A speed spectrum picking method based on a DBSCAN clustering algorithm and a medium
By using the DBSCAN clustering algorithm to classify velocity spectrum datasets and remove outliers, the velocity spectrum picking process is optimized, solving the problems of low efficiency and low reliability in existing technologies, and achieving efficient and reliable velocity spectrum picking.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-10-21
- Publication Date
- 2026-03-17
AI Technical Summary
Existing velocity spectrum picking methods suffer from low efficiency and unreliability, especially manual picking which is extremely time-consuming and automatic picking methods are severely affected by outliers and noise.
The DBSCAN clustering algorithm was used to classify the velocity spectrum dataset, select a velocity pick-up point set, and remove outlier data. The picking results were optimized by setting thresholds and preset conditions.
It improves the efficiency and reliability of velocity spectrum picking, effectively solving the problems of low efficiency and low reliability in existing technologies.
Smart Images
Figure CN116028830B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of seismic exploration, and more specifically, relates to a velocity spectrum picking method and medium based on the DBSCAN clustering algorithm. Background Technology
[0002] When analyzing residual velocity spectra, manual picking is commonly used. While the reliability of manual picking results is high due to the experienced processors, it is inefficient and time-consuming. As the scale of seismic data processing increases, manual picking becomes unsustainable in a short period. With the continuous development of seismic exploration technology, automatic velocity spectrum picking has made significant progress. Neural network methods learn different features in the velocity spectrum to achieve automatic picking. However, this method requires setting a large number of labels and training the network with a large amount of data. Clustering algorithms do not require training; they are unsupervised machine learning methods. The main idea is to identify energy clusters based on spatial location and then select the center of the energy cluster as the picking result. K-means clustering is the most commonly used clustering analysis method, but this method requires pre-setting the number of clusters and is easily affected by outliers and noise points, resulting in less than ideal classification results. In addition, there are methods that use channel sets to pick accurately superimposed velocity spectra and methods that use finite difference algorithms to solve for the optimal trajectory. However, all methods cannot avoid the influence of outliers and noise points, and not all energy clusters in the velocity spectrum need to be picked.
[0003] Therefore, it is hoped that a velocity spectrum picking method based on the DBSCAN clustering algorithm can be invented to effectively solve the problems of low efficiency and low reliability caused by the picking methods in the existing technology. Summary of the Invention
[0004] The purpose of this invention is to propose a velocity spectrum picking method based on the DBSCAN clustering algorithm to solve the problems of low efficiency and low reliability caused by the picking methods in the existing technology.
[0005] To achieve the above objectives, this invention provides a velocity spectrum picking method based on the DBSCAN clustering algorithm, comprising:
[0006] Obtain the dataset used for remaining velocity picking;
[0007] The DBSCAN clustering algorithm is applied to classify the dataset to obtain the classified dataset;
[0008] Based on the remaining velocity spectrum, a set of velocity pickup points is selected from the classified dataset;
[0009] Remove abnormal data from the speed pickup point set.
[0010] Optionally, the dataset used to obtain the remaining velocity pickup includes:
[0011] Define the velocity spectrum matrix as follows:
[0012]
[0013] Where s ij Let v be the energy value of the point in the i-th row and j-th column, v be the column number of the velocity spectrum matrix, g be the row number of the velocity spectrum matrix, i = 1, 2, ..., v, j = 1, 2, ..., g;
[0014] A first target threshold k is set, and multiple energy values in the velocity spectrum matrix Sv×g whose element values are greater than the first target threshold k are selected. These multiple energy values form a first matrix, which is:
[0015]
[0016] For the first matrix S1 v×g Dimensionality reduction is performed to obtain the dataset X used for residual velocity picking, wherein the dataset X is:
[0017] X={x ij = (i, j)|s ij ≠0} #
[0018] ={x1, x2, ..., x n}
[0019] Where (i, j) are the coordinates of the data points, and n is the total number of data points in the dataset.
[0020] Optionally, the classified dataset C is:
[0021] C = {c i}, i = 1, 2, ..., m
[0022] Among them, c i Let m represent a class in the classified dataset X, and let c represent the total number of classes in the dataset X. i ={x i1 x i2 , ..., x in}, where in is less than n.
[0023] Optionally, selecting a set of velocity pickup points from the classified dataset based on the residual velocity spectrum includes:
[0024] Step 31: Determine the interval height h based on the remaining velocity spectrum;
[0025] Step 32: For each class c in the classified dataset i Perform the following steps:
[0026] Determine the class c i Longitudinal span h i The longitudinal span h i for
[0027] h i =|a i -b i |
[0028] Where a i For the class c i The minimum y-coordinate of the data points, b i For the class c i The maximum ordinate of the data points in the middle;
[0029] The longitudinal span hi is compared with the interval height h, and the initial velocity pickup point is determined based on the comparison result;
[0030] Step 33: Sort the initial velocity pickup points of all classes in ascending order of depth to obtain the velocity pickup point set V, wherein the velocity pickup point set V is:
[0031] V = {v1, v2, ..., v} i , ..., v t},
[0032] Among them, v i For the class c i The initial velocity pickup point, t is the number of velocity pickup points in the velocity pickup point set V.
[0033] Optionally, comparing the longitudinal span hi with the interval height h and determining the initial velocity pickup point based on the comparison result includes:
[0034] If hi ≤ h, then class c is selected. i The point with the highest energy value is the initial velocity pickup point;
[0035] If hi > h, then class c will be... i Divided into h i / h intervals, and select the point with the largest energy value in each interval as the initial position velocity pickup point.
[0036] Optionally, the process of removing outlier data from the velocity pickup point set includes:
[0037] Remove speed pickup points from the speed pickup point set V that do not meet the first preset condition;
[0038] Remove velocity pickup points from the velocity pickup point set V that are not within the preset interval;
[0039] Remove speed pickup points from the speed pickup point set V that do not meet the second preset condition.
[0040] Optionally, the step of removing speed pickup points from the speed pickup point set V that do not meet the first preset condition includes:
[0041] Set the minimum distance Dis between two adjacent speed pickup points;
[0042] Two velocity pickup points v can be randomly selected from the velocity pickup point set V. i and v j If |v i -v j If | <Dis, then the two velocity pickup points v are removed. i and v j Points with relatively low energy values.
[0043] Optionally, the preset interval is obtained by deviating from the center line of the velocity spectrum by a predetermined distance along the depth direction to both sides.
[0044] Optionally, the step of removing speed pickup points from the speed pickup point set V that do not meet the second preset condition includes:
[0045] For the velocity pickup point vi, calculate its corresponding preset function value, whereby the preset function is:
[0046]
[0047] Where m i For speed pickup point v i The perpendicular distance to the center line of the velocity spectrum, g i For speed pickup point v i to the adjacent velocity pickup point v j gradient, S i For speed pickup point v i Energy value, The average value of the vertical distance. Energy value S i The average value, For the gradient g i The average value, σm is the vertical distance m i Standard deviation, σ s Energy value S i Standard deviation, σ g For the gradient g i Standard deviation;
[0048] W iCompared with the preset threshold W0, if W i If W ≤ W0, then the velocity pickup point vi is discarded.
[0049] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the velocity spectrum picking method based on the DBSCAN clustering algorithm.
[0050] The beneficial effects of this invention are as follows:
[0051] The present invention discloses a velocity spectrum picking method based on the DBSCAN clustering algorithm, which includes: acquiring a dataset for residual velocity picking; classifying the dataset using the DBSCAN clustering algorithm to obtain a classified dataset; selecting a velocity picking point set from the classified dataset based on the residual velocity spectrum; and removing outlier data from the velocity picking point set. This invention uses the DBSCAN clustering algorithm for automatic velocity picking, continuously adjusting and modifying the velocity picking points, and removing those that do not meet the requirements, ultimately obtaining better velocity picking results. Combining the DBSCAN clustering algorithm with automatic velocity picking not only improves the velocity picking speed but also enhances the reliability of the velocity picking results, effectively solving the problems of low work efficiency caused by manual velocity picking and low reliability of picking results caused by existing automatic velocity picking methods.
[0052] Other features and advantages of the present invention will be described in detail in the following detailed description section. Attached Figure Description
[0053] The above and other objects, features and advantages of the present invention will become more apparent from the more detailed description of exemplary embodiments of the invention in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of the invention.
[0054] Figure 1 A flowchart of a velocity spectrum picking method based on the DBSCAN clustering algorithm according to an embodiment of the present invention is shown.
[0055] Figure 2 A graph showing the relationship between a preset interval and velocity pickup points in a velocity spectrum pickup method based on the DBSCAN clustering algorithm according to an embodiment of the present invention is illustrated.
[0056] Figure 3 A schematic diagram of a high signal-to-noise ratio velocity spectrum in the prior art is shown.
[0057] Figure 4 A schematic diagram of a velocity spectrum picking method based on the DBSCAN clustering algorithm according to an embodiment of the present invention is shown for velocity spectra with high signal-to-noise ratio.
[0058] Figure 5 A schematic diagram of the velocity spectrum with low signal-to-noise ratio in the prior art is shown.
[0059] Figure 6 A schematic diagram of a velocity spectrum picking method based on the DBSCAN clustering algorithm according to an embodiment of the present invention is shown for velocity spectra with low signal-to-noise ratio. Detailed Implementation
[0060] Preferred embodiments of the invention will now be described in more detail. While preferred embodiments of the invention are described below, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
[0061] A velocity spectrum picking method based on the DBSCAN clustering algorithm according to the present invention includes:
[0062] Obtain the dataset used for remaining velocity picking;
[0063] The DBSCAN clustering algorithm is applied to classify the dataset, resulting in a classified dataset.
[0064] Based on the remaining velocity spectrum, a set of velocity pickup points is selected from the classified dataset;
[0065] Remove outlier data from the speed pick-up point set.
[0066] Specifically, the velocity spectrum picking method based on the DBSCAN clustering algorithm of the present invention includes: acquiring a dataset for residual velocity picking; classifying the dataset using the DBSCAN clustering algorithm to obtain a classified dataset; selecting a set of velocity picking points in the classified dataset based on the residual velocity spectrum; and removing outlier data from the velocity picking point set. The present invention uses the DBSCAN clustering algorithm for automatic velocity picking, continuously adjusting and modifying the velocity picking points, and removing those that do not meet the requirements, ultimately obtaining better velocity picking results. Combining the DBSCAN clustering algorithm with automatic velocity picking not only improves the velocity picking speed but also enhances the reliability of the velocity picking results, effectively solving the problems of low work efficiency caused by manual velocity picking and low reliability of picking results caused by existing automatic velocity picking methods.
[0067] Furthermore, DBSCAN (Density-Based Spatial Clustering of Application with Noise) is a density-based clustering method. It defines a cluster as the largest set of density-connected points, enabling the division of regions based on density into clusters, and can discover clusters of arbitrary shapes in noisy data.
[0068] The DBSCAN clustering algorithm divides sample data into three classes based on the neighborhood radius r and the minimum number of points in the neighborhood (MinP):
[0069] a. Center point: If a point has at least MinP points within its radius r, then that point is called the center point;
[0070] b. Boundary point: If a point has fewer than MinP points within its radius r, and is also within the neighborhood of the center point;
[0071] c. Noise point: A point that is neither a center point nor a boundary point is a noise point.
[0072] The clustering process of the DBSCAN algorithm:
[0073] (1) Randomly select a data point that has not yet been classified. If the number of points in the neighborhood of the point is greater than MinP, then mark the point as the center point; if it is less than MinP, then mark the point as a noise point.
[0074] (2) When a center point is found, add the points in its neighborhood to the cluster. If a center point exists in the neighborhood, add the points in the neighborhood of that center point to the cluster. And so on, including the center points in the neighborhoods of the neighborhoods. If a noisy point is added, change the mark of that noisy point to a boundary point.
[0075] Repeat the process in (1)-(2) above until all points are marked, then the clustering process ends.
[0076] In one example, the dataset used to obtain the remaining velocity pickup includes:
[0077] Define the velocity spectrum matrix as follows:
[0078]
[0079] Where s ij Let v be the energy value of the point in the i-th row and j-th column, v be the number of columns in the velocity spectrum matrix, g be the number of rows in the velocity spectrum matrix, i = 1, 2, ..., v, j = 1, 2, ..., g;
[0080] Set the first target threshold k and filter out the velocity spectrum matrix S. v×gMultiple energy values whose element value is greater than the first target threshold k are combined to form a first matrix, which is as follows:
[0081]
[0082] For the first matrix S1 v×g After dimensionality reduction, we obtain dataset X for residual velocity picking. Dataset X is:
[0083] X={x ij = (i, j)|s ij ≠0} #
[0084] ={x1, x2, ..., x n}
[0085] Where (i, j) are the coordinates of the data points, and n is the total number of data points in the dataset.
[0086] In one example, the classified dataset C is:
[0087] C = {c i}, i = 1, 2, ..., m
[0088] Where ci is a class in the classified dataset C, m represents the total number of classes in dataset X, and c i ={x i1 x i2 , ..., x in}, where in is less than n.
[0089] Specifically, each energy cluster will be assigned to a different class based on its radius r and the minimum number of points MinP in its neighborhood. Ideally, points within the same energy cluster will be assigned to the same class, and noise points will be automatically removed. However, classes containing fewer points (i.e., very small energy clusters) can be removed as needed. Each class represents at least one energy cluster; some clusters with overlapping or intersecting energies may be assigned to the same class.
[0090] In one example, based on the residual velocity spectrum, the set of velocity pickup points selected from the classified dataset includes:
[0091] Step 31: Determine the interval height h based on the remaining velocity spectrum;
[0092] Step 32: For each class c in the classified dataset i Perform the following steps:
[0093] Determine class c i Longitudinal span h i Longitudinal span h i for
[0094] h i =|a i -b i |
[0095] Where a i For class c i The minimum y-coordinate of the data points, b i For class c i The maximum ordinate of the data points in the middle;
[0096] The longitudinal span hi is compared with the interval height h, and the initial velocity pickup point is determined based on the comparison result;
[0097] Step 33: Sort the initial velocity pickup points of all classes in ascending order of depth to obtain the velocity pickup point set V. The velocity pickup point set V is:
[0098] V = {v1, v2, ..., v} i , ..., v t},
[0099] Among them, v i For class c i The initial velocity pickup point, t is the number of velocity pickup points in the velocity pickup point set V.
[0100] In one example, comparing the longitudinal span hi with the interval height h, and determining the initial velocity pickup point based on the comparison result includes:
[0101] If hi ≤ h, then class c is selected. i The point with the highest energy value is the initial velocity pickup point;
[0102] If hi > h, then class c i Divided into h i / h intervals, and select the point with the largest energy value in each interval as the initial position velocity pickup point.
[0103] In one example, removing outlier data from the velocity pickup point set includes:
[0104] Remove velocity pickup points from the velocity pickup point set V that do not meet the first preset condition;
[0105] Remove velocity pickup points from the velocity pickup point set V that are not within the preset range;
[0106] Remove velocity pickup points from the velocity pickup point set V that do not meet the second preset condition.
[0107] In one example, velocity pickup points in the velocity pickup point set V that do not meet the first preset condition are excluded, including:
[0108] Set the minimum distance Dis between two adjacent speed pickup points;
[0109] Select any two velocity pickup points v from the velocity pickup point set V. i and v j If |v i -v j If | <Dis, then remove the two velocity pickup points v. i and v j Points with relatively low energy values.
[0110] In one example, a preset interval is obtained by deviating from the center line of the velocity spectrum to both sides by a predetermined distance along the depth direction.
[0111] Specifically, because the degree of focusing of energy clusters varies at different depths, a depth-dependent function is determined to restrict the location of the pickup point. This is achieved by setting a pair of straight lines: y1 = k1x + b1 and y2 = -k1x + b2, which restrict the velocity pickup point. These two functions essentially expand the pickup area outwards from the first pickup point; the greater the depth, the larger the pickup range. Figure 2 As shown, function y1 is determined by (x0, y0) and (x1, y1), and function y2 is determined by (x0, y0) and (x2, y2). (x0, y0) is the first pick-up point, located on the center line of the velocity spectrum. Points a0 and b0 are not within the region enclosed by y1 and y2, so these two points are removed.
[0112] In one example, velocity pickup points in the velocity pickup point set V that do not meet the second preset condition are excluded, including:
[0113] For speed pickup point v i Calculate the corresponding preset function value. The preset function is:
[0114]
[0115] Where m i For speed pickup point v i The perpendicular distance to the center line of the velocity spectrum, g i For speed pickup point v i to the adjacent velocity pickup point v j gradient, S i For speed pickup point v i Energy value, This represents the average vertical distance. Energy value S i The average value, For gradient g i The average value, σ m Vertical distance m i Standard deviation, σs Energy value S i Standard deviation, σ g For gradient g i Standard deviation;
[0116] W i Compared with the preset threshold W0, if W i If W ≤ W0, then the velocity pickup point v is discarded. i .
[0117] Specifically, the farther the velocity pickup point is from the center line, the smaller the energy value, and the greater the gradient between the two points, the easier it is for that point to be eliminated.
[0118] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the velocity spectrum picking method based on the DBSCAN clustering algorithm.
[0119] Example 1
[0120] Figure 1 A flowchart of a velocity spectrum picking method based on the DBSCAN clustering algorithm according to an embodiment of the present invention is shown.
[0121] like Figure 1 The flowchart shown below illustrates the velocity spectrum acquisition method based on the DBSCAN clustering algorithm, including:
[0122] Step 1: Obtain the dataset for remaining speed picking;
[0123] The dataset used to obtain the remaining velocity pickup includes:
[0124] Define the velocity spectrum matrix as follows:
[0125]
[0126] Where s ij Let v be the energy value of the point in the i-th row and j-th column, v be the number of columns in the velocity spectrum matrix, g be the number of rows in the velocity spectrum matrix, i = 1, 2, ..., v, j = 1, 2, ..., g;
[0127] Set the first target threshold k and filter out the velocity spectrum matrix S. v×g Multiple energy values whose element value is greater than the first target threshold k are combined to form a first matrix, which is as follows:
[0128]
[0129] For the first matrix S1 v×g After dimensionality reduction, we obtain dataset X for residual velocity picking. Dataset X is:
[0130] X={x ij = (i, j)|s ij ≠0} #
[0131] ={x1, x2, ..., x n}
[0132] Where (i, j) are the coordinates of the data points, and n is the total number of data points in the dataset.
[0133] The classified dataset C is as follows:
[0134] C = {c i}, i = 1, 2, ..., m
[0135] Among them, c i Let m represent a class in the classified dataset X, and c represent the total number of classes in dataset X. i ={x i1 x i2 , ..., x in}, where in is less than n.
[0136] Step 2: Apply the DBSCAN clustering algorithm to classify the dataset, obtaining the classified dataset;
[0137] Step 3: Based on the remaining velocity spectrum, select a set of velocity pickup points in the classified dataset;
[0138] Among them, the set of velocity pickup points selected from the classified dataset based on the residual velocity spectrum includes:
[0139] Step 31: Determine the interval height h based on the remaining velocity spectrum;
[0140] Step 32: For each class c in the classified dataset i Perform the following steps:
[0141] Determine class c i Longitudinal span h i Longitudinal span h i for
[0142] h i =|a i -b i |
[0143] Where a i For class c i The minimum y-coordinate of the data points, b i For class c i The maximum ordinate of the data points in the middle;
[0144] The vertical span hi The initial velocity pickup point is determined based on the comparison result with the interval height h.
[0145] Step 33: Sort the initial velocity pickup points of all classes in ascending order of depth to obtain the velocity pickup point set V. The velocity pickup point set V is:
[0146] V = {v1, v2, ..., v} i , ..., v t},
[0147] Among them, v i For class c i The initial velocity pickup point, t is the number of velocity pickup points in the velocity pickup point set V.
[0148] Wherein, the longitudinal span h i Compared with the interval height h, the initial velocity pickup points are determined based on the comparison results, including:
[0149] If hi ≤ h, then class c is selected. i The point with the highest energy value is the initial velocity pickup point;
[0150] If hi > h, then class c i Divided into h i / h intervals, and select the point with the largest energy value in each interval as the initial position velocity pickup point.
[0151] Step 4: Remove outlier data from the speed pick-up point set.
[0152] The process of removing outlier data from the velocity pickup point set includes:
[0153] Remove velocity pickup points from the velocity pickup point set V that do not meet the first preset condition;
[0154] Remove velocity pickup points from the velocity pickup point set V that are not within the preset range;
[0155] Remove velocity pickup points from the velocity pickup point set V that do not meet the second preset condition.
[0156] Among them, the velocity pickup points in the velocity pickup point set V that do not meet the first preset condition are excluded, including:
[0157] Set the minimum distance Dis between two adjacent speed pickup points;
[0158] Select any two velocity pickup points v from the velocity pickup point set V. i and v j If |v i -v j If | <Dis, then remove the two velocity pickup points v.i and v j Points with relatively low energy values.
[0159] Specifically, a preset interval is obtained by deviating from the center line of the velocity spectrum to both sides by a predetermined distance along the depth direction.
[0160] Velocity pickup points in the velocity pickup point set V that do not meet the second preset condition are excluded, including:
[0161] For speed pickup point v i Calculate the corresponding preset function value. The preset function is:
[0162]
[0163] Where m i For speed pickup point v i The perpendicular distance to the center line of the velocity spectrum, g i For speed pickup point v i to the adjacent velocity pickup point v j gradient, S i For speed pickup point v i Energy value, This represents the average vertical distance. Energy value S i The average value, For gradient g i The average value, σ n Vertical distance m i Standard deviation, σ s Energy value S i Standard deviation, σ g For gradient g i Standard deviation;
[0164] W i Compared with the preset threshold W0, if W i If W ≤ W0, then the velocity pickup point vi is discarded.
[0165] Specifically, such as Figure 3-6 As shown, the velocity spectrum picking method based on the DBSCAN clustering algorithm of the present invention is applied to, for example... Figure 3 The high signal-to-noise ratio velocity spectrum shown is composed of Figure 4 As shown, large energy clusters were picked up using velocity pickup, and... Figure 3 The in-phase axis is flattened; the velocity spectrum picking method based on the DBSCAN clustering algorithm of this invention is applied to, for example, Figure 5 The low signal-to-noise ratio velocity spectrum shown is due to Figure 6 As shown, by picking up the energy clusters at high speed, a larger energy cluster is also picked up, and at the same time, it also... Figure 5 The in-phase axis in the middle is flattened.
[0166] Example 2
[0167] This disclosure provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned velocity spectrum picking method based on the DBSCAN clustering algorithm.
[0168] A computer-readable storage medium according to embodiments of the present disclosure stores non-transitory computer-readable instructions. When these non-transitory computer-readable instructions are executed by a processor, all or part of the steps of the methods described in the foregoing embodiments of the present disclosure are performed.
[0169] The aforementioned computer-readable storage media include, but are not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or portable hard drive), media with built-in rewritable non-volatile memory (e.g., memory card), and media with built-in ROM (e.g., ROM cartridge).
[0170] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.
Claims
1. A velocity spectrum picking method based on DBSCAN clustering algorithm, characterized in that, The method comprises: acquiring a data set for residual velocity pick-up; applying a DBSCAN clustering algorithm to classify the data set to obtain a classified data set; selecting a velocity pick-up point set from the classified data set based on a residual velocity spectrum; eliminating abnormal data in the velocity pick-up point set; the selecting of the velocity pick-up point set from the classified data set based on the residual velocity spectrum comprises: step 31: determining an interval height h according to the residual velocity spectrum; Step 32: For each class c in the classified dataset i perform the following steps: determining the class c i a longitudinal span h i of the class c i is wherein a i is the minimum longitudinal coordinate of the data points in the class c i is the maximum longitudinal coordinate of the data points in the class c i is the minimum longitudinal coordinate of the data points in the class c i is the maximum longitudinal coordinate of the data points in the class c determining an initial velocity pick-up point in dependence on a comparison of the longitudinal span h i determining an initial velocity pick-up point in dependence on a comparison of the longitudinal span h step 33: sorting all the initial velocity pick-up points of the classes in order from low to high in depth to obtain the velocity pick-up point set V, the velocity pick-up point set V being: ， wherein v i is the initial velocity pickup point of the class c i , and t is the number of velocity pickup points in the set V of velocity pickup points. the eliminating of the velocity pick-up points in the velocity pick-up point set V that do not satisfy a first preset condition comprises: setting a minimum distance Dis between adjacent two velocity pick-up points; Any two speed pickup points v i and v j are selected from the speed pickup point set V , if i and v j are the two speed pickup points with the smallest energy value, the one with smaller energy value is removed. obtaining a preset interval by deviating from a center line of the velocity spectrum in a depth direction to both sides by a predetermined distance, and eliminating the velocity pick-up points that are not in the preset interval; the eliminating of the velocity pick-up points in the velocity pick-up point set V that do not satisfy a second preset condition comprises: For the speed pick-up point v i , the corresponding preset function value is calculated, which is: Where m i For speed pickup point v i The perpendicular distance to the center line of the velocity spectrum, g i For speed pickup point v i to the adjacent velocity pickup point v j gradient, S i For speed pickup point v i Energy value, The average value of the vertical distance. Energy value S i The average value, For the gradient g i The average value, The vertical distance m i standard deviation Energy value S i standard deviation For the gradient g i Standard deviation; The W i If compared with a preset threshold W0, if The speed pickup point v i is rejected.
2. The velocity spectrum picking method based on DBSCAN clustering algorithm according to claim 1, characterized in that, the acquiring of the data set for residual velocity pick-up comprises: defining a velocity spectrum matrix, the velocity spectrum matrix being: ， wherein is the first row, the first column of the energy values of the points, v is the number of columns of the velocity spectrum matrix, g is the number of rows of the velocity spectrum matrix, ; Setting a first target threshold k, screening out the speed spectrum matrix S v×g The plurality of energy values whose middle element values are greater than the first target threshold k, the plurality of energy values constitute a first matrix, and the first matrix is: ； performing dimensionality reduction on the first matrix S1 v×g performing dimensionality reduction on the first matrix S1 v×g performing dimensionality reduction on the first matrix S1 v×g performing dimensionality reduction on the first matrix S1 v×g performing dimensionality reduction on the first matrix S1 v×g perform wherein are coordinates of data points, n is the total number of data points in the dataset.
3. The velocity spectrum picking method based on DBSCAN clustering algorithm according to claim 2, characterized in that, the classified data set C is: where c i is one class in the classified dataset C, m represents the total number of classifications of the dataset X, where in is less than n.
4. The velocity spectrum picking method based on DBSCAN clustering algorithm of claim 1, wherein, said determining the initial velocity pick-up point comprises determining the initial velocity pick-up point based on a comparison of the longitudinal span h i determining the initial velocity pick-up point based on a comparison of the longitudinal span h with the spacing height h comprises If , the point with the maximum energy value in the class c i is selected as the initial speed pickup point. If , the class c i is divided into h i / h intervals, and the point with the maximum energy value in each interval is selected as the initial speed pickup point.
5. A computer readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to implement the velocity spectrum pick-up method based on the DBSCAN clustering algorithm in any one of claims 1-4.