter counter, the fault and non-fault label matrices are established at the same time, and the data set matrix and label are connected based on the counter. In this way, the segmentation of a single Inline 2D seismic profile image and the production of data sets and labels are completed. Figure 5 Partial cross-sectional images of the single Inline fault and non-fault data sets.

[0053] S5. After completing the segmentation of a two-dimensional seismic profile image in the Inline direction, continue to complete the next two-dimensional seismic profile image in the Inline direction in the same cutting and searching manner.

[0054] S6. In the process of segmenting and searching for all data sets and labels, you can use a loop or parallel method to cut. The loop Inline method is to cut all the Inline 2D seismic profile images in sequence; the parallel cutting method is relatively fast, but requires higher computing power of the computer.

[0055] S7. After cutting all the two-dimensional cross-sectional images, the three-dimensional fault data set, fault labels, three-dimensional non-fault data set, and non-fault labels obtained according to the exposed information of the work area can be obtained.

[0056] Based on the above embodiments, it can be seen that the present invention has the following improvements over the prior art:

[0057] Through data dimensionality reduction, the 3D seismic data is compressed into 2D plane data, and then cut along the Inline direction to materialize the 3D seismic data into 2D cross-sectional image data of Time and CDP in each Inline direction. This skips the traditional 3D seismic data processing that stays on data calculation and realizes the direct materialization of image features.

[0058] Through the n×n patch image segmentation method, each intercepted patch is indexed by its center point, which can establish an association between the image features and data, so that the image feature capture will not lose the actual data as the foundation. In addition, this segmentation method can better simulate the function of human vision to capture the information of two-dimensional images, making it easier for computers to obtain more information about the difference between fault points and non-fault points when establishing recognition models.

[0059] Through the n×n patch image segmentation method, the Inline, CDP, and Time information of the center point can be quickly captured. While intercepting the amplitude information, using Inline, CDP, and Time as index information, the two-dimensional image data and the revealed breakpoint information can be quickly linked, thereby saving time and effort in making data set labels.

[0060] Through the patch image segmentation method of the center point, the image features of the edge part are retained, while the edge information is not lost, the amount of calculation and the calculation time are reduced, thus improving the operation efficiency of the overall network.

[0061] By putting the intercepted two-dimensional matrix back into the three-dimensional matrix with counters, the label information and the two-dimensional matrix can be easily indexed, and in the established convolutional neural network, it is directly input as a three-dimensional matrix. The input of the three-dimensional matrix can process the two-dimensional image of each dimension at the same time, thereby realizing parallel computing in the network, reducing computing time, and greatly improving computing efficiency.

[0062] In addition to the above, it should be noted that "one embodiment", "another embodiment", "embodiment", etc. mentioned in this specification refer to the specific features, structures or characteristics described in conjunction with the embodiment included in at least one embodiment generally described in this application. The same expression appearing in multiple places in the specification does not necessarily refer to the same embodiment. Further, when describing a specific feature, structure or characteristic in conjunction with any embodiment, it is claimed that the realization of such feature, structure or characteristic in conjunction with other embodiments also falls within the scope of the present invention.

[0063] Although the present invention is described herein with reference to a number of illustrative embodiments of the present invention, it will be appreciated that those skilled in the art may devise many other modifications and implementations that fall within the scope and spirit of the principles disclosed herein. More specifically, within the scope of the present disclosure, drawings, and claims, a variety of variations and improvements may be made to the components and / or layout of the subject combination layout. In addition to variations and improvements made to the components and / or layout, other uses will also be apparent to those skilled in the art.

Claims

1. A method for annotating and segmenting image datasets based on deep learning for fault prediction, characterized in that The method comprises the following steps: Acquire original three-dimensional seismic data in the work area to be measured, wherein the original three-dimensional seismic data records the amplitude information of each point in the work area to be measured in three dimensions: Inline, CDP, and Time; The original three-dimensional seismic data is subjected to dimensionality reduction processing, and the original two-dimensional matrix M is obtained by the following formula: tow-dim ; M tow-dim =Time×(Inline×CDP) According to the direction of Inline, the original two-dimensional matrix M tow-dim Cut and obtain multiple cutting two-dimensional matrices M Inline ; For each of the cutting two-dimensional matrix M Inline Cut again to generate multiple two-dimensional small windows M of the current window patch Each of the two-dimensional small image windows M patch Time, Inline, and CDP information are all recorded; Through each two-dimensional small window M patch Recorded Inline, CDP, Time information, search for the corresponding Inline, CDP, Time information in the disclosed information of the work area to be tested, and determine the current two-dimensional small window M patch Is it the required data? If the judgment result is yes, the current two-dimensional small window M patch Add the counter parameter as an index and put it into a new three-dimensional matrix; otherwise, continue to judge the current cutting two-dimensional matrix M Inline The next two-dimensional small window M in patch ; According to the counter parameter, the fault and non-fault label matrices are established at the same time, and the data set matrix and the label matrix are connected according to the counter parameter to complete the current cutting two-dimensional matrix M Inline Cutting and dataset,labeling; When all the two-dimensional matrix M is cut Inline After the cutting and data set and label production are completed, the three-dimensional fault data set, fault label, three-dimensional non-fault data set and non-fault label obtained according to the revealed information of the work area to be measured are obtained.

2. The image data set annotation and segmentation method based on fault prediction deep learning according to claim 1 is characterized by: The original three-dimensional seismic data in the work area to be measured includes disclosed information of the work area to be measured, and the disclosed information of the work area to be measured includes disclosed fault information and disclosed non-fault information.

3. The image data set annotation and segmentation method based on fault prediction deep learning according to claim 1 or 2, characterized in that: For each cutting two-dimensional matrix M Inline The cutting method is: use an n×n window and cut the two-dimensional matrix M Inline Move up, and after determining the center point of the window, extend n / 2-1 points to the left, n / 2 points to the right, n / 2 points upward, and n / 2 points downward.

4. The image data set annotation and segmentation method based on fault prediction deep learning according to claim 1 or 2, characterized in that: The original three-dimensional seismic data is obtained from Geoframe software.

5. The image data set annotation and segmentation method based on fault prediction deep learning according to claim 1, characterized in that: The newly created three-dimensional matrix is ​​a two-dimensional small window M patch After adding the counter parameter as an index, the counter×n×n three-dimensional data matrix is ​​rebuilt, and the two-dimensional matrix M is cut successively. Inlin After averaging, the data are placed into the seismic data volume of the three-dimensional data matrix in sequence according to the self-increment of the parameter counter; The newly created three-dimensional matrix includes a fault three-dimensional matrix and a non-fault three-dimensional matrix.

6. The image data set annotation and segmentation method based on fault prediction deep learning according to claim 1 or 5, characterized in that The method further comprises: According to the counter parameter, the three-dimensional fault data set and the three-dimensional non-fault data set are divided into training sets and test sets, and training labels and test labels are formed at the same time. The training set, test set, training labels and test labels are put into the convolutional neural network for deep learning of fault prediction.

Images