CN106951924B - Seismic coherence body image fault automatic identification method and system based on AdaBoost algorithm - Google Patents
Seismic coherence body image fault automatic identification method and system based on AdaBoost algorithm Download PDFInfo
- Publication number
- CN106951924B CN106951924B CN201710186136.XA CN201710186136A CN106951924B CN 106951924 B CN106951924 B CN 106951924B CN 201710186136 A CN201710186136 A CN 201710186136A CN 106951924 B CN106951924 B CN 106951924B
- Authority
- CN
- China
- Prior art keywords
- image
- image blocks
- fault
- classification
- adaboost algorithm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/285—Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention provides an automatic seismic coherence body image fault identification method and system based on an AdaBoost algorithm, wherein the method comprises the following steps: acquiring image blocks in a seismic coherent body fault image, wherein the image blocks comprise fault body image blocks, fault edge image blocks and fault background image blocks; classifying the image blocks through a pre-trained classification network to obtain a classification result, wherein the classification network is composed of a plurality of classifiers and is obtained through learning of an AdaBoost algorithm; and according to the classification result, realizing automatic identification of the seismic coherence tomography image. The invention trains the classification network by adopting an ensemble learning method, and classifies the image blocks by using the classification network, so that the automatic fault identification and extraction results are more accurate.
Description
Technical Field
The invention relates to the field of seismic coherence tomography identification, in particular to an automatic seismic coherence tomography image tomography identification method and system based on an AdaBoost algorithm.
Background
Seismic coherent interpretation is the most prominent technique for achieving geological fault location and attribute analysis. However, the inevitable external factors such as equipment, environment and human factors, and the influence of complex interference and noise generated by the external factors on different geological structures exist in the field acquisition process of seismic data, and the data degradation caused by the influence is often nonlinear. For this reason, the stratigraphic background and fault target data in the seismic tomographic image are often irregularly distributed complex data. In the face of such complex data, the conventional methods based on image edge detection/contour extraction, image segmentation/target extraction and the like are difficult to realize accurate identification and positioning of the fault, and often require more manual editing work. In addition, the proportion of fault layers and background data in the seismic coherence image is unbalanced, the background data sparsity image is good, the fault is relatively complex, and certain difficulty is brought to automatic fault identification.
It is considered that the artificial intelligence method such as machine learning has been well applied in various disciplines and fields related to signal and information processing. Particularly, the ensemble learning method does not make any assumption on the prior distribution of the training data, and has strong robustness on the data distribution condition. The method can effectively solve the classification tasks of unbalanced category and uneven attribute by directly working in the probability space of input data. Obviously, there is a need to develop an application technology of the ensemble learning method in seismic coherence tomography from the classification point of view.
Therefore, the existing seismic fault method in the seismic coherent body image has the defect that the automatic fault identification and extraction result is inaccurate due to irregular fault distribution, complex and changeable fault shape, unbalanced fault and background data ratio and the like.
Disclosure of Invention
Aiming at the technical problems, the invention provides an earthquake coherent body image fault automatic identification method and system based on an AdaBoost algorithm.
In order to solve the technical problems, the technical scheme provided by the invention is as follows:
in a first aspect, the invention provides an automatic seismic coherence image fault identification method based on an AdaBoost algorithm, which comprises the following steps:
step S1, acquiring image blocks in the seismic coherent body fault image, wherein the image blocks comprise fault body image blocks, fault edge image blocks and fault background image blocks;
step S2, classifying the image blocks through a pre-trained classification network to obtain a classification result, wherein the classification network is composed of a plurality of classifiers and is obtained through learning of an AdaBoost algorithm;
and step S3, automatically recognizing the seismic coherence tomography image according to the classification result.
The invention discloses an earthquake coherent body image fault automatic identification method based on an AdaBoost algorithm, which adopts the technical scheme that: acquiring image blocks in a seismic coherent body fault image, wherein the image blocks comprise fault body image blocks, fault edge image blocks and fault background image blocks; classifying the image blocks through a pre-trained classification network to obtain a classification result, wherein the classification network is composed of a plurality of classifiers and is obtained through learning of an AdaBoost algorithm; and according to the classification result, realizing automatic identification of the seismic coherence tomography image.
The earthquake coherent body image fault automatic identification method based on the AdaBoost algorithm adopts an ensemble learning method to train a classification network, and classifies image blocks by using the classification network, so that the automatic identification and extraction results of faults are more accurate.
Further, the training process of the classification network is as follows:
selecting N off-line training samples to form a first training sample, and learning the first training sample to obtain a first weak classifier, wherein the off-line training samples are image blocks in a seismic coherence tomography image;
forming a second training sample by the wrongly-divided sample in the first weak classifier and other off-line training samples, and obtaining a second weak classifier through learning the second training sample, wherein the second training sample is formed by N off-line training samples;
and forming a third training sample by using the sample which is wrongly divided in the first weak classifier and the second weak classifier and other off-line training samples, obtaining a third weak classifier by learning the third training sample, and forming a classification network formed by three second classes of classifiers, wherein the third training sample is formed by N off-line training samples.
Further, in step S2, specifically, the method includes:
classifying the image blocks through classifiers in the classification network to obtain the image blocks processed by each classifier;
calculating the similarity between the image block and the image block processed by each classifier to obtain a plurality of similarities;
calculating the voting result of each classifier according to the plurality of similarities;
and obtaining a classification result according to the voting result, wherein the classification result is the class with the largest accumulated voting result.
Further, the size of the image blocks is at least one of 3 × 3, 5 × 5, 7 × 7, and 9 × 9.
Further, the selection ratio of the image blocks of the fault body, the image blocks of the fault edge and the image blocks of the fault background in the image blocks is 3: 2: 1.
in a second aspect, the present invention provides an automatic seismic coherence image fault identification system based on an AdaBoost algorithm, including:
the image block acquisition module is used for acquiring image blocks in the seismic coherence tomography image, wherein the image blocks comprise tomography body image blocks, tomography edge image blocks and tomography background image blocks;
the image block classification module is used for classifying image blocks through a pre-trained classification network to obtain a classification result, wherein the classification network is composed of a plurality of classifiers and is obtained through learning of an AdaBoost algorithm;
and the automatic fault image identification module is used for realizing automatic identification of the seismic coherence tomography image according to the classification result.
The invention provides an earthquake coherent body image fault automatic identification system based on an AdaBoost algorithm, which has the technical scheme that: firstly, an image block acquisition module is used for acquiring image blocks in a seismic coherence tomography image, wherein the image blocks comprise tomography body image blocks, tomography edge image blocks and tomography background image blocks;
then, an image block classification module is used for classifying image blocks through a pre-trained classification network to obtain a classification result, wherein the classification network is composed of a plurality of classifiers and is obtained through learning of an AdaBoost algorithm; and finally, automatically identifying the seismic coherence tomography image according to the classification result by a tomography image automatic identification module.
The earthquake coherent body image fault automatic identification system based on the AdaBoost algorithm adopts an ensemble learning method to train a classification network, and classifies image blocks by using the classification network, so that the automatic identification and extraction results of faults are more accurate.
Further, the image block classification module is specifically configured to train a classification network, and specifically includes:
selecting N off-line training samples to form a first training sample, and learning the first training sample to obtain a first weak classifier, wherein the off-line training samples are image blocks in a seismic coherence tomography image;
forming a second training sample by the wrongly-divided sample in the first weak classifier and other off-line training samples, and obtaining a second weak classifier through learning the second training sample, wherein the second training sample is formed by N off-line training samples;
and forming a third training sample by using the sample which is wrongly divided in the first weak classifier and the second weak classifier and other off-line training samples, obtaining a third weak classifier by learning the third training sample, and forming a classification network formed by three second classes of classifiers, wherein the third training sample is formed by N off-line training samples.
Further, the image block classification module is specifically configured to:
classifying the image blocks through classifiers in the classification network to obtain the image blocks processed by each classifier;
calculating the similarity between the image block and the image block processed by each classifier to obtain a plurality of similarities;
calculating the voting result of each classifier according to the plurality of similarities;
and obtaining a classification result according to the voting result, wherein the classification result is the class with the largest accumulated voting result.
Further, the size of the image blocks is at least one of 3 × 3, 5 × 5, 7 × 7, and 9 × 9.
Further, the selection ratio of the image blocks of the fault body, the image blocks of the fault edge and the image blocks of the fault background in the image blocks is 3: 2: 1.
drawings
In order to more clearly illustrate the detailed description of the invention or the technical solutions in the prior art, the drawings that are needed in the detailed description of the invention or the prior art will be briefly described below.
Fig. 1 shows a flow chart of an automatic seismic coherence image fault identification method based on an AdaBoost algorithm according to an embodiment of the present invention;
FIG. 2 is a schematic diagram illustrating a multi-class classification network of an automatic seismic coherence tomography image fault identification method based on an AdaBoost algorithm according to an embodiment of the present invention;
fig. 3 is a schematic diagram illustrating an automatic seismic coherence image fault identification system based on an AdaBoost algorithm according to an embodiment of the present invention.
Detailed Description
Embodiments of the present invention will be described in detail below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and therefore are only examples, and the protection scope of the present invention is not limited thereby.
Example one
Fig. 1 shows a flow chart of an automatic seismic coherence image fault identification method based on an AdaBoost algorithm according to an embodiment of the present invention; as shown in fig. 1, in an embodiment, a seismic coherence image fault automatic identification method based on an AdaBoost algorithm includes:
step S1, acquiring image blocks in the seismic coherent body fault image, wherein the image blocks comprise fault body image blocks, fault edge image blocks and fault background image blocks;
step S2, classifying the image blocks through a pre-trained classification network to obtain a classification result, wherein the classification network is composed of a plurality of classifiers and is obtained through learning of an AdaBoost algorithm;
and step S3, automatically recognizing the seismic coherence tomography image according to the classification result.
The invention discloses an earthquake coherent body image fault automatic identification method based on an AdaBoost algorithm, which adopts the technical scheme that: acquiring image blocks in a seismic coherent body fault image, wherein the image blocks comprise fault body image blocks, fault edge image blocks and fault background image blocks; classifying the image blocks through a pre-trained classification network to obtain a classification result, wherein the classification network is composed of a plurality of classifiers and is obtained through learning of an AdaBoost algorithm; and according to the classification result, realizing automatic identification of the seismic coherence tomography image.
The invention discloses an earthquake coherent body image fault automatic identification method based on an AdaBoost algorithm, which adopts an ensemble learning method to train a classification network, wherein Adaboost is an iterative algorithm.
And changing the distribution of irregular fault distribution, complex and changeable fault shapes, faults and background data through a classification network, and determining the weight of each sample according to whether the classification of each sample in each training set is correct and the accuracy of the last overall classification. And (4) sending the new data set with the modified weight value to a lower-layer classifier for training, and finally fusing the classifiers obtained by each training as a final decision classifier. The use of the Adaboost classifier may exclude some unnecessary training data features and overlay critical training data. Through continuous training, the data classification capability can be improved. The invention classifies the image blocks by using the classification network obtained by Adaboost training, and can make the automatic identification and extraction result of the fault more accurate.
Referring to fig. 2, the training process of the classification network is:
selecting N off-line training samples to form a first training sample, and learning the first training sample to obtain a first weak classifier, wherein the off-line training samples are image blocks in the seismic coherence tomography image;
forming a second training sample by the wrongly divided sample in the first weak classifier and other off-line training samples, and obtaining a second weak classifier through learning the second training sample, wherein the second training sample is formed by N off-line training samples;
and forming a third training sample by using the sample which is wrongly divided in the first weak classifier and the second weak classifier and other off-line training samples, obtaining a third weak classifier by learning the third training sample, and forming a classification network formed by three second classes of classifiers, wherein the third training sample is formed by N off-line training samples.
Through continuous training, three two-class classifiers shown in fig. 2 are obtained to form a classification network, the fault image blocks are classified, a more accurate classification result is obtained, and further, the automatic fault identification and extraction result is more accurate.
Specifically, step S2 specifically includes:
classifying the image blocks through classifiers in a classification network to obtain the image blocks processed by each classifier;
calculating the similarity between the image block and the image block processed by each classifier to obtain a plurality of similarities;
the specific process of calculating the similarity between the image block and the image block processed by each classifier is as follows:
in the probability theory and the statistical theory, the Hellinger distance is used for measuring the similarity of two probability distributions, so the Hellinger distance which is fit by the mean image block gray level Gaussian distribution of the image blocks to various samples is calculated to express the similarity between the image blocks and the image blocks processed by each classifier, and the calculation formula of the Hellinger distance is as follows:
whereinFitting parameters for the gray level Gaussian distribution of the current image block;
and fitting parameters for the i-th sample average image block gray level Gaussian distribution.
Calculating the voting result of each classifier according to the plurality of similarities;
the voting result is calculated by the following formula:
wherein v isk＝{vk(i) The vote value of the kth second-class classifier is obtained;
and obtaining a classification result according to the voting result, wherein the classification result is the class with the maximum accumulated voting result.
Specifically, the size of the image block is at least one of 3 × 3, 5 × 5, 7 × 7, and 9 × 9.
According to related researches, the minimum description scale of the pixels at the edges of the fault layers in the seismic coherent body image is 3 x 3, while the complete representation of the edges of the image usually needs the scale of 9 x 9 image blocks, and for this reason, the invention selects 3 x 3, 5 x 5, 7 x 7 and 9 x 9 scales respectively to select the sample image blocks.
Preferably, as shown by comparing the results of the experiment, if the image block is too small, such as 3 × 3, the continuous fault is split in the classification result, and if the image block is too large, such as 9 × 9, the fault edge and the fault body are both expanded in the experiment result. Only in the case of a moderate image block size selection, such as 5 × 5 or 7 × 7, the final classification result can accurately identify slice bodies and slice edges. Therefore preferred image blocks in the present invention have a size of 5 x 5 or 7 x 7.
Preferably, the selection ratio of the image blocks of the fault body, the image blocks of the fault edge and the image blocks of the fault background in the image blocks is 3: 2: 1.
considering the representativeness of sample data and the sparsity problem of corresponding categories, namely the situation that background data sparsity image pairs are good and faults are relatively complex, the selection ratio of fault body image blocks, fault edge image blocks and fault background image blocks in the image blocks is 3: 2: 1. therefore, the problem that the automatic identification and extraction result of the fault is inaccurate due to the fact that the quantity of the image blocks of the fault body, the image blocks of the fault edge and the image blocks of the fault background is not balanced can be solved.
Fig. 3 is a schematic diagram illustrating an automatic seismic coherence image fault identification system based on an AdaBoost algorithm according to an embodiment of the present invention. As shown in fig. 3, the present invention provides an automatic seismic coherent body image fault identification system 10 based on AdaBoost algorithm, including:
the image block acquisition module 101 is used for acquiring image blocks in the seismic coherence tomography image, wherein the image blocks comprise tomography image blocks, tomography edge image blocks and tomography background image blocks;
the image block classification module 102 is configured to classify image blocks through a pre-trained classification network to obtain a classification result, where the classification network is formed by multiple classifiers and is obtained through learning of an AdaBoost algorithm;
and the automatic tomographic image identification module 103 is used for automatically identifying the seismic coherence tomography image according to the classification result.
The invention provides an earthquake coherent body image fault automatic identification system 10 based on an AdaBoost algorithm, which has the technical scheme that: firstly, an image block acquisition module is used for acquiring image blocks in a seismic coherence tomography image, wherein the image blocks comprise tomography body image blocks, tomography edge image blocks and tomography background image blocks;
then, an image block classification module is used for classifying image blocks through a pre-trained classification network to obtain a classification result, the classification network is composed of a plurality of classifiers, and the classification network is obtained through learning of an AdaBoost algorithm; and finally, automatically identifying the seismic coherence tomography image according to the classification result by using a tomography image automatic identification module.
The earthquake coherent body image fault automatic identification system based on the AdaBoost algorithm adopts an integrated learning method to train a classification network, Adaboost is an iterative algorithm, the core idea is to train different classifiers (weak classifiers) aiming at a training set formed by the same earthquake coherent body image, and then the weak classifiers are collected to form a classification network, so that the classification capability of the classification network is stronger.
And changing the distribution of irregular fault distribution, complex and changeable fault shapes, faults and background data through a classification network, and determining the weight of each sample according to whether the classification of each sample in each training set is correct and the accuracy of the last overall classification. And (4) sending the new data set with the modified weight value to a lower-layer classifier for training, and finally fusing the classifiers obtained by each training as a final decision classifier. The use of the Adaboost classifier may exclude some unnecessary training data features and overlay critical training data. Through continuous training, the data classification capability can be improved. The invention classifies the image blocks by using the classification network obtained by Adaboost training, and can make the automatic identification and extraction result of the fault more accurate.
Referring to fig. 2, the image block classification module 102 is specifically configured to train a classification network, and specifically includes:
selecting N off-line training samples to form a first training sample, and learning the first training sample to obtain a first weak classifier, wherein the off-line training samples are image blocks in the seismic coherence tomography image;
forming a second training sample by the wrongly divided sample in the first weak classifier and other off-line training samples, and obtaining a second weak classifier through learning the second training sample, wherein the second training sample is formed by N off-line training samples;
and forming a third training sample by using the sample which is wrongly divided in the first weak classifier and the second weak classifier and other off-line training samples, obtaining a third weak classifier by learning the third training sample, and forming a classification network formed by three second classes of classifiers, wherein the third training sample is formed by N off-line training samples.
Through continuous training, three two-class classifiers shown in fig. 2 are obtained to form a classification network, the fault image blocks are classified, a more accurate classification result is obtained, and further, the automatic fault identification and extraction result is more accurate.
Specifically, the image block classification module 102 is specifically configured to:
classifying the image blocks through classifiers in a classification network to obtain the image blocks processed by each classifier;
calculating the similarity between the image block and the image block processed by each classifier to obtain a plurality of similarities;
the specific process of calculating the similarity between the image block and the image block processed by each classifier is as follows:
in the probability theory and the statistical theory, the Hellinger distance is used for measuring the similarity of two probability distributions, so the Hellinger distance which is fit by the mean image block gray level Gaussian distribution of the image blocks to various samples is calculated to express the similarity between the image blocks and the image blocks processed by each classifier, and the calculation formula of the Hellinger distance is as follows:
wherein
Fitting parameters for the gray level Gaussian distribution of the current image block;
and fitting parameters for the i-th sample average image block gray level Gaussian distribution.
Calculating the voting result of each classifier according to the plurality of similarities;
the voting result is calculated by the following formula:
wherein v isk＝{vk(i) The vote value of the kth second-class classifier is obtained;
and obtaining a classification result according to the voting result, wherein the classification result is the class with the maximum accumulated voting result.
Specifically, the size of the image block is at least one of 3 × 3, 5 × 5, 7 × 7, and 9 × 9.
According to related researches, the minimum description scale of the pixels at the edges of the fault layers in the seismic coherent body image is 3 x 3, while the complete representation of the edges of the image usually needs the scale of 9 x 9 image blocks, and for this reason, the invention selects 3 x 3, 5 x 5, 7 x 7 and 9 x 9 scales respectively to select the sample image blocks.
Preferably, as shown by comparing the results of the experiment, if the image block is too small, such as 3 × 3, the continuous fault is split in the classification result, and if the image block is too large, such as 9 × 9, the fault edge and the fault body are both expanded in the experiment result. Only in the case of a moderate image block size selection, such as 5 × 5 or 7 × 7, the final classification result can accurately identify slice bodies and slice edges. Therefore preferred image blocks in the present invention have a size of 5 x 5 or 7 x 7.
Preferably, the selection ratio of the image blocks of the fault body, the image blocks of the fault edge and the image blocks of the fault background in the image blocks is 3: 2: 1.
considering the representativeness of sample data and the sparsity problem of corresponding categories, namely the situation that background data sparsity image pairs are good and faults are relatively complex, the selection ratio of fault body image blocks, fault edge image blocks and fault background image blocks in the image blocks is 3: 2: 1. therefore, the problem that the automatic identification and extraction result of the fault is inaccurate due to the fact that the quantity of the image blocks of the fault body, the image blocks of the fault edge and the image blocks of the fault background is not balanced can be solved.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; such modifications and substitutions do not depart from the spirit and scope of the present invention, and they should be construed as being included in the following claims and description.
Claims (10)
1. The seismic coherence image fault automatic identification method based on the AdaBoost algorithm is characterized by comprising the following steps of:
step S1, acquiring image blocks in the seismic coherent body fault image, wherein the image blocks comprise fault body image blocks, fault edge image blocks and fault background image blocks;
step S2, classifying the image blocks through a pre-trained classification network to obtain a classification result, wherein the classification network is composed of a plurality of classifiers and is obtained through learning of an AdaBoost algorithm;
and step S3, automatically recognizing the seismic coherence tomography image according to the classification result.
2. The seismic coherence image fault automatic identification method based on the AdaBoost algorithm is characterized in that,
the training process of the classification network comprises the following steps:
selecting N off-line training samples to form a first training sample, and learning the first training sample to obtain a first weak classifier, wherein the off-line training samples are image blocks in a seismic coherence tomography image;
forming a second training sample by the wrongly-divided sample in the first weak classifier and other off-line training samples, and obtaining a second weak classifier through learning the second training sample, wherein the second training sample is formed by N off-line training samples;
and forming a third training sample by using the sample which is wrongly divided in the first weak classifier and the second weak classifier and other off-line training samples, obtaining a third weak classifier by learning the third training sample, and forming a classification network formed by three second classes of classifiers, wherein the third training sample is formed by N off-line training samples.
3. The seismic coherence image fault automatic identification method based on the AdaBoost algorithm is characterized in that,
the step S2 specifically includes:
classifying the image blocks through classifiers in the classification network to obtain the image blocks processed by each classifier;
calculating the similarity between the image block and the image block processed by each classifier to obtain a plurality of similarities;
calculating the voting result of each classifier according to the plurality of similarities;
and obtaining a classification result according to the voting result, wherein the classification result is the class with the largest accumulated voting result.
4. The seismic coherence image fault automatic identification method based on the AdaBoost algorithm is characterized in that,
the size of the image blocks is at least one of 3 × 3, 5 × 5, 7 × 7, and 9 × 9.
5. The seismic coherence image fault automatic identification method based on the AdaBoost algorithm is characterized in that,
the selection ratio of the image blocks of the fault body, the image blocks of the fault edge and the image blocks of the fault background in the image blocks is 3: 2: 1.
6. an earthquake coherent body image fault automatic identification system based on an AdaBoost algorithm is characterized by comprising the following steps:
the image block acquisition module is used for acquiring image blocks in the seismic coherence tomography image, wherein the image blocks comprise tomography body image blocks, tomography edge image blocks and tomography background image blocks;
the image block classification module is used for classifying image blocks through a pre-trained classification network to obtain a classification result, wherein the classification network is composed of a plurality of classifiers and is obtained through learning of an AdaBoost algorithm;
and the automatic fault image identification module is used for realizing automatic identification of the seismic coherence tomography image according to the classification result.
7. The seismic coherence image fault automatic identification system based on the AdaBoost algorithm as claimed in claim 6,
the image block classification module is specifically used for training a classification network, and specifically comprises the following steps:
selecting N off-line training samples to form a first training sample, and learning the first training sample to obtain a first weak classifier, wherein the off-line training samples are image blocks in a seismic coherence tomography image;
forming a second training sample by the wrongly-divided sample in the first weak classifier and other off-line training samples, and obtaining a second weak classifier through learning the second training sample, wherein the second training sample is formed by N off-line training samples;
and forming a third training sample by using the sample which is wrongly divided in the first weak classifier and the second weak classifier and other off-line training samples, obtaining a third weak classifier by learning the third training sample, and forming a classification network formed by three second classes of classifiers, wherein the third training sample is formed by N off-line training samples.
8. The seismic coherence image fault automatic identification system based on the AdaBoost algorithm as claimed in claim 6,
the image block classification module is specifically configured to:
classifying the image blocks through classifiers in the classification network to obtain the image blocks processed by each classifier;
calculating the similarity between the image block and the image block processed by each classifier to obtain a plurality of similarities;
calculating the voting result of each classifier according to the plurality of similarities;
and obtaining a classification result according to the voting result, wherein the classification result is the class with the largest accumulated voting result.
9. The seismic coherence image fault automatic identification system based on the AdaBoost algorithm as claimed in claim 6,
the size of the image blocks is at least one of 3 × 3, 5 × 5, 7 × 7, and 9 × 9.
10. The seismic coherence image fault automatic identification system based on the AdaBoost algorithm as claimed in claim 6,
the selection ratio of the image blocks of the fault body, the image blocks of the fault edge and the image blocks of the fault background in the image blocks is 3: 2: 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710186136.XA CN106951924B (en) | 2017-03-27 | 2017-03-27 | Seismic coherence body image fault automatic identification method and system based on AdaBoost algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710186136.XA CN106951924B (en) | 2017-03-27 | 2017-03-27 | Seismic coherence body image fault automatic identification method and system based on AdaBoost algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106951924A CN106951924A (en) | 2017-07-14 |
| CN106951924B true CN106951924B (en) | 2020-01-07 |
Family
ID=59473783
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710186136.XA Active CN106951924B (en) | 2017-03-27 | 2017-03-27 | Seismic coherence body image fault automatic identification method and system based on AdaBoost algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106951924B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109086773B (en) * | 2018-08-29 | 2022-03-04 | 电子科技大学 | Fault plane identification method based on full convolution neural network |
| CN111582359B (en) * | 2020-04-28 | 2023-04-07 | 新疆维吾尔自治区烟草公司 | Image identification method and device, electronic equipment and medium |
| CN113130018B (en) * | 2021-04-20 | 2023-05-12 | 东北石油大学 | Lithology recognition method based on reservoir meta-target invariant feature description |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6018498A (en) * | 1998-09-02 | 2000-01-25 | Phillips Petroleum Company | Automated seismic fault detection and picking |
| CN103412331A (en) * | 2013-08-30 | 2013-11-27 | 电子科技大学 | Automatic extraction method for three-dimensional earthquake fault |
| CN103592681A (en) * | 2013-09-16 | 2014-02-19 | 电子科技大学 | Signal classification based seismic image horizon tracking method |
| CN105093289A (en) * | 2014-05-22 | 2015-11-25 | 中国石油化工股份有限公司 | Seismic horizon identification method based on image processing |
| CN105659292A (en) * | 2013-11-26 | 2016-06-08 | 界标制图有限公司 | Global grid building unfaulting sequence for complex fault-network topologies |
| CN106154327A (en) * | 2016-08-17 | 2016-11-23 | 中国石油化工股份有限公司 | A kind of method improving hidden fault recognizing precision |
-
2017
- 2017-03-27 CN CN201710186136.XA patent/CN106951924B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6018498A (en) * | 1998-09-02 | 2000-01-25 | Phillips Petroleum Company | Automated seismic fault detection and picking |
| CN103412331A (en) * | 2013-08-30 | 2013-11-27 | 电子科技大学 | Automatic extraction method for three-dimensional earthquake fault |
| CN103592681A (en) * | 2013-09-16 | 2014-02-19 | 电子科技大学 | Signal classification based seismic image horizon tracking method |
| CN105659292A (en) * | 2013-11-26 | 2016-06-08 | 界标制图有限公司 | Global grid building unfaulting sequence for complex fault-network topologies |
| CN105093289A (en) * | 2014-05-22 | 2015-11-25 | 中国石油化工股份有限公司 | Seismic horizon identification method based on image processing |
| CN106154327A (en) * | 2016-08-17 | 2016-11-23 | 中国石油化工股份有限公司 | A kind of method improving hidden fault recognizing precision |
Non-Patent Citations (5)
| Title |
|---|
| Adaboost算法分类器设计及其应用;许剑 等;《四川理工学院学报（自然科学版）》;20140228;第27卷(第01期);第208-214页 * |
| Automated fault detection without seismic processing;Mauricio Araya-Polo;《THE LEADING EDGE》;20170302;第36卷(第03期);第28-31页 * |
| Semi-automatic detection of faults in 3D seismic data;Kristofer M 等;《Geophysical Prospecting》;20051231;第533-542页 * |
| 基于三维地震数据相干信息的断层识别方法研究;李转;《中国优秀硕士学位论文全文数据库 基础科学辑》;20160715(第07期);第55-63页 * |
| 藏汉跨语言话题模型构建与对齐方法研究;孙媛 等;《中文信息学报》;20170131;第31卷(第1期);第102-111页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106951924A (en) | 2017-07-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107657279B (en) | Remote sensing target detection method based on small amount of samples | |
| EP3478728B1 (en) | Method and system for cell annotation with adaptive incremental learning | |
| CN106408030B (en) | SAR image classification method based on middle layer semantic attribute and convolutional neural networks | |
| CN106226050B (en) | A kind of TFDS fault picture automatic identifying method based on convolutional neural networks | |
| Sirmacek et al. | Urban area detection using local feature points and spatial voting | |
| CN104091321B (en) | It is applicable to the extracting method of the multi-level point set feature of ground laser radar point cloud classifications | |
| CN108537751B (en) | Thyroid ultrasound image automatic segmentation method based on radial basis function neural network | |
| CN106846316A (en) | A kind of GIS inside typical defect automatic distinguishing method for image | |
| CN105574550A (en) | Vehicle identification method and device | |
| CN108171119B (en) | SAR image change detection method based on residual error network | |
| CN108333468B (en) | The recognition methods of bad data and device under a kind of active power distribution network | |
| CN103761531A (en) | Sparse-coding license plate character recognition method based on shape and contour features | |
| CN110135514B (en) | Workpiece classification method, device, equipment and medium | |
| CN114972922B (en) | Coal gangue sorting and identifying method, device and equipment based on machine learning | |
| CN106951924B (en) | Seismic coherence body image fault automatic identification method and system based on AdaBoost algorithm | |
| CN105160303A (en) | Fingerprint identification method based on mixed matching | |
| CN111275198A (en) | Bearing abnormity detection method and system | |
| CN104680185B (en) | Hyperspectral image classification method based on boundary point reclassification | |
| CN114776304B (en) | Method, device, equipment and medium for identifying abnormities of deep sea mineral products | |
| US20200034649A1 (en) | Object tracking system, intelligent imaging device, object feature extraction device, and object feature extraction method | |
| CN106529470A (en) | Gesture recognition method based on multistage depth convolution neural network | |
| CN110321890B (en) | Digital instrument identification method of power inspection robot | |
| CN108805181B (en) | Image classification device and method based on multi-classification model | |
| CN106548195A (en) | A kind of object detection method based on modified model HOG ULBP feature operators | |
| CN104537392B (en) | A kind of method for checking object based on the semantic part study of identification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |