CN110349242A - The oil exploration datum target automatic tracing and method for visualizing recommended automatically based on more seed points - Google Patents
The oil exploration datum target automatic tracing and method for visualizing recommended automatically based on more seed points Download PDFInfo
- Publication number
- CN110349242A CN110349242A CN201910506168.2A CN201910506168A CN110349242A CN 110349242 A CN110349242 A CN 110349242A CN 201910506168 A CN201910506168 A CN 201910506168A CN 110349242 A CN110349242 A CN 110349242A
- Authority
- CN
- China
- Prior art keywords
- point
- seed
- density
- data
- seed point
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- 230000000007 visual effect Effects 0.000 claims abstract description 8
- 230000008569 process Effects 0.000 claims abstract description 7
- 238000004458 analytical method Methods 0.000 claims abstract description 6
- 238000005516 engineering process Methods 0.000 claims abstract description 4
- 239000013076 target substance Substances 0.000 claims abstract description 4
- 230000015572 biosynthetic process Effects 0.000 claims description 18
- 238000012800 visualization Methods 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 11
- 238000012360 testing method Methods 0.000 claims description 11
- 230000000694 effects Effects 0.000 claims description 7
- 238000007794 visualization technique Methods 0.000 claims description 6
- 241001269238 Data Species 0.000 claims description 5
- 238000005259 measurement Methods 0.000 claims description 4
- 238000005266 casting Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 238000009877 rendering Methods 0.000 claims description 3
- 230000001360 synchronised effect Effects 0.000 claims description 3
- 238000013461 design Methods 0.000 claims description 2
- 238000001914 filtration Methods 0.000 claims description 2
- 238000000605 extraction Methods 0.000 abstract description 8
- 230000004807 localization Effects 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 19
- 238000005755 formation reaction Methods 0.000 description 15
- 230000008859 change Effects 0.000 description 2
- 230000002452 interceptive effect Effects 0.000 description 2
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 2
- 239000003208 petroleum Substances 0.000 description 2
- 240000001439 Opuntia Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000739 chaotic effect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000035800 maturation Effects 0.000 description 1
- 239000003345 natural gas Substances 0.000 description 1
- 239000003209 petroleum derivative Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T15/00—3D [Three Dimensional] image rendering
- G06T15/005—General purpose rendering architectures
Landscapes
- Engineering & Computer Science (AREA)
- Computer Graphics (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention discloses a kind of oil exploration datum target automatic tracing and method for visualizing recommended automatically based on more seed points, specifically comprise the following steps: that (1) pre-processes the artificial earthquake back wave slice of data that oil exploration personnel are obtained by artificial earthquake wave technology;(2) more seed points are recommended;(3) the more seed points obtained using recommendation, carry out more seed point target substance automatic tracings;(4) exploration and analysis of result are tracked.The present invention can be improved the extraction efficiency of buried channel, extract the accuracy and integrality of result, the integrality for wherein extracting result, which is mainly reflected in, solves the problems, such as data characteristics localization, and final result can be scanned for and be analyzed, it can largely reduce visual confusion and removal ambient noise.
Description
Technical field
The present invention relates to artificial earthquake data to explain visualization technique field, recommends multiple seed points automatically, and according to pushing away
The more seed point automatic tracings buried channel (UFP) recommended is proposed a kind of petroleum recommended automatically based on more seed points and surveyed
Visit datum target automatic tracing and method for visualizing.
Background technique
During oil exploration, in order to predict that the distribution situation of petroleum or natural gas, common technology are sent out to stratum
Artificial Seismic Wave is penetrated, and volume of data is obtained according to the hourage of back wave, amplitude and waveform, to judge buried target body
Spatial position and structure.In analysis complex data, especially science data, such as earthquake reflective data, ground location thunder
Up to data, Reservoir Data and weather analogue data etc., visualization has extremely strong potentiality.In recent years, pass through oil exploration number
The distribution situation that earth formation and subsurface material can be obtained is explained according to (visiting ground data), and by current data with vision
This very intuitive way is presented and is supplied to user, and carries out interactive exploration.
In recent years, in order to improve the robustness of UFP extraction process and extract the accuracy of result, Liu etc. proposes a kind of base
In progressive visualization method.But this method needs all seeds being placed on each local dense region.A this step one
The method of step placement seed point, is also very dependent on the experience and pertinent arts of user.Also, due to being difficult to pass through interaction
Select the most intensive point in some areas, the accuracy that more seed points are placed be not often it is very high, there is an urgent need to have more intelligence with
Automatic more seed point Placement Strategies.
In general, there are also many limitations for the method for extraction UFP at present, such as: it is inefficient, it is endless to extract result
It is whole, extract result inaccuracy.This is mainly due to four data characteristicses of oil exploration data, it may be assumed that noise is big, it is discontinuous, differentiate
Rate is low and feature localization caused by.
Summary of the invention
Technical problem to be solved by the present invention lies in provide a kind of oil exploration number recommended automatically based on more seed points
According to target automatic tracing and method for visualizing, efficiency, the integrality and standard of extraction result that multiple UFP are extracted parallel can be improved
True property, and solve data characteristics Localization Problems, and can final result be explored and be analyzed, largely
Reduce visual confusion and removal noise structure.
In order to solve the above technical problems, the present invention provides a kind of oil exploration data mesh recommended automatically based on more seed points
Automatic tracing and method for visualizing are marked, is included the following steps:
(1) oil exploration personnel are carried out by the artificial earthquake back wave slice of data that artificial earthquake wave technology obtains pre-
Processing;All slices are arranged according to true physical space, to obtain three-dimensional data;The said three-dimensional body that system will be handled well
Data are loaded into system, provide data branch for the recommendation of subsequent seed point, seed point tracking, tracking result visualization and exploration
It holds;
(2) more seed points are recommended;To solve the problems, such as that data characteristics localizes, each regional area in three-dimensional data
A seed point is needed, at least to guarantee that all local buried channels completely are tracked out;It is measured and is counted by comprehensive distance
It calculates, primary filtration obtains candidate seed point, and is directed to these candidate seed points, further provides a kind of based on the close of kernel function
Spend gradient calculation method;
(3) the more seed points obtained using recommendation, carry out more seed point target substance automatic tracings;Per thread corresponding one
All buried channel structures finally are tracked out by a seed point in global three-dimensional data;Tracing process using area increases
Long algorithm, multiple threads increase, to track all voxels of entire data from respective seed point to neighborhood direction;
(4) exploration and analysis of result are tracked;It is hidden intuitively to edit and exploring more seed point trackings as a result, reducing vision
Gear, devises the graph structure of different node types to explore and analyze tracking result.
Preferably, in step (1), artificial earthquake back wave slice of data is pre-processed, is specifically comprised the following steps:
(11) original seismic exploration data is subjected to slice arrangement, the reflection interval between being sliced is 1 millisecond;
(12) all data slicers are aligned one by one with actual physics space by arranging well data according to reflection interval, it is right
Data after neat are three-dimensional data;
(13) these three-dimensional datas are exactly transferred on GPU by the effect of visualization of three-dimensional space from CPU, using Ray-
The volume rendering algorithm of casting carries out volume drawing.
Preferably, in step (2), a kind of more seed density gradient calculation methods based on kernel function are specifically included as follows
Step:
(21) pattern match is carried out according to the feature of initial seed point or mode;Specific method is according to initial seed point
It is (comprehensive to obtain comprehensive measurement value in entire data field for intensity value and the local density values and density level bands angle value of its position
Distance) it is less than all candidate seed points of some threshold value;Carry out comprehensive measurement value calculate when, intensity value, local density values and
The weight of density level bands angle value is defaulted as identical, and user can also be adjusted according to data feature itself;
(22) a kind of density gradient calculation method based on kernel function is designed;Specific method is carried out from candidate seed point
Density value calculates, and the candidate point density center (point of thick) of each regional area will become the consequently recommended kind in each region
Sub-, all regional areas recommend seed point that will constitute seed point set;
(23) user is allowed to calculate density gradient using different kernel functions according to different application scenarios, using different
Kernel function mainly uses different weights in distance metric.
Preferably, in step (22), a kind of density gradient calculation method based on kernel function specifically comprises the following steps:
The dot density calculating of each x based on kernel function is written as:
Wherein xiThe all the points in x neighborhood for being r for radius, dtIt is public for the sum of kernel function k (x) all density values calculated
In formula (1) the distance between each point be defined as candidate seed point density contrast or density gradient it is poor, in order to make formula (1) formula most
Greatly to get most point off density is arrived, its derivative is calculated:
Enabling formula (2) is 0, obtains the maximum point of density value in x neighborhood, which is the smallest point of density contrast in x neighborhood;
After obtaining the maximum point of density value in x neighborhood, using it as new data point x ', equally can in x ' neighborhood
A most point off density is obtained, always iteration, until new most point off density position is almost consistent (small with upper one most point off density position
In the threshold value of some very little) until;The final most point off density obtained when iteration stopping, the seed point of as current regional area,
It is also the point recommended in seed point set, is used for subsequent seed point automatic tracing.
Preferably, in step (23), more seed points select the implementation method of different IPs when recommending automatically, specifically include following step
It is rapid:
(231) 5 kinds of optional kernel functions are used, are Gaussian core, Epanechnikov core, Rectangular respectively
Core, Logistic core and Silverman core, it is different that the interior density put of different kernel representation neighborhoods calculates weight;
(232) these kernel functions are tested using multiple reference datas, passes through the regional area of the seed point of recommendation
Integrality (it is recommended that each regional area at least needs a seed point), accuracy (density value, intensity value maximum needs
Have seed point) and three indexs of performance on overall merit, last test the result shows that, Gaussian core effect we survey
It is showed in multiple data of examination optimal.
Preferably, in step (3), a kind of more seed method for automatic tracking for supporting Multi-thread synchronization function are specifically included
Following steps:
(31) earth formation is tracked using the method that region increases (region growing), and it is carried out more
Thread synchronization processing is to obtain seed point tracking result;
(32) each seed point will start an individual worker thread, for finding out the candidate in current seed point neighborhood
Point, these candidate points will not only meet strength range condition, also meet density conditions;If candidate point meets these conditions,
It is then finally referred in the point set of prospect earth formation, otherwise will be regarded as background dot or noise spot;New candidate point is again
Region growth can be carried out as new seed point;In order to avoid the repetition of identical voxel is tracked, all threads will be synchronized, it is any
One will not be considered again by the voxel that worker thread was tracked.
Preferably, in step (4), figure (graph) method for visualizing with different node types is designed, is specifically described
Are as follows:
Because the earth formation that more seed back tracking methods are extracted is not necessarily UFP, in order to reduce view in graphic searching data
Feel chaotic, when seed point is too many, the entire subgraph of selection can be selected or cancelled by clicking hub node.
It is of the invention main to have the beneficial effect that
The present invention does not need to place seed step by step manually, it is only necessary on ground before automatically extracting earth formation
It shakes and roughly places an initial seed on any slice of data, according to the feature mode of selected initial seed, using being based on
The method of kernel function recommends seed automatically;The seed generates scheme to initial seed position and insensitive, unless placing at the beginning
The complete mistake of seed point location, such as when extracting UFP structure, initial seed point is placed in sheet sand structure.
The present invention can extract the corresponding UFP structure of multiple regional areas parallel in entire earth formation;With tradition side
Method compares, and parallel extraction algorithm effect is very high;Traditional semiautomatic extraction method based on the cutting of interactive body can only once mention
Take a UFP structure;And this method of body cutting is quite time-consuming, and part steps also need prolonged artificial participation, artificially
It is often also not high to participate in part precision;Therefore, it is compared with the traditional method, efficiency of the present invention greatly improves, and precision degree also obtains
Greatly promoted;In addition to only needing oil exploration expert to place an initial seed point according to their domain knowledge at the beginning
Outside, whole process is the automatic calculating process of computer entirely, i.e., more seed points are recommended and more seed point automatic tracings automatically.
In addition to this, present invention itself is not limited to extract UFP, can also extract other earth formations, such as triangle
Continent, flood plain, the structures such as sheet sand.
Detailed description of the invention
Fig. 1 is that the more seed points of the present invention are recommended and the overall flow schematic diagram of UFP automatic tracing automatically.
Fig. 2 is that seed point of the present invention recommends to obtain seed point schematic illustration automatically.
Fig. 3 is the automatic different kernel function schematic diagrames for recommending to use when seed point of the present invention.
Fig. 4 is present invention pretreatment original data flow schematic diagram.
Fig. 5 is the more automatic recommended flowsheet schematic diagrames of seed of the present invention.
Fig. 6 is the more seed point automatic tracing flow diagrams of the present invention.
Fig. 7 (a) is the incomplete non-optimal seed and chase after that Rectangular kernel function generates in data set I of the present invention
Track result schematic diagram.
Fig. 7 (b) is that user of the present invention can delete several uninterested seeds and corresponding tracking result schematic diagram.
Fig. 7 (c) is that user of the present invention can add multiple seeds according to the domain knowledge of oneself and add corresponding tracking
Result schematic diagram.
Fig. 7 (d) is the incomplete non-optimal seed of rectangle kernel function generation and corresponding to chase after in data set II of the present invention
Track result schematic diagram.
Fig. 7 (e) is that user of the present invention can add multiple seeds according to the domain knowledge of oneself and corresponding tracking result
Schematic diagram.
Fig. 8 (a) is that the present invention selects earth formation schematic diagram of all nodes to show all retrospects.
Fig. 8 (b) is that the present invention cancels the three node schematic diagrames chosen in circle.
Fig. 8 (c) is after the present invention cancels selection super node, to hide all earth formations shown in No. 33 node subgraph
Schematic diagram.
Fig. 9 (a) automatically generates schematic diagram by gaussian kernel function for seed of the present invention.
Fig. 9 (b1) is all node schematic diagrames in user of the present invention exploration graph structure.
Fig. 9 (b2) is the corresponding exploration result schematic diagram of the present invention.
Fig. 9 (c1) is that user of the present invention cancels the node schematic diagram selected in circle.
Fig. 9 (c2) is that user of the present invention cancels the node schematic diagram selected in circle.
Fig. 9 (d) is the visualization result schematic diagram that cross display pattern of the present invention is combined with the earth formation of extraction.
Specific embodiment
As shown in Figure 1, a kind of oil exploration datum target automatic tracing recommended automatically based on more seed points and visualization
Method specifically comprises the following steps:
(1) artificial earthquake reflected waveform data is pre-processed;
(2) propose that a kind of more seed density gradient calculation methods based on kernel function are recommended based on continuous Scale-space theory
More seed points;
(3) to solve the problems, such as that data characteristics localizes, propose that a kind of more seeds for supporting Multi-thread synchronization function are automatic
Method for tracing;
It (4) is intuitive editor and exploration tracking as a result, design has the graph structure of different node types.
As shown in figure 4, detailed process is as follows for step (1): firstly, original seismic exploration data is carried out slice arrangement
(reflection interval between slice is 1 millisecond).Then well data are arranged so that all data according to reflection interval
Slice is aligned with actual physics space.Data after alignment are three-dimensional data.Last three-dimensional space can
Exactly these three-dimensional datas are transferred on GPU from CPU depending on change effect and carry out volume drawing, at present the existing maturation of volume drawing
Algorithm, we are using Ray-casting volume rendering algorithm.
As shown in figure 5, the specific implementation steps are as follows for step (2): firstly, being calculated using the region growing of tracking UFP structure
Method.Since the noise of seismic data is big, discontinuity, differentiates the features such as low, each regional area at least needs a seed
Point, we recommend multiple seed points out according to continuous Scale-space theory automatically thus, recommend automatically for subsequent more seed points
Service is provided.
One initial seed is placed on user and currently wants on any position of the target UFP extracted that the position only needs
It is a rough position, because experiments have shown that entire algorithm is to initial seed point position and insensitive.For user, (petroleum is surveyed
Visit domain expert) for, the position of the initial seed point can be simply placed according to their domain knowledge.Due to earthquake number
According to discontinuous, noise it is big, differentiate the characteristics of low feature, especially feature localize, there is only Mr. Yus for all target substances
A little regional areas, and underground distribution and it is discontinuous.So each regional area at least needs a seed point to carry out
Subsequent more seed point trackings, so needing multiple seed points in entire artificial earthquake data;For the spy of continuous scale space
Point recommends seed candidate point according to the feature mode of initial seed automatically.Feature mode includes the intensity value of seed point itself, institute
Locate the density value of the candidate point of regional area and the gradient value of the density.For this purpose, devising a kind of density based on kernel function
Gradient Iteration numerical procedure, single-step iteration are as shown in Figure 2.
Wherein, the specific implementation steps are as follows for a kind of density gradient calculation method based on kernel function:
The dot density calculating of each x based on kernel function can be written as:
Wherein xiThe all the points in x neighborhood for being r for radius, dtIt is public for the sum of kernel function k (x) all density values calculated
In formula (1) the distance between each point can be defined as candidate seed point density contrast or density gradient it is poor, in order to make (1) formula most
Greatly to get most point off density is arrived, its derivative is calculated:
Enabling formula (2) is 0, is calculated by fine granularity, is further obtained all on the slice with similar features mode
Point.
Wherein, the specific implementation steps are as follows for the kernel function: firstly, devising 5 kinds to test core result accuracy
Kernel function is Gaussian core, Epanechnikov core, Rectangular core, Logistic core and Silverman respectively
Core.Then these kernel functions are tested, finally finds that Gaussian core result precision is relatively high, effect is more preferable.It is all
Candidate kernel function is as shown in Figure 3.Default uses Gaussian core in the present invention.
As shown in fig. 6, the specific implementation steps are as follows for step (3): firstly, using region growing algorithm to earth formation into
Row tracking, and carry out Multi-thread synchronization processing to it and obtain seed point tracking result.Secondly as the feature of data localizes,
Further growth of the regional area since a seed point, each seed are realized using the data point expanded function of tracing algorithm
Point will start an individual worker thread.Finally, the repetition in order to avoid identical voxel is tracked, all threads are synchronized.
In order to test the function that user deletes seed point, we use one of them non-optimal Rectangular core letter
Number recommends some non-optimal seed points out, as shown in Fig. 7 (a) to simulate this operational requirements.Subgraph in Fig. 7 (a)-(e)
Left side shows seed, and right side shows corresponding more seed tracking results.User can increase according to domain knowledge deletes seed point
It is finely adjusted.Fig. 7 (a)-(e) shows the test result of the gradual seed fine tuning of data set I and data set II.Seed is raw
Cheng Hou, each seed will start a thread automatic tracing UFP structure.In order to reduce visual confusion, removal domain expert according to
The noise daughter structure of domain knowledge assessment devises the visualization of figure, allow user's exploration automatic tracing UFP structure and its
Distribution.User can choose or cancel selection node and carrys out the corresponding sub- volume structure of show or hide.
Fig. 7 (a)-(e) is the test result that the present invention is obtained using test data set.It include that automatic recommendation obtains in figure
Seed point (left side of each subgraph) and more seed point tracking results (right side of each subgraph).Fig. 7 (a) (left figure) is data
Collect the incomplete non-optimal seed and tracking result (Fig. 7 (a) right figure) that Rectangular kernel function generates in I.Fig. 7 (b) is
User can delete several uninterested seeds (circle) (left figure) and corresponding tracking result (right figure).Fig. 7 (c) is to use
Family can add multiple seeds (in circle) according to the domain knowledge of oneself, and add corresponding tracking result (right figure).Fig. 7
(d) (right for the incomplete non-optimal seed (left figure) of rectangle kernel function generation and corresponding tracking result in data set II
Figure).Fig. 7 (e) is that user can add multiple seeds (circle according to the domain knowledge of oneself and corresponding tracking result (right figure)
Circle).
Fig. 8 (a)-(c) is result of this test case to data set I by figure visualization Exploring Analysis automatic tracing.Fig. 8
(a) earth formation for all nodes of selection to show all retrospects.Fig. 8 (b) is to cancel the three nodes (neck chosen in circle
Domain expert thinks that they are noises).Fig. 8 (c) is after cancelling selection super node, to hide shown in No. 33 node subgraph and own
Earth formation.
Fig. 9 (a)-(d) is result of this test case to data set II by figure visualization Exploring Analysis automatic tracing.Figure
9 (a) are automatically generated for seed by gaussian kernel function.Fig. 9 (b1-b2) is all nodes in user's exploration graph structure (b1), and
Corresponding exploration result (b2).For the node in user's cancellation selection circle, (domain expert thinks that it is one to Fig. 9 (c1-c2)
Noisy earth formation).Fig. 9 (d) is the visualization result that cross display pattern is combined with the earth formation of extraction.
Fig. 8 (a)-(c) be using figure visualization function to data set I automatic tracing to the mistake explored of UFP structure
Journey.The figure that Fig. 9 (a)-(d) is II for data sets visualizes heuristic process.Test result shows in the visual help of figure
Under, it can utmostly reduce visual occlusion and eliminate non-targeted structure of matter bring visual occlusion.
Claims (7)
1. a kind of oil exploration datum target automatic tracing recommended automatically based on more seed points and method for visualizing, feature are existed
In specifically comprising the following steps:
(1) the artificial earthquake back wave slice of data that oil exploration personnel are obtained by artificial earthquake wave technology is located in advance
Reason;All slices are arranged according to true physical space, to obtain three-dimensional data;The said three-dimensional body number that system will be handled well
According to being loaded into system, data branch is provided for the recommendation of subsequent seed point, seed point tracking, tracking result visualization and exploration
It holds;
(2) more seed points are recommended;To solve the problems, such as that data characteristics localizes, in three-dimensional data, each regional area is at least
A seed point is needed, to guarantee that all local buried channels are tracked out to come completely;By comprehensive distance metric calculation, just
Candidate seed point is obtained by filtration in step, and is directed to these candidate seed points, further provides a kind of density level bands based on kernel function
Spend calculation method;
(3) the more seed points obtained using recommendation, carry out more seed point target substance automatic tracings;Per thread is one kind corresponding
It is sub-, finally all buried channel structures are tracked out in global three-dimensional data;Tracing process using area, which increases, to be calculated
Method, multiple threads increase, to track all voxels of entire data from respective seed point to neighborhood direction;
(4) exploration and analysis of result are tracked;For intuitively edit and explore more seed point trackings as a result, reduce visual occlusion,
The graph structure of different node types is designed to explore and analyze tracking result.
2. the oil exploration datum target automatic tracing recommended automatically based on more seed points as described in claim 1 and visualization
Method, which is characterized in that in step (1), artificial earthquake back wave slice of data is pre-processed, following step is specifically included
It is rapid:
(11) original seismic exploration data is subjected to slice arrangement, the reflection interval between each slice is 1 millisecond;
(12) all data slicers are aligned by arranging well data according to reflection interval one by one with actual physics space, are aligned it
Data afterwards are three-dimensional data;
(13) these three-dimensional datas are exactly transferred on GPU by the effect of visualization of three-dimensional space from CPU, using Ray-
The volume rendering algorithm of casting carries out volume drawing.
3. the oil exploration datum target automatic tracing recommended automatically based on more seed points as described in claim 1 and visualization
A kind of method, which is characterized in that in step (2), more seed density gradient calculation methods based on kernel function are specifically included as follows
Step:
(21) pattern match is carried out according to the feature of initial seed point or mode;Specific method is the intensity according to initial seed point
The local density values and density level bands angle value of value and its position obtain comprehensive measurement value in entire data field and are less than some
All candidate seed points of threshold value;When carrying out the calculating of comprehensive measurement value, the power of intensity value, local density values and density level bands angle value
It is defaulted as again identical, user can also be adjusted according to data feature itself;
(22) a kind of density gradient calculation method based on kernel function is designed;Specific method is to carry out density from candidate seed point
Value calculates, and the candidate point density center of each regional area will become the consequently recommended seed point in each region, all regional areas
Recommend seed point that will constitute seed point set;
(23) allow user to calculate density gradient using different kernel functions according to different application scenarios, use different core letters
Number mainly uses different weights in distance metric.
4. the oil exploration datum target automatic tracing recommended automatically based on more seed points as claimed in claim 3 and visualization
A kind of method, which is characterized in that in step (22), density gradient calculation method based on kernel function specifically comprises the following steps:
The dot density calculating of each x based on kernel function is written as:
Wherein xiThe all the points in x neighborhood for being r for radius, dtFor the sum of kernel function k (x) all density values calculated, formula
(1) in the distance between each point be defined as candidate seed point density contrast or density gradient it is poor, in order to keep formula (1) formula maximum,
Most point off density is obtained, its derivative is calculated:
Enabling formula (2) is 0, obtains the maximum point of density value in x neighborhood, which is the smallest point of density contrast in x neighborhood;
After obtaining the maximum point of density value in x neighborhood, using it as new data point x ', one is similarly obtained in x ' neighborhood
Most point off density, iteration always, until new most point off density position and almost consistent upper one most point off density position;Iteration is stopped
The final most point off density obtained when only, the seed point of as current regional area, and recommend a point in seed point set,
For subsequent seed point automatic tracing.
5. the oil exploration datum target automatic tracing recommended automatically based on more seed points as claimed in claim 3 and visualization
Method, which is characterized in that in step (23), more seed points select the implementation method of different IPs when recommending automatically, specifically include as follows
Step:
(231) use 5 kinds of optional kernel functions, be respectively Gaussian core, Epanechnikov core, Rectangular core,
Logistic core and Silverman core, it is different that the interior density put of different kernel representation neighborhoods calculates weight;
(232) these kernel functions are tested using multiple reference datas, the regional area by the seed point of recommendation is complete
Overall merit in three property, accuracy and performance indexs, last test the result shows that, Gaussian core effect is in the more of test
It is showed in a data optimal.
6. the oil exploration datum target automatic tracing recommended automatically based on more seed points as described in claim 1 and visualization
A kind of method, which is characterized in that in step (3), more seed method for automatic tracking for supporting Multi-thread synchronization function are specifically included
Following steps:
(31) earth formation is tracked using the method that region increases, and carries out Multi-thread synchronization processing to it to obtain kind
Sub- point tracking result;
(32) each seed point will start an individual worker thread, for finding out the candidate point in current seed point neighborhood,
These candidate points will not only meet strength range condition, also meet density conditions;If candidate point meets these conditions, most
It is referred in the point set of prospect earth formation at last, otherwise will be regarded as background dot or noise spot;New candidate point again can conduct
New seed point carries out region growth;In order to avoid the repetition of identical voxel is tracked, all threads will be synchronized, any one is by work
The voxel tracked as thread will not be considered again.
7. the oil exploration datum target automatic tracing recommended automatically based on more seed points as described in claim 1 and visualization
Method, which is characterized in that in step (4), design the figure method for visualizing with different node types, specifically describe are as follows:
It is mixed in order to reduce vision in graphic searching data because the earth formation that more seed back tracking methods are extracted is not necessarily UFP
Disorderly, when seed point is too many, the entire subgraph of selection is selected or cancelled by clicking hub node.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910506168.2A CN110349242A (en) | 2019-06-12 | 2019-06-12 | The oil exploration datum target automatic tracing and method for visualizing recommended automatically based on more seed points |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910506168.2A CN110349242A (en) | 2019-06-12 | 2019-06-12 | The oil exploration datum target automatic tracing and method for visualizing recommended automatically based on more seed points |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110349242A true CN110349242A (en) | 2019-10-18 |
Family
ID=68181855
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910506168.2A Pending CN110349242A (en) | 2019-06-12 | 2019-06-12 | The oil exploration datum target automatic tracing and method for visualizing recommended automatically based on more seed points |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110349242A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113505798A (en) * | 2021-06-22 | 2021-10-15 | 浙江工业大学 | Time-varying data feature extraction and tracking method |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150219779A1 (en) * | 2014-02-03 | 2015-08-06 | Schlumberger Technology Corporation | Quality control of 3d horizon auto-tracking in seismic volume |
CN107741602A (en) * | 2017-09-29 | 2018-02-27 | 中国石油化工股份有限公司 | A kind of reservoir method for automatic tracking based on seismic wave characteristic |
-
2019
- 2019-06-12 CN CN201910506168.2A patent/CN110349242A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150219779A1 (en) * | 2014-02-03 | 2015-08-06 | Schlumberger Technology Corporation | Quality control of 3d horizon auto-tracking in seismic volume |
CN107741602A (en) * | 2017-09-29 | 2018-02-27 | 中国石油化工股份有限公司 | A kind of reservoir method for automatic tracking based on seismic wave characteristic |
Non-Patent Citations (2)
Title |
---|
丁晓凤: "基于MEAN_SHIFT的多模板目标跟踪算法的研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
魏强等: "基于CUDA的地震数据层位面自动追踪算法", 《计算机技术与发展》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113505798A (en) * | 2021-06-22 | 2021-10-15 | 浙江工业大学 | Time-varying data feature extraction and tracking method |
CN113505798B (en) * | 2021-06-22 | 2024-03-22 | 浙江工业大学 | Feature extraction and tracking method for time-varying data |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP2496967B1 (en) | Method for creating a hierarchically layered earth model | |
US9128204B2 (en) | Shape-based metrics in reservoir characterization | |
AU2009341103B2 (en) | Classifying potential hydrocarbon reservoirs using electromagnetic survey information | |
CN104636980B (en) | Collect the geophysics characterizing method of condition for channel reservoir type oil gas | |
US20150309197A1 (en) | Method and System for Geophysical Modeling of Subsurface Volumes Based on Label Propagation | |
Raska | 3D geologic subsurface modeling within the Mackenzie Plain, Northwest Territories, Canada | |
Sun et al. | GIS-based regional assessment of seismic site effects considering the spatial uncertainty of site-specific geotechnical characteristics in coastal and inland urban areas | |
CN109425899A (en) | A kind of prediction technique and device of the distribution of carbonate rock fault belt | |
Ismail et al. | Unsupervised machine learning and multi-seismic attributes for fault and fracture network interpretation in the Kerry Field, Taranaki Basin, New Zealand | |
US20160377752A1 (en) | Method of Digitally Identifying Structural Traps | |
CN110349242A (en) | The oil exploration datum target automatic tracing and method for visualizing recommended automatically based on more seed points | |
US11080525B1 (en) | Methods of identifying flying objects using digital imaging | |
Clausolles et al. | Generating variable shapes of salt geobodies from seismic images and prior geological knowledge | |
CN106157369B (en) | Geological surface triangulation methodology based on sparse structure interpretation data | |
Mushayandebvu et al. | Subsurface geometry of the Revell Batholith by constrained geophysical modelling, NW Ontario, Canada | |
Kozak | Comparison of fracture detection methods applied on the Kerry 3D seismic, Taranaki Basin, New Zealand | |
CN116256801B (en) | Deep oil gas accurate navigation fault characterization method and system based on image fusion | |
EP2715403A1 (en) | Two-way wave equation targeted data selection for improved imaging of prospects among complex geologic structures | |
Stork | Wave equation illumination using RTM | |
Li et al. | Virtual Reality and Computer-Aided 3D Seismic Exploration Data Acquisition Planning and Design System Optimization | |
Halpert | Interpreter-driven automatic image segmentation and model evaluation | |
Luo | Advanced application of ground-penetrating radar in underground tree root systems detection and mapping | |
Cardinal | Site Identification, Delineation, and Evaluation through Quantitative Spatial Analysis: Geostatistical and GIS Methods to Facilitate Archaeological Resource Assessment | |
CN116299717A (en) | Three-dimensional depicting method and system for delta diversion river sand | |
CN113759421A (en) | Method for researching earthquake structure morphological analysis based on aftershock positioning data |
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 |