[go: up one dir, main page]

EP1540506A1 - Construction d'un modele de reconnaissance de forme appliquee a l'exploration et a la production petrolieres - Google Patents

Construction d'un modele de reconnaissance de forme appliquee a l'exploration et a la production petrolieres

Info

Publication number
EP1540506A1
EP1540506A1 EP03764505A EP03764505A EP1540506A1 EP 1540506 A1 EP1540506 A1 EP 1540506A1 EP 03764505 A EP03764505 A EP 03764505A EP 03764505 A EP03764505 A EP 03764505A EP 1540506 A1 EP1540506 A1 EP 1540506A1
Authority
EP
European Patent Office
Prior art keywords
pattern
data
processing unit
central processing
level
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.)
Withdrawn
Application number
EP03764505A
Other languages
German (de)
English (en)
Other versions
EP1540506A4 (fr
Inventor
Robert Wentland
Peter Whitehead
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chroma Energy Inc
Original Assignee
Chroma Energy Inc
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Priority claimed from US10/308,860 external-priority patent/US7162463B1/en
Application filed by Chroma Energy Inc filed Critical Chroma Energy Inc
Publication of EP1540506A1 publication Critical patent/EP1540506A1/fr
Publication of EP1540506A4 publication Critical patent/EP1540506A4/fr
Withdrawn legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V11/00—Prospecting or detecting by methods combining techniques covered by two or more of main groups G01V1/00 - G01V9/00
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/28—Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/98—Detection or correction of errors, e.g. by rescanning the pattern or by human intervention; Evaluation of the quality of the acquired patterns
    • G06V10/987—Detection or correction of errors, e.g. by rescanning the pattern or by human intervention; Evaluation of the quality of the acquired patterns with the intervention of an operator

Definitions

  • the present invention relates generally to oil exploration and production.
  • the present invention relates to constructing pattern recognition templates in combination with geological, geophysical and engineering data processing, analysis and interpretation for hydrocarbon exploration, development, or reservoir management on digital computers.
  • a variety of mathematical manipulations of the data are performed by computer to form displays that are used by an interpreter, who interprets the data in view of facts and theories about the subsurface. The interpretations may lead to decisions for bidding on leases or drilling of wells.
  • ⁇ A commonly used measurement for studying the subsurface of the earth under large geographical areas is seismic signals (acoustic waves) that are introduced into the subsurface and reflected back to measurement stations on or near the surface of the earth.
  • seismic data can also be used to obtain detailed information regarding producing oil or gas reservoirs and to monitor changes in the reservoir caused by fluid movement.
  • Description of neural network modeling for seismic pattern recognition or seismic facies analysis in an oil reservoir is described, for example, in “Seismic-Pattern Recognition Applied to an Ultra Deep-Water Oilfield,” Journal of Petroleum Technology August, 2001 , page 41). Time-lapse seismic measurements for monitoring fluid movement in a reservoir are well known.
  • the fluid displacement may be caused by natural influx of reservoir fluid, such as displacement of oil by water or gas, or may be caused by injection of water, steam or other fluids. Pressure depletion of a reservoir may also cause changes in seismic wave propagation that can be detected. From these data, decisions on where to drill wells, production rates of different wells and other operational decisions may be made.
  • the neural network technique usually assumes that all significant combinations of rock type are known before analysis is started so that they can be used as a training set. This assumption is usually acceptable when analyzing fully developed fields but breaks down when only a few or no wells have been drilled. Common implementations of the neural network technique usually assume selection of the location of the geology of interest is an input that is determined prior to the analysis and often selects it using an analysis gate of fixed thickness.
  • geology of interest As the geology of interest is not always well known, the geology of interest should be a product of the analysis, not an input. Moreover, geology of interest rarely has a fixed thickness. The thickness varies significantly as the depositional process varies from place to place, sometimes by an amount that is sufficient to significantly degrade the result of the neural network analysis. This form of analysis includes information extraction and information classification in a single step that has little of no user control.
  • U.S. Patent No. 6,336,943 Bl discloses a neural network-based system for delineating spatially dependent objects in the subsurface from seismic data. The application of neural networks to seismic data interpretation has been widely investigated.
  • U.S. Patent No. 6,336,596 Bl discloses the use of a Voxel Coupling Matrix, which is developed using a finite number of neighboring voxels forming a textile. "Texture attributes" are developed. The attribute volumes are then used as inputs into an attribute- trace classification method to produce a seismic inte ⁇ retation volume. The inte ⁇ retation volume is displayed using distinct colors to represent various numbers of classes of reflection patterns present within the seismic volume. The aforementioned technique is an example of a class of image analysis called texture matching.
  • U.S. Patent No. 6,151,555 discloses a workstation computer system and an associated method and program storage device.
  • U.S. Patent No. 6,131 ,071 discloses a method for processing seismic data to provide improved quantification and visualization of subtle seismic thin bed tuning effects and other lateral rock discontinuities.
  • a reflection from a thin bed has a characteristic expression and the frequency domain that is indicative of the thickness of the bed.
  • the method may be applied to any collection of spatially related seismic traces.
  • Other methods of presentation of seismic data are disclosed in the patent and technical literature.
  • 3D seismic produces images of structures and features of the subsurface of the earth over very large geographical areas, it does not inte ⁇ ret those images.
  • a trained geoscientist or specialist performs the inte ⁇ retation.
  • reliance upon a relatively few qualified individuals increases the cost of the inte ⁇ retation process and limits the number of inte ⁇ retations that can be made within a given period.
  • a pattern recognition process would be repeated for large amounts of data in a screening process, with the results displayed in an intuitive manner so that the specialist can quickly perform quality control on the results, and correct noise induced errors, if any.
  • What is needed is a way to capture in a template and reuse the information used to perform a pattern analysis including the computation of the feature, pattern, and texture extraction, the decision surfaces required to classify the data, and the parameters required to visualize the results.
  • a data analysis specialist should not be required to rely on analysis of non-visual measures of object characteristics.
  • the information describing the visual characteristics of seismic data should be stored in a way that allows the data specialist to interact with the information to infer and extract geological information and to make a record of the exploration process. Templates with various uses and for various geological settings should be stored and retrieved for later reuse.
  • the present invention solves many of the shortcomings of the prior art by providing an apparatus, system, and method for identifying hyperdimensional fragments in a PDB that identify geological targets of interest, stored them as a part of a template, optimizing the templates, verifying the templates, and storing the templates of the known data in a template database ("PDB").
  • PDB template database
  • the present invention is applicable to a variety of applications where large amounts of information are generated. These applications include many forms of geophysical and geological data analysis including but not limited to 3D seismic.
  • the method of the present invention first builds a pattern database for a target data set with known targets and if possible known non-targets. It then builds a template for identifying targets using either a previously determined template or a set of default parameters. Next the method repeatedly tests the template, identifies cases where it fails, and refines the template until failures are minimized. The result is stored in a template database.
  • Templates are feature, pattern, and texture decision surfaces that are used by the associated classifiers to find like structures.
  • Known patterns that are captured in templates can then be compared, in an automated fashion, to new data to detect similar patterns and hence find the desired features in the new data.
  • the templates also contain all of the processing and display parameters required to start with an initial data set and create a final product in a batch data computer run without human intervention.
  • Figure 1 is a block diagram of the apparatus of the present invention.
  • Figure 2 is a flowchart illustrating an embodiment of a method of creating a template.
  • Figure 3 is a flowchart illustrating an embodiment of a method of identifying target and non-target hyperdimensional fragments when well data is available.
  • Figure 4 is a flowchart illustrating an embodiment of a method of identifying target and non-target hyperdimensional fragments in the absence of well data.
  • Figure 5a is a flowchart illustrating an embodiment of a method of modifying a template to reduce misclassifications or false positives.
  • Figure 5b is a flowchart illustrating an embodiment of a method of improving feature attribute and statistic selection and related decision surface selection.
  • Figure 5c is a flowchart illustrating an embodiment of a method of improving pattern attribute and statistic selection and related decision surface selection.
  • Figure 5d is a flowchart illustrating an embodiment of a method of improving texture attribute and statistic selection and related decision surface selection.
  • the present invention may be susceptible to various modifications and alternative forms. Specific embodiments of the present invention are shown by way of example in the drawings and are described herein in detail. It should be understood, however, that the description set forth herein of specific embodiments is not intended to limit the present invention to the particular forms disclosed. Rather, all modifications, alternatives, and equivalents falling within the spirit and scope of the invention as defined by the appended claims are intended to be covered.
  • the present invention includes a system for and method of extracting, organizing, and classifying features, patterns, and textures from a data set.
  • the data and the information extracted therefrom, is organized as a pattern hierarchy and stored in a pattern database.
  • the present invention also provides a system for the segmentation and the analysis of geological objects, for example, by identifying, extracting, and dissecting the best estimate of hydrocarbon filled reservoir rocks from band-limited acoustical impedance (“RAI”) data computed from 3D seismic data or, if available, broadband acoustical impedance (“Al”) computed from 3D seismic data, stacking velocities, well logs, and user supplied subsurface structural models.
  • RAI band-limited acoustical impedance
  • Al broadband acoustical impedance
  • the present invention includes a system for capturing the knowledge of the geoscientists operating the present invention in templates and reusing the templates for automated mining of large volumes of data for additional geological objects.
  • the following are definitions of terms that are used in the description of the present invention. Terms not defined herein retain their common usage.
  • An abstraction process is a process of successive identification of features, patterns, and textures within data and storing the results of each step into a layer within a. pattern database.
  • Auto-track An auto-track is the process of building objects by identifying spatially connected zones with a common attribute, collection of attributes, a common hyperdimensional fragment, or satisfying a template.
  • Attribute is a characteristic or measurement of data.
  • pu ⁇ oses of the present invention it is defined as the measurements used to characterize the data at a given level of the pattern pyramid.
  • attributes can characterize the cuts or the boundary representation of objects after segmentation. Examples include local curvature measurements on the exterior of a segmented geological reservoir in geoscience, or colon in medical studies. Examples are features, patterns, and textures.
  • Attribute Location The attribute location is the physical starting and ending location in the data set defining the location of an attribute.
  • Azimuth See Dip / Azimuth.
  • Binding strength is a threshold which is supplied by the operator that sets the degree of match required to determine if the known data matches the target data.
  • the binding strength allows the present invention to recognize patterns in data that contains defects or noise or where a perfect match is not required. It is implemented as the threshold of values for each level of a hyperdimensional fragment.
  • Classifiers are a computational method (algorithm) that sorts out data into various sets with common characteristics or classes. See template.
  • Classification is the process of applying classifiers or templates. For the pu ⁇ oses of the present invention it is the process of identifying hyperdimensional fragments in the pattern database of the target data set that match, within a given binding strength, the hyperdimensional fragments of a known data set. Cut / Cutting: Cut and cutting is the process of subdividing the data into a collection of one dimensional fragments.
  • the cutting criteria are a set of methods and related parameters for creating fragments. The specific method depends on the nature of the data and the objective of the analysis. The simplest example is to choose a fixed length of data samples. A more complex method is to perform cuts based on changes in the data values such as sign changes, identification of edges, and others. Either of the above examples result in cuts with varying spatial lengths. Variable length cuts give rise to a topological surface that has measurable characteristics or attributes. While cuts of variable spatial lengths are possible, cuts of uniform length are also possible.
  • Data mining is the process of applying templates from a template database to one or many target data sets creating output objects, in scenes, that satisfy the template.
  • the decision surface is a surface that separates two or more classes during classification. For the present invention it exists each of the feature lev e , pattern level, and texture level and is defined by a hyperdimensional fragment plus a non-zero binding strength that performs classification at each of the levels of abstraction.
  • a template is a concatenation of the decision surfaces for each level of abstraction.
  • a dip or azimuth is a measurement system used to measure the orientation of geological formations. It refers to the orientation of a plane that is tangent to the surface of a rock layer at a given map location. They are two angles defining the orientation of a vector that points in a direction that maximizes the vectors dip magnitude and lies on the tangent plane.
  • the dip angle is referenced to the horizon, which is a horizontal plane oriented so that its normal points toward the center of the earth at the location of interest. Dip is the angle the vector makes relative to the horizon. When the vector is projected onto the horizon plane, strike is the angle from the projected vector to a unit vector pointing north.
  • An earth manifold is, mathematically, a manifold is a topological space that is only locally Euclidean. It consists of a collection of local coordinate charts and a set of transformations between the charts. The transformations allow motion from one chart to the next.
  • the manifold is also required to be piecewise continuous. Intuitively it is a space that was distorted.
  • One example of a manifold is the globe manifold, which is the 3D space representing the surface of the planet earth. It has been mapped using a set of maps (locally Euclidean coordinate charts) and a convention (overlapping of the maps) for moving from the edge of one map to the corresponding edge of the next (set of transformations).
  • each fragment, or a related coordinate neighborhood has a coordinate chart defined by the local strike and dip of the rock layers (plus the normal to both) and rules for connecting them together along the rock layers.
  • pattern analysis is performed on a manifold.
  • Edge occurs where the data or attribute(s) of the data changes significantly at a given spatial location. In other words, an edge occurs at locations where the derivative of the attribute(s) has a peak. In reality, edges are often not sha ⁇ and are obscured by various noises in the data. Also, see segmentation and cutting.
  • a feature is the smallest inte ⁇ retable and classifiable measurement that can be made on the data. It is one member of a feature set. Also, see feature set, visual feature set, and feature space.
  • Feature Location T e feature location is the physical starting and ending location that defines the spatial position of a feature.
  • Feature Set feature set is a set of feature attributes that represent the state of the data or image at each feature location in the image. Different feature sets are selected to classify data in various ways. Also, see feature, visual feature set, and feature space.
  • Feature Space The feature space is, Mathematically and specifically topologically, a feature set is represented by a vector space (state space) where each axes of the space is a feature attribute.
  • the smallest feature attribute set which is also the computationally most efficient set, is described mathematically as a set, or basis of the vector space, where each member of the set is linearly independent. Linearly independent means that one axes, or set member, is not a linear combination of the other set members.
  • Each feature attribute represents a degree of freedom of the image. When all of the degrees of freedom are represented as feature attributes, then the feature set, which is also the basis of the vector space, are described as spanning.
  • Intuitively spanning means that a feature set can be recombined to exactly recreate the data from which it was measured.
  • a feature set capable of spanning the feature space is defined, but to reduce computation time only the features required to solve the problem are computed.
  • Fragment A fragment is a one-dimensional interval that has a physical size and spatial location. It is the smallest interval at the given spatial location within which attributes can be measured. Fragments may be defined as having a fixed physical length or include a fixed number of attribute values. They may also have a variable physical length where the fragment is cut using cutting criteria that are a function of the data, attributes, or statistics of any lower level in the pattern pyramid. Variable length fragments usually lay between data value, attribute, or statistic edges. Fragments are cut at each of the feature, pattern, or texture levels of the pattern pyramid. Also, see fragment sequence.
  • Fragment Orientation When the data being analyzed has more than one dimension a fragment orientation needs to chosen while cutting the data into fragments. For example 3D seismic is often measured in a three dimensional space with three axes, line, xline, and time. Other 3D data is often measured in three-dimensional space in three axes, x, y, and z. Thus, the fragment orientation can be aligned along any of these three axes. Another option is to align the fragment orientation along the earth manifold in the dip, strike, or normal (to the rock layers) direction, or in an alternate coordinate system, such as a geology or tissue aligned manifold.
  • a fragment sequence is the sequence of data, attribute, or statistic values that occur within a fragment.
  • Global Statistic A global statistic is a statistical comparison of the value of the attribute of interest at a particular location in a data set to the value of this attribute at all locations of the data set. Also, see statistic and local statistic.
  • Hyperdimensional Fragment is a fragment that extends vertically through the various levels of a pattern pyramid. It represents an ordered collection of attributes containing members from each level. At each level, the hyperdimensional fragment contains a single point and the associated attribute value. If required, a binding strength, which is a set of thresholds representing a range of values about the attributes at each level above the data level, can be used. It is defined for both known data and target data and is used for the simultaneous classification of features, patterns, and textures. See template and classification. Mathematically, topologically, it is a fiber view of the tangent spaces represented by the levels of the pattern pyramid. In terms of pattern recognition, it represents a classifier derived from the known data that is used to perform classification of the target data.
  • Known Data are a specified portion of a geophysical data set containing either a geological analog that is a known hydrocarbon deposit or an example composed of a subset of the geophysical data that is identified by a geoscientist as a potential hydrocarbon deposit. It is chosen to include primarily the item of interest and little else.
  • Local Statistic is a statistical comparison of the value of the attribute of interest at a particular location to the value of this attribute in a local coordinate neighborhood. The size of the path is operator selected. Also, see statistic and global statistic.
  • An object is a spatially connected body within a scene that has a common attribute, hyperdimensional fragment (collection of attributes), or fits a template.
  • Software objects are not to be confused with an object of the present invention. Software objects retain their standard meaning.
  • An object space is a manifold space that represents either the exterior boundary of an object or the interior of an object.
  • An example of a 3D manifold is the surface on the outside of an object.
  • Another example is the 3D manifold representing rock layers within the earth.
  • An object attributes are the measurable properties that is associated with an object 's boundary representation or outside edge.
  • An example of an object attribute is the local curvature of the exterior surface of an object.
  • Pattern A pattern is a naturally occurring repetition of feature attributes in a fragment sequence. Also, see Pattern Space.
  • Pattern Database is a database that consists of several levels of attributes (usually features, patterns, and textures) within a data set. It can be a relational database containing the attributes. It can also be an object database containing the attributes as parameters and the reduction process computations as methods plus other parameters and methods if required for data handling and/or display. Pattern Location: The pattern location is the physical starting and ending location that defines the spatial position of a pattern.
  • Pattern Pyramid is a diagram that represents the pattern database.
  • the pyramid sits on a broad rectangular base representing the data.
  • a triangle sets on the base that decreases in width upward.
  • the triangle has 3 levels consisting from bottom to top of features, patterns, and textures.
  • the width at each level represents the number of fragments at that level, which decreases as the level of abstraction increases.
  • the levels of the pattern pyramid represent tangent spaces of the data set.
  • Pattern recognition is the analysis of data for making a decision. It involves making measurements on the data (pattern attributes and data statistics), analyzing the measurements (classification) and making a decision (computing a decision surface).
  • Pattern Space Physical space is determined by a transformation from feature space to pattern space.
  • Pattern space is an abstract vector space (state space) where each axes of the space represents a degree of freedom of the patterns in the image.
  • Each location in the fragment sequence (N F ) and each member of the associated feature set represent the degrees of freedom (N M )-
  • the pattern space has a dimension of D
  • a fragment sequence of length 3 with only 1 measured feature is a 3D space.
  • Physical Space Physical space is the space within which the data is measured or sampled. It is usually a Euclidean space.
  • attribute calculations are performed on the physical space.
  • the axes of the physical space are inline, xline, and time.
  • the time axes refers to the two way travel time of sound through the earth. Sometimes time is replaced by subsurface depth.
  • a scene is a spatial region that is viewable and contains one or more objects. It is analogous to a room that is being viewed by an observer where the room contains furniture objects and people objects, all of which can be viewed.
  • a scene is a collection of objects. It is implemented as an integer data cube that contains integer numbers that represent object numbers plus a null value. Data points containing the null value represent null zones, which are not assigned to objects. Zones containing object numbers are assigned to the objects associated with the specific numbers.
  • Statistic is a method of analyzing the attributes of each layer in the PDB for the pu ⁇ ose of analyzing the probability of a selected occurrence of the attribute occurring elsewhere in the entire data set or in a selected local coordinate neighborhood. Also, see local statistic, and global statistic. Statisizing is the process of applying the statistic method.
  • Target data is a geophysical data set that is to be analyzed. Examples include seismic, electrical, magnetic, optical, or other form of measurements that measure rock properties. Target data is usually in the form of a 3D voxel cube although higher or lower spatial dimensions or analog data may also be used. The data that is most frequently used is 3D seismic data. To allow analysis of the rock layers rather than their interfaces, band-limited acoustical impedance is usually used.
  • a template is a sorting device for selecting a subset of a data collection where each member of the subset matches the template. It is implemented in software as a collection of one or more decision surfaces.
  • a template contains a hyperdimensional fragment representing the pertinent contents of the pattern database of a known data plus the binding strength. It is applied to the hyperdimensional fragments of the PDB of a target data set to identify targets.
  • a template can contain decision surfaces that are related to object attributes that are properties of the objects boundary representation. An example is selecting objects for which the local curvatures lie within a given range of values as defined by decision surfaces in the template.
  • Template database is a database that contains one or more templates. It can be stored as data files, in a relational database, or in an object database.
  • Texture Intuitively, a texture is the visual characteristic of a cloth that is composed of closely interwoven threads where the threads have unique patterns. It is a measurement of the order of repeated patterns in a series of spatially adjacent fragment sequences. Also, see Texture Space.
  • Visual feature set is a feature set designed to classify data based on its visual appearance.
  • An instance of a visual feature set is the one a shipping company uses to classify packages.
  • the feature set is measurements of boxes including length, width, height, and shape (square, rectangular, etc.).
  • An instance of a visual feature set that is used by seismic stratigraphers and inte ⁇ reters to analyze band-limited acoustical impedance (inverted seismic) includes distance between zero crossings, maximum amplitude, and shape (skewed upward, skewed downward, single peak, multiple peaks, shoulder). Also, see features.
  • Voxel cube A voxel cube is a 3D regular, ordered, quadmesh representing discreet measurements in a Euclidean space. A voxel cubes are also referred to as 3D pixels.
  • the first step of the pattern recognition method of the present invention is feature extraction.
  • Feature extraction comes in many forms, and tends to be specific to the type of problem encountered. For example, in seismic data analysis, geological features are extracted. Most traditional methods of feature extraction for seismic data involve mathematical algorithms that focus on the measurements of the sound rather than on the visual appearance of the displayed data. Most geophysicists, however, think of geology in a visual way, which makes analysis and inte ⁇ retation of traditionally extracted seismic signal features difficult. Many other examples and uses for the feature extraction and imaging technique of the present invention will be apparent upon examination of this specification.
  • a mathematical representation of features describes the local state of a system.
  • the features are then represented as a vector in an abstract vector space or tangent space called the feature state space.
  • the axes of the state space are the degrees of freedom of the system, in this case the features of the image.
  • the features, axes of the state space be linearly independent.
  • the features have the capacity to "span the signal," or to describe all seismic attributes such that, for example, a geophysicist could accurately re-create the underlying geology.
  • geological features are extracted for performing pattern recognition on a seismic data set.
  • Feature descriptors of seismic data tend to be one-dimensional, measuring only one aspect of the image, such as measuring only properties of the signal at specific locations in the signal. These feature descriptors taken singly do not yield enough information to adequately track geology.
  • the relationship these measurements have with their local neighbors contains information about depositional sequences which is also very important geological information.
  • the relationship features have with their neighbors and the total data set also needed to be analyzed.
  • the present invention utilizes a hierarchical data structure called a pattern pyramid that is stored in a pattern database (PDB).
  • PDB pattern database
  • the pattern database employs a process that is based on DNA-like pseudo sequencing to process data and places the information into a pattern database.
  • This database contains the data plus features and their relationship with the data and, in addition, information on how the features relate with their neighbors and the entire data set in the form of pattern, textures, and related statistics.
  • the basic concept of the pattern pyramid is that complex systems can be created from simple small building blocks that are combined with a simple set of rules.
  • the building blocks and rules exhibit polymo ⁇ hism in that their specific nature varies depending on their location or situation, in this case the data being analyzed and the objective of the analysis.
  • the basic building block used by the present invention is a fragment sequence built from a one-dimensional string of data samples.
  • a pattern pyramid is built using fragment sequences (simple building blocks) and an abstraction process (simple rules).
  • the specific definition of the building blocks, cutting criteria exhibits polymo ⁇ hism in that the algorithm varies depending on the data being analyzed and the goal of the analysis.
  • the abstraction process exhibits polymo ⁇ hism in that the algorithm depends on the data being analyzed and the goal of the analysis.
  • a pattern database is built for known data, which functions as a reference center for estimating the locations in the target data that are potential hydrocarbon deposits.
  • the estimation is accomplished by building a pattern database for the target data using the same computations as for the known data and comparing the pattern databases.
  • the pattern pyramids have several levels of abstraction which may include features, patterns, and textures.
  • the present invention not only performs pattern recognition, but is also capable of performing feature recognition, texture recognition, and data comparison all at the same time as required for solving the problem.
  • the hyperdimensional fragment selected by the geoscientist operating the present invention captures the operators' knowledge of what is a desirable outcome, or in other words what a hydrocarbon filled reservoir looks like.
  • the hyperdimensional fragments and associate abstraction process parameters can be saved as a template into a template database.
  • One or more templates can be checked out from the library and applied to large volumes of target data to identify targets.
  • Targets which have been segmented out of the data set are stored as objects in a collection of objects called a scene. The objects, along with additional data the geoscientist adds to them, become a list of drilling opportunities.
  • the lead After a full analysis of the reservoir, charge, and trap plus risk analysis, economic analysis, and drilling location selection the lead becomes a prospect that is ready to be drilled. The probability of success is highest when there is strong evidence that a reservoir, charge, and trap all exist, that they exist in the same drillable location, and that they can be profitable exploited.
  • Our objective is to construct a pattern recognition process and associated tools that identify a location with all of the constituent parts of a lead and to quantify them to covert a lead into a prospect.
  • the present invention identifies a potential reservoir through feature analysis, identifies hydrocarbon indications through pattern and texture analysis, and identifies the presence of a depositional process that deposits reservoir rock though texture analysis.
  • the final step of associating and validating the three components of an oil field is usually accomplished by a geoscientist. After a lead has been identified the pattern database, along with appropriate visualization, could be used to perform reservoir dissection. This is a study of the internal characteristics of the reservoir to estimate the economics and convert the lead into a prospect.
  • the present invention is capable of being used to improve reservoir characterization, which is the estimation of rock properties (rock type, porosity, permeability, etc.), and fluid properties (fluid type, fluid saturations, etc.).
  • Rock types and properties are a function of the geologic process that deposited them.
  • the local features, patterns and textures contain information about depositional processes.
  • the rock type and property estimations can be improved by including the feature, pattern, and texture information while estimating them.
  • the present invention could be used for portions of data processing. Examples include but are not limited to, automatic stacking velocity picking, automatic migration velocity picking, noise identification, and noise muting.
  • the present invention is also capable of performing data set registration and comparison by successively aligning textures, patterns, and features. When applied to seismic it includes registering shear data to compressional data, registering 4D seismic data, registering angle stacks for AVA analysis, and others. 3D Seismic First Pass Lead Identification
  • This example performs first pass lead identification through simultaneous identification of a potential reservoir through feature analysis, identification of hydrocarbon indications through pattern and texture analysis, and identification of the presence of a depositional process, that deposits reservoir rock, though texture analysis.
  • One way to do this is to use a known data set, which represents a successful lead or example lead, and compare the target data to known data.
  • the goal is to identify reservoirs that occur in all forms of traps.
  • the overall process starts by building a pattern database with successive levels of abstraction (features, patterns, and textures) for the known data.
  • the pattern database building process has been applied to a set of known data, and the minimum set of attributes that characterize the known data has been identified, the pattern database is applied to a set of data to be analyzed (the "target data").
  • the data of each set are subjected to the same series of steps within the abstraction process.
  • an affinity or binding strength is selected by the operator which determines how closely the known data has to match the target data to result in a target being identified.
  • the binding strength helps to identify features, patterns, and textures in the target data that adequately match, but do not exactly match, the desired features, patterns, and textures in the known data.
  • the pattern database for the known data is compared to that of the target data. This is performed by identifying a hyperdimensional fragment from the known data pattern database that adequately and reasonably uniquely characterizes the known data.
  • This hyperdimensional fragment relates the data at the location where the hydrocarbons were found or are expected to be found to the set of features, patterns, and textures which were derived from it.
  • the hyperdimensional fragment and associated abstraction process parameters can be combined into a template. Templates can be used immediately or stored in a template database on one or more mass storage devices, and then retrieved when needed. When templates are applied to target data sets the resulting targets are identified. These targets are stored as objects which represent leads.
  • the leads objects are the locations in the target data sets which have a potential reservoir identified through feature analysis, potential hydrocarbon indications identified through pattern and texture analysis, and the potential presence of a depositional process that deposits reservoir rock identified though texture analysis.
  • a collection of objects are stored in a scene.
  • the scene represents the physical locations of the leads identified by the present invention in this example. Geological and other required properties of the leads can be stored with them.
  • a collection of templates can be created and stored in a template database. These may be sequentially applied to one or many target data sets in a process called data mining. When multiple templates are applied to the same target data set, the results are several scenes each containing lead objects. The scenes and their associated objects, one scene from each template, can be combined by performing Boolean operations on the scenes containing the objects creating one composite scene.
  • a 3D seismic data set exemplary embodiment of the layers of abstraction associated with the method of the present invention is illustrated in Figure la.
  • Each level of the pattern pyramid represents a level of abstraction.
  • the input data lie at the bottom of the pyramid 100.
  • the width at the base of each layer is generally indicative of the number of data samples or fragments involved within that stage of the method of the present invention.
  • a fragment sequence is a one dimensional, ordered, spatially sequential, set of data values that cover multiple data samples and becomes larger with each higher level of abstraction.
  • the pattern pyramid 100 contains three layers of abstraction above the data level 108 (see Figure la).
  • the abstraction process is first applied to the data level to generate the e ⁇ twre level 106. Thereafter, the abstraction process is applied (at least once) to the feature layer data to generate the/? ⁇ tter « level 104. Next, the abstraction process is applied (at least once) to the pattern layer data to generate the texture level 103. While the exemplary embodiment illustrated in Figure la has three layers of abstraction above the data level 108, only one layer is required. On the other hand, should the analysis call for it, any number of layers may be generated above the data level 108. How many layers are generated, or how they are generated is problem-specific.
  • the pattern pyramid shown in Figure la corresponds to a single fragment orientation during analysis.
  • Some data sets with more than one spatial dimension may require analysis in more than one fragment orientation to achieve the desired results.
  • Seismic data has a strong preferred direction caused by the geology of the subsurface of the earth.
  • Another example of data with a preferred direction is wood grain.
  • the analysis can give very different results depending on the fragment orientation relative to the preferred direction of the data. Successful analysis of this data might require using fragments with more than on alignment.
  • sides can be added to the pyramid as shown in Figure lb. Each side is associated with a fragment alignment direction.
  • the example in Figure lb shows three views (oblique 112, top 114, and side 1 16) of a 3D pattern pyramid.
  • the example shows a pattern pyramid for 3D seismic data which has 3 spatial dimensions consisting of the inline axes, xline axes, and time axes. Each direction has an associated side on the pattern pyramid, an inline side 118, an xline side 1 19, and a time side 117. Because geology does not always align itself with the coordinate system on which the data is collected, this orientation will result in a pattern recognition analysis where the largest effect is the structure of the earth. When analyzing the trap component of an oil field this is very useful. If the goal is to not want to analyze geological structure and instead analyze the earth's stratigraphy, a different coordinate system is needed. To accomplish that goal, the fragments need to be aligned with the earth manifold, along dip, strike, and normal to the layers.
  • the pattern database building process identifies the minimum set of attributes (features, patterns, and textures) of one or several examples of known data so that, when the known data is compared to the target data, only the desired characteristics need to be considered.
  • the results of each step are represented in the pattern pyramid 120 as shown in Figure lc and are stored in the pattern database.
  • the process starts at the data layer which for seismic data can contain a lower layer of pre-stack seismic data 142 setting under a layer of post-stack seismic data 140.
  • the pattern database contains several layers of abstraction that are built sequentially starting at features, proceeding through patterns, and finally ending with textures, the highest level of abstraction. There may be one or several layers of each type. Not all of the layers are required.
  • the pattern database can be built only up to the pattern pyramid level required to solve the problem.
  • the creation of each layer includes one or more steps of cutting, computing attributes, and computing statistics.
  • Each layer has a cut 138, 132, and 126, computed attributes 136, 130, and 124, plus computed statistics 135, 128, and 122.
  • the precise methods of cutting, computing attributes, and computing statistics changes from layer to layer, and can change within the layers. They specific computations in the abstraction process are designed to capture the minimum set of feature level attributes 136, feature level statistics 135, pattern level attributes 130, pattern level statistics 128, texture level attributes 124, and texture level statistics 122 required to solve the problem.
  • Figure lc illustrates how a particular point of space in the input data 140 and 142, represented by the point 156, has corresponding points 154 and 152 in the feature layer, 150 and 148 in the pattern layer, plus 146 and 144 in the texture layer.
  • the ordered set of points 156, 154, 152, 150, 148, 146, and 144 forms a trajectory called a hyperdimensional fragment of the data point 156 in question.
  • the pattern pyramid has a set of hyperdimensional fragments that associate each data sample to the features, patterns, and textures to which it contributed. Because the type of abstraction analysis is problem specific, so too is the resultant hyperdimensional fragment.
  • the binding strength or affinity When comparing the known data hyperdimensional fragment to the collection of target data hyperdimensional fragments the amount of similarity required to consider them matched is determined by the binding strength or affinity.
  • This invention implements the concept of a binding strength by setting a range of acceptable feature, pattern, and texture values at each pattern pyramid level that the hyperdimensional fragment passes through. The result is that exact matches are no longer required but similar matches are allowed.
  • the hyperdimensional fragment and associated threshold becomes a template that is used for object identification. Making a comparison between the known data and the target data is accomplished by applying the template to the target data.
  • the comparison is accomplished by searching through all of the hyperdimensional fragments in the target data set and determining if the feature, pattern, and texture values though which they pass are the same within the binding strength as the values in the known data hyperdimensional fragment.
  • Templates can be stored in a template database and retrieved for later use on any target data set.
  • the result of applying a template to a target data set pattern database is a scene that contains null values where matches did not occur and a value representing matched where matches did occur.
  • the next step is to identify all data connected points where matches occurred and assign them to an object. This is accomplished by stepping through all of the points that are marked as matched and performing an auto-track that assigns all connected points that are marked as matched to an object. This is repeated until all points that are marked as matched have been assigned to connected objects.
  • the result is a scene containing connected objects that represent potential hydrocarbon deposits. These objects represent a simultaneous analysis of how well they represent a potential reservoir through feature analysis, represent hydrocarbon indications through pattern and texture analysis, and include the presence of a depositional process that deposits reservoir rock though texture analysis.
  • Objects can have associated properties.
  • a 3D manifold also referred to as a shrink-wrap
  • Topological properties of the object surface such as local curvature, can be measured and stored as an object property.
  • Templates can be precomputed from known data sets, stored in a template database, and used the pattern databases for one or many target data sets creating resultant scenes containing objects which satisfy the templates. This process is often referred to as data mining. The collection of objects becomes a lead inventory.
  • the PDB comparison is performed by comparing hyperdimensional fragments.
  • the binding strength is specified for each level of the pattern pyramid where it was not already specified during pattern database construction usually by using the quick clustering technique above.
  • this step is performed for the first time it is often performed interactively during visualization of a target data set and the related pattern database.
  • the template is applied to the target data set. This step is often referred to as applying a scene construction tool. After this is accomplished the spatially connected objects are computed using another tool which is also referred to as a scene tool.
  • the template computed above is saved in the template database.
  • the appropriate templates are checked out and applied to all of the data in the geographical region being analyzed.
  • the resulting scenes and associated templates are combined using Boolean operations which are usually referred to as Boolean scene tools.
  • the final product is a lead inventory which is associated with a scene containing a list of multiple leads (objects) and lead parameters.
  • the lead parameters include lead names, locations, spatial sizes, global statistics, local statistics, and other useful information as required by the operator.
  • the template needs to contain associated information which allows the PDB to be created. This includes the definition of the PDB contents including all parameters and information required to create the appropriate PDB on which the template is applied.
  • the template needs to contain associated information which allows the results of applying the template plus the associated PDB to be visualized and analyzed for quality control. Examples include color tables, opacity tables, and other display parameters that may be used to display each of the levels of the PDB. In addition it might contain false color image combinations of information from several layers of the PDB. Other display information can be included.
  • the present invention is preferably implemented as a set of one or more software processes on a digital computer system.
  • the present invention may also be implemented purely in hardware, or may be virtually any combination of hardware and software.
  • the following description of the apparatus represents the best mode for practicing the present invention, and the elimination of one or more of the components described herein may incur a corresponding reduction in the effectiveness or the capability of the present invention.
  • future computer components may offer increased performance, or integration of tasks leading to enhanced performance of the present invention with the same or fewer numbers of components.
  • FIG. 16 illustrates an enhanced personal computer (“PC") 160 used for extracting features from a signal.
  • Enhanced PC 160 includes a main unit 162, a high- resolution display 180, a VGA cable 182, an optional CD-ROM drive 172, an optional 8mm (or other type) tape drive 174, a mouse 176, and a keyboard 178.
  • Main unit 162 further includes one or more CPUs 164, a high-speed memory 166, a network card 168, a high-speed graphics card 170, and an internal and/or external hard drive 180.
  • the hard drive 180 can be replaced with any suitable mass storage device, such as a storage area network (“SAN”), RAM, tape, drum, bubble or any other mass storage media.
  • SAN storage area network
  • Hard drive 180 stores, for example, a seismic and SEG-Y format database 184, a pattern database (“PDB”) 186 (also called a knowledge hierarchy), well data, culture data, other supporting adapt and documents, one or more applications 188, and a template library 190.
  • High-speed memory 166 is used to accelerate processing.
  • High-speed graphics card 170 is preferably an ultrahigh-speed graphics card like the Intense 3D Wildcat (manufactured by 3DLabs of Huntsville, Alabama).
  • High-resolution display 180 is the highest resolution display currently available in order to support the applications, which are intensely graphic in nature, and is electrically connected to main unit 162 by VGA cable 182. Also electrically connected to main unit 162 are CD-ROM drive 172, 8mm tape drive 174, mouse 176, and keyboard 178.
  • seismic data enters the enhanced PC 160 via, for example, the 8mm tape drive 174, the CD-ROM drive 172 and/or the network card 168.
  • This seismic data is stored in, for example, SEG-Y format in database 184 and is processed by CPU 164 using applications 188, with mouse 176, and keyboard 178 as input devices and high-speed memory 164 to facilitate processing.
  • the processed seismic data is then stored in a PDB 188 format.
  • a PDB contains a collection of data volumes.
  • the collection includes a 3D seismic data volume, multiple associated pattern, feature, texture volumes, and multiple scene volumes.
  • the data values are stored so that they can be addressed as spatially vertical columns or horizontal slabs with the columns and slabs made up of subsets called bricks.
  • a stack of bricks that extend from the top of the cube to the bottom is a column.
  • a mosaic of bricks that extends horizontally across the volume is a slab.
  • the brick size is chosen to optimize data access, for example, 64 by 64 samples in size.
  • the samples are 8-bit integer, 32-bit floating point, or any other desired format.
  • Each volume contains metadata including:
  • sample value scaling properties additive minimum and maximum values of scaled sample values
  • the PDB collection, and associated metadata can be stored as files on a file system, as information in a database, or as a combination of the two.
  • a seismic template is created for each geoscientist, and this template is stored in template library 190.
  • the seismic data is viewed on the high-resolution display 180.
  • the seismic data is stored in template library 190, and output to 8mm tape drive 174 or CD-ROM 172, or transmitted via the network card 168.
  • the methods are executed using object-oriented programming, which allows reflection coefficient ("RFC”) data, acoustic impedance (“Al”), and other calculated feature extraction information to be stored either as parameters and methods or as results, according to the available memory and processing capability of the host system. If the full results of the seismic data analysis are stored in the PDB 188, the memory requirement is measured in terabytes, which is more memory capacity than many systems have. If the parameters and methods for generating the seismic analysis are stored instead, the system must have enormous processing capability and high-speed memory 166 in order to rapidly calculate the analyzed seismic data when the seismic object is executed.
  • RRC reflection coefficient
  • Al acoustic impedance
  • the present invention employs the above-identified apparatus for various pu ⁇ oses.
  • the method of the present invention will now be illustrated via a method of creating a template of the present invention as illustrated in Figure 2 by method 200.
  • the operator might perform only portions of this method or variations of the method as is appropriate to the problem being solved and the operator's individual working style. This is a trial-and-error method which starts at a best guess starting point and modifies the pattern recognition database contents and parameters until a solution is reached
  • Step 205 Method 200 starts. The method starts generally at step 205.
  • Step 210 Select a starting template or use defaults.
  • the operator creates a starting point for construction of a pattern database.
  • the starting point will be refined to create a PDB that is adapted to solving the specific problem at hand.
  • the starting point could be provided by a pre-existing template taken from the template library. This is done by selecting a template that was built to work in a similar geological setting. If a previously built template does not exist the staring point could be the set of application defaults. The rate at which this trial-and-error method converges on a solution depends on how close this staring point is to the best solution.
  • Step 215 Create a Pattern Database.
  • the operator uses the a pattern recognition technique to create the pattern database.
  • Step 220 Have wells been drilled? In this decision step, the system operator determines if wells have been drilled within the area covered by the seismic data. If wells have been drilled, then the method 200 proceeds to step 225; otherwise, the method 200 proceeds to step 230. Step 225: Create lists of target and non-target Hyperdimensional Fragments
  • Step 230 Create lists of target and non-target HDF from identified leads using method 400 of Figure 4.
  • the operator performs method 400 of Figure 4.
  • Step 235 Compute binding strength.
  • the operator analyzes the list of target and non-target hyperdimensional fragments ("HDF") to compute the binding strength.
  • Binding strengths can be defined in many ways with the specific definition depending on the problem being solved. An example of a binding strength is to compute a range around the HDF values for each level of abstraction in the pattern pyramid within which any other HDFs are considered to be a match. A common way to do this is to use either the local statistic or global statistic to determine the mean as the HDF and some percentage of the standard deviation to set the binding strength.
  • Step 240 Create a template.
  • the operator creates a template.
  • Step 245 Apply template.
  • the operator uses the pattern recognition method, such as the one applied in step 215. If the pattern database contents as described in the template is the same as the current pattern database, then the pattern recognition method is skipped.
  • Step 250 Visualize and verify the leads. In this step, the operator uses the
  • Chroma Vision application to visualize the scene created by step 245.
  • the application is instructed to use the default visualization parameters from the template. If the operator wishes, other pattern database information can be viewed with the scene displayed as an overlay using the default false color images and visualization parameters from the template. The operator is optionally able to modify the display of the other pattern database information to create custom displays as required to verify the leads.
  • Step 255 False positives?
  • the system operator determines if the scene contains objects which are false positives. If this step the operator uses his previous experience and knowledge of geological processes to identify objects which have the same seismic response as targets but are in fact not a desired target. If false positives are identified, then the method 200 proceeds to step 260; otherwise, the method 200 switches to step 275.
  • step 265, and 270 are causing a reduction in the number of false positives and if not are further reductions possible. If the analysis loop is still converging on an optimal solution, then the method 200 proceeds to step 265; otherwise the best template has been found and the method 200 switches to step 275.
  • Step 265 Add false positives to list of non-targets.
  • the operator adds the false positives to the list of non-targets.
  • Step 270 Update template using method 500 of Figure 5.
  • the operator updates the template.
  • the operator performs method 500 of Figure 5.
  • method 200 switches to step 245.
  • Step 275 Add template to the template library.
  • the operator executes a software application which places the template by this method into a template library.
  • Step 295 Method 200 ends. The method 200 ends generally at step 295.
  • the present invention employs the above-identified apparatus for various pu ⁇ oses.
  • the method of the present invention will now be illustrated via the identifying target and non-target hyperdimensional fragments when well data is available method of the present invention as illustrated in Figure 3 by method 300. This method is usually performed as step 225 of method 200.
  • Step 305 Method 300 starts. The method starts generally at step 305.
  • Step 310 Initialization.
  • the operator or an application the operator is executing prepares a list of wells which penetrate the earth within the boundaries of the 3D seismic being analyzed.
  • a pointer or selector for the list is initialized so that the first time step 320 is executed the first well in the list is selected.
  • Step 315 Next well.
  • the operator or the application proceeds to the next well in the list which is the first well when this step is performed for the first time
  • Step 320 Obtain well data.
  • the operator obtains the well logs and related data. If the well data has not been previously processed and analyzed the processing and analysis is performed by a petrophysicist.
  • Step 325 Identify location(s) of reservoirs in the well(s).
  • the operator prepares a list of reservoir locations in the well.
  • the entries in the list are obtained from the petrophysical analysis results, from recorded perforation zones, or by analyzing the well data.
  • Step 330 Initialization.
  • the operator or an application the operator is executing sets a pointer or selector for the reservoir list to initialize it so that the first time step 335 is executed the first reservoir in the list is selected.
  • Step 335 Next reservoir. In this step, the operator or the application proceeds to the next reservoir in the well. It is the first reservoir when this step is performed for the first time with a new well.
  • Step 340 Identify fragment location.
  • the operator or the application the operator is executing identifies the feature fragment location in the PDB which is in the same spatial location as the reservoir in the well.
  • Step 345 hydrocarbons?
  • the system operator determines if the well encountered hydrocarbons at the reservoir being analyzed. If the reservoir contains hydrocarbons, then the method 300 proceeds to step 350; otherwise the method 300 switches to step 360.
  • Step 350 Identify hyperdimensional fragment for target.
  • the operator or the application the operator is executing identifies the hyperdimensional fragment for the target.
  • the hyperdimensional fragment consists of the data value(s) plus all associated feature value(s), pattern values(s), texture value(s) and associated statistics for all levels in the PDB.
  • Step 355 Add to the target hyperdimensional fragment collection.
  • the operator or the application the operator is executing adds the hyperdimensional fragment to the collection of target hyperdimensional fragments and then the method 300 proceeds to step 370.
  • Step 360 Identify hyperdimensional fragment for non-target.
  • the operator or the application the operator is executing identifies the hyperdimensional fragment for the non-target.
  • the hyperdimensional fragment consists of the data value(s) plus all associated feature value(s), pattern values(s), texture value(s) and associated statistics for all levels in the PDB.
  • Step 365 Add to the non-target hyperdimensional fragment collection.
  • the operator or the application the operator is executing adds the hyperdimensional fragment to the collection of non-target hyperdimensional fragments.
  • Step 345 More reservoirs?
  • the system operator or the application the system operator is executing determines if there are more reservoirs in the well being analyzed. If there are more reservoirs, then the method 300 returns to step 335; otherwise the method 300 continues to step 375.
  • Step 375 More wells?
  • the system operator or the application the system operator is executing determines if there are more wells in the well list. If there are more wells, then the method 300 returns to step 315; otherwise the method 300 continues to step 395.
  • Step 395 Method 300 ends. The method 300 ends generally at step 395.
  • the present invention employs the above-identified apparatus for various pu ⁇ oses.
  • the method of the present invention will now be illustrated via the identifying target and non-target hyperdimensional fragments in the absence of well data method of the present invention as illustrated in Figure 4 by method 400. This method is usually performed as step 230 of method 200.
  • Step 405 Method 400 starts.
  • the method starts generally at step 400.
  • the operator or an application the operator is executing prepares a list of examples which may be obtained from previously analyzed data set(s) or visually identified by the operator with the decision based on previous experience.
  • the examples consist of geology which might contain hydrocarbons and geology which is unlikely to contain hydrocarbons.
  • a pointer or selector for the list is initialized so that the first time step 415 is executed the first example in the list is selected.
  • Step 420 lead?
  • the system operator determines if the example is an example of a lead which could contain hydrocarbons or is an example of data which is not a lead or is unlikely to contain hydrocarbons. If the example is a lead, then the method 400 switches to step 440; otherwise the method 400 proceeds to step 425.
  • Step 425 Identify fragment location.
  • the operator or the application being executed by the operator identifies the feature fragment location(s) at the same spatial location as the example which does not qualify as a lead.
  • Step 430 Identify hyperdimensional fragment for non-target.
  • the operator or the application the operator is executing identifies the hyperdimensional fragment for the non-target.
  • the hyperdimensional fragment consists of the data value(s) plus all associated feature value(s), pattern values(s), texture value(s) and associated statistics for all levels in the PDB.
  • Step 435 Add to the non-target hyperdimensional fragment collection. In this step, the operator or the application the operator is executing adds the hyperdimensional fragment to the collection of non-target hyperdimensional fragments and method 400 switches to step 450.
  • Step 440 Identify fragment location.
  • the operator or the application being executed by the operator identifies the feature fragment location(s) at the same spatial location as the example which qualifies as a lead.
  • Step 445 Identify hyperdimensional fragment for target.
  • the operator or the application the operator is executing identifies the hyperdimensional fragment for the target.
  • the hyperdimensional fragment consists of the data value(s) plus all associated feature value(s), pattern values(s), texture value(s) and associated statistics for all levels in the PDB.
  • Step 450 Add to the target hyperdimensional fragment collection.
  • the operator or the application the operator is executing adds the hyperdimensional fragment to the collection of target hyperdimensional fragments.
  • Step 455 More examples?
  • the system operator or the application the system operator is executing determines if there are more examples in the example list. If there are more examples, then the method 400 returns to step 415; otherwise the method 400 continues to step 495.
  • Step 495 Method 400 ends. The method 400 ends generally at step 495.
  • Method 500 starts. The method starts generally at step 502.
  • Step 504 Identify misclassifications.
  • the operator executes a visualization application to visualize and analyze the leads and underlying data identified by the template.
  • Misclassifications may include leads which should not have been identified as leads, or false positives, and locations in the data which should have been classified as a lead but was not, with the decision based on the geoscientist's knowledge and previous experience.
  • the misclassifications are placed in a list and a pointer is initialized so that the next time step 505 is executed the first misclassification in the list is selected.
  • Step 505 Next misclassifications.
  • the operator or a computer application being executed by the operator increments the pointer to the current selection in the misclassification list to the next misclassification. The first time this step is executed the result is that the first misclassification is selected.
  • Step 506 Misclassification is a target hyperdimensional fragment?
  • the system operator or the application the system operator is executing determines if the misclassification is a target hyperdimensional fragment or should be a lead. The operator bases this decision on the operator's knowledge and previous experience. If the misclassification is a target hyperdimensional fragment, then the method 500 proceeds to step 508; otherwise the method 500 jumps to step 510.
  • Step 508 Identify most similar non-target HDF and use as contrast HDF.
  • the system operator or the application the system operator is executing checks the list of non-target hyperdimensional fragments and identifies one which is most similar to the target hyperdimensional fragment being studied. This selection becomes a contrast hyperdimensional fragment used for comparison and the method 500 skips to step 515.
  • Step 510 Identify most similar target HDF and use as contrast HDF.
  • the system operator or the application the system operator is executing checks the list of target hyperdimensional fragments and identifies one which is most similar to the non-target hyperdimensional fragment being studied. This selection becomes a contrast hyperdimensional fragment used for comparison.
  • Step 515 Visualize PDB and compare misclassification to the contrast.
  • the operator executes a visualization application to visualize and analyze the misclassification and contrast along with the PDB contents.
  • Step 520 Difference lies between feature cuts?
  • the system operator or the application the system operator is executing determines if the two examples have the greatest difference between feature cuts. Another way to accomplish the same objective is to determine if hyperdimensional fragments have the greatest difference at the feature level. If the greatest difference between the misclassification and contrast occurs between feature cuts, then the method 500 proceeds to step 540 of Figure 5b; otherwise the method 500 proceeds to step 525.
  • Step 525 Difference lies between pattern cuts?
  • the system operator or the application the system operator is executing determines if the two examples have the greatest difference between pattern cuts. Another way to accomplish the same objective is to determine if hyperdimensional fragments have the greatest difference at the pattern level. If the greatest difference between the misclassification and contrast occurs between pattern cuts, then the method 500 proceeds to step 560 of Figure 5c; otherwise the method 500 proceeds to step 580 of Figure 5d.
  • Step 542 Modify feature attribute decision surfaces to discriminate between the misclassification and contrast.
  • the system operator or the application the system operator is executing modifies the parameters used to define the decision surface in the template, which classifies between leads (targets) and non-leads (non-targets), so that it lies between the misclassification and contrast.
  • the computer application is instructed to perform the reclassification.
  • Step 544 Available feature attributes can discriminate?
  • the system operator or the application the system operator is executing determines if there are feature level attributes which might be able to discriminate between the misclassification and the contrast but were not placed in the PDB. If there are feature attributes which could discriminate, then the method 500 proceeds to step 546; otherwise the method 500 switches to step 548.
  • Step 546 Add a feature attribute to the PDB.
  • the system operator executes an application which modifies the PDB by adding a feature attribute.
  • Step 555 Step 548: Available feature statistics can discriminate?
  • the system operator or the application the system operator is executing determines if there are feature level statistics which might be able to discriminate between the misclassification and the contrast but were not placed in the PDB. If there are feature statistics which could discriminate, then the method 500 proceeds to step 550; otherwise the method 500 switches to step 552.
  • Step 550 Add a feature statistic to the PDB. In this step, the system operator executes an application which modifies the PDB by adding a feature statistic. When this step is completed method 500 proceeds to step 555.
  • Step 552 Add a feature attribute or statistic plug-in to the software.
  • the system operator defines an algorithm which discriminates between the misclassification and the contrast and adds the algorithm to the software application either by adding a software plug-in, as a dynamically linked library, or by modifying the application, recompiling, and re-linking.
  • Step 550 Add a feature statistic to the PDB.
  • the system operator executes the enhanced application which modifies the PDB by adding a feature attribute or statistic to the PDB.
  • Step 555 More misclassifications?
  • the system operator or the application the system operator is executing determines if there are more misclassifications in the misclassification list. If there are more misclassifications, then the method 500 returns to step 505 of Figure 5a; otherwise the method 500 switches to step 595.
  • Step 560 Patterns in the PDB can discriminate?
  • the system operator or the application the system operator is executing determines if the pattern level attributes in the PDB are capable of discriminating between the misclassification and the contrast. If the features can discriminate, then the method 500 proceeds to step 562; otherwise the method 500 switches to step 564.
  • Step 562 Modify pattern attribute decision surfaces to discriminate between the misclassification and contrast.
  • the system operator or the application the system operator is executing modifies the parameters used to define the decision surface in the template, which classifies between leads (targets) and non-leads (non-targets), so that it lies between the misclassification and contrast.
  • the computer application is instructed to perform the reclassification.
  • Step 564 Available pattern attribute discriminates?
  • the system operator or the application the system operator is executing determines if there are pattern level attributes which might be able to discriminate between the misclassification and the contrast but were not placed in the PDB. If there are pattern attributes which could discriminate, then the method 500 proceeds to step 566; otherwise the method 500 switches to step 568.
  • Step 566 Add a pattern attribute to the PDB.
  • the system operator executes an application which modifies the PDB by adding a pattern attribute.
  • this step is completed method 500 proceeds to step 575.
  • Step 568 Available pattern statistics can discriminate?
  • the system operator or the application the system operator is executing determines if there are pattern level statistics which might be able to discriminate between the misclassification and the contrast but were not placed in the PDB. If there are pattern statistics which could discriminate, then the method 500 proceeds to step 570; otherwise the method 500 switches to step 572.
  • Step 570 Add a pattern statistic to the PDB.
  • the system operator executes an application which modifies the PDB by adding a pattern statistic.
  • method 500 proceeds to step 575.
  • the system operator defines an algorithm which discriminates between the misclassification and the contrast and adds the algorithm to the software application either by adding a software plug-in, as a dynamically linked library, or by modifying the application, recompiling, and re-linking.
  • Step 574 Add a pattern statistic to the PDB.
  • the system operator executes the enhanced application which modifies the PDB by adding a pattern attribute or statistic to the PDB.
  • Step 575 More misclassifications?
  • the system operator or the application the system operator is executing determines if there are more misclassifications in the misclassification list. If there are more misclassifications, then the method 500 returns to step 505 of Figure 5a; otherwise the method 500 switches to step 595.
  • Step 580 Textures in the PDB can discriminate?
  • the system operator or the application the system operator is executing determines if the texture level attributes in the PDB are capable of discriminating between the misclassification and the contrast. If the textures can discriminate, then the method 500 proceeds to step 582; otherwise the method 500 switches to step 584.
  • Step 582 Modify texture attribute decision surfaces to discriminate between the misclassification and contrast.
  • step 594 the system operator or the application the system operator is executing modifies the parameters used to define the decision surface in the template, which classifies between leads (targets) and non-leads (non-targets), so that it lies between the misclassification and contrast.
  • the computer application is instructed to perform the reclassification.
  • Step 584 Available texture attribute discriminates?
  • the system operator or the application the system operator is executing determines if there are texture level attributes which might be able to discriminate between the misclassification and the contrast but were not placed in the PDB. If there are texture attributes which could discriminate, then the method 500 proceeds to step 586; otherwise the method 500 switches to step 588.
  • Step 586 Add a texture attribute to the PDB.
  • the system operator executes an application which modifies the PDB by adding a texture attribute.
  • this step is completed method 500 proceeds to step 594.
  • Step 588 Available texture statistics can discriminate?
  • the system operator or the application the system operator is executing determines if there are texture level statistics which might be able to discriminate between the misclassification and the contrast but were not placed in the PDB. If there are texture statistics which could discriminate, then the method 500 proceeds to step 590; otherwise the method 500 switches to step 592.
  • Step 590 Add a texture statistic to the PDB.
  • the system operator executes an application which modifies the PDB by adding a texture statistic.
  • method 500 proceeds to step 594.
  • Step 592 Add a texture attribute or statistic plug-in to the software.
  • the system operator defines an algorithm which discriminates between the misclassification and the contrast and adds the algorithm to the software application either by adding a software plug-in, as a dynamically linked library, or by modifying the application, recompiling, and re-linking.
  • Step 593 Add a texture statistic to the PDB.
  • the system operator executes the enhanced application which modifies the PDB by adding a pattern attribute or statistic to the PDB.
  • Step 594 More misclassifications?
  • the system operator or the application the system operator is executing determines if there are more misclassifications in the misclassification list. If there are more misclassifications, then the method 500 returns to step 505 of Figure 5a; otherwise the method 500 switches to step 595.
  • Step 495 Method 400 ends.
  • the method 500 ends generally at step 595.

Landscapes

  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Geophysics (AREA)
  • Quality & Reliability (AREA)
  • General Life Sciences & Earth Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Geophysics And Detection Of Objects (AREA)

Abstract

Un appareil, un système ainsi qu'un ensemble de procédés sont présentés pour l'élaboration de modèles pouvant être utilisés dans l'exploration et la production de gisements de pétrole et de gaz. Un ordinateur personnel (160) est utilisé pour exécuter des calculs sur des données connues (184) afin de générer des modèles hyperdimensionnels lesquels sont stockés dans une base de données de formes (186). Les modèles hyperdimensionnels sont générés dans une série en couches de phases de coupe, d'attribution et de statistique utilisant des données connues pour représenter des gisements de pétrole et de gaz. Les modèles hyperdimensionnels sont stockés dans une base de données de formes (188). Des procédés sont prévus pour identifier des erreurs de classification et aussi pour déterminer la qualité des modèles dans la base de données de formes.
EP03764505A 2002-07-12 2003-07-14 Construction d'un modele de reconnaissance de forme appliquee a l'exploration et a la production petrolieres Withdrawn EP1540506A4 (fr)

Applications Claiming Priority (7)

Application Number Priority Date Filing Date Title
US39596002P 2002-07-12 2002-07-12
US39595902P 2002-07-12 2002-07-12
US395959P 2002-07-12
US395960P 2002-07-12
US308860 2002-12-03
US10/308,860 US7162463B1 (en) 2002-07-12 2002-12-03 Pattern recognition template construction applied to oil exploration and production
PCT/US2003/021723 WO2004008338A1 (fr) 2002-07-12 2003-07-14 Construction d'un modele de reconnaissance de forme appliquee a l'exploration et a la production petrolieres

Publications (2)

Publication Number Publication Date
EP1540506A1 true EP1540506A1 (fr) 2005-06-15
EP1540506A4 EP1540506A4 (fr) 2008-07-09

Family

ID=30119069

Family Applications (1)

Application Number Title Priority Date Filing Date
EP03764505A Withdrawn EP1540506A4 (fr) 2002-07-12 2003-07-14 Construction d'un modele de reconnaissance de forme appliquee a l'exploration et a la production petrolieres

Country Status (4)

Country Link
EP (1) EP1540506A4 (fr)
AU (1) AU2003249051A1 (fr)
NO (1) NO20050791L (fr)
WO (1) WO2004008338A1 (fr)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8538702B2 (en) 2007-07-16 2013-09-17 Exxonmobil Upstream Research Company Geologic features from curvelet based seismic attributes

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001092973A (ja) * 1999-09-24 2001-04-06 Fujitsu Ltd 画像解析装置及び方法並びにプログラム記録媒体
US20020031268A1 (en) * 2001-09-28 2002-03-14 Xerox Corporation Picture/graphics classification system and method

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001092973A (ja) * 1999-09-24 2001-04-06 Fujitsu Ltd 画像解析装置及び方法並びにプログラム記録媒体
US20020031268A1 (en) * 2001-09-28 2002-03-14 Xerox Corporation Picture/graphics classification system and method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
MOTTL V ET AL: "Pattern recognition in spatial data: a new method of seismic explorations for oil and gas in crystalline basement rocks" PATTERN RECOGNITION, 2000. PROCEEDINGS. 15TH INTERNATIONAL CONFERENCE ON SEPTEMBER 3-7, 2000; [PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION. (ICPR)], LOS ALAMITOS, CA, USA,IEEE COMPUT. SOC, US, vol. 2, 3 September 2000 (2000-09-03), pages 315-318, XP010533818 ISBN: 978-0-7695-0750-7 *
See also references of WO2004008338A1 *

Also Published As

Publication number Publication date
EP1540506A4 (fr) 2008-07-09
AU2003249051A1 (en) 2004-02-02
WO2004008338A1 (fr) 2004-01-22
NO20050791L (no) 2005-02-14

Similar Documents

Publication Publication Date Title
US7162463B1 (en) Pattern recognition template construction applied to oil exploration and production
US7188092B2 (en) Pattern recognition template application applied to oil exploration and production
US7184991B1 (en) Pattern recognition applied to oil exploration and production
US7308139B2 (en) Method, system, and apparatus for color representation of seismic data and associated measurements
US8447524B2 (en) Method for geophysical and geological interpretation of seismic volumes using chronological panning
US20060184488A1 (en) Method and system for trace aligned and trace non-aligned pattern statistical calculation in seismic analysis
US20050288863A1 (en) Method and system for utilizing string-length ratio in seismic analysis
US7295706B2 (en) Pattern recognition applied to graphic imaging
US8219322B2 (en) Processing of stratigraphic data
Qi et al. Seismic attribute selection for machine-learning-based facies analysis
US8213261B2 (en) Method for geophysical and geological interpretation of seismic volumes in the domains of depth, time, and age
CA2481907C (fr) Procede d'analyse morphologique d'objets sismiques
US20130151161A1 (en) Seismic Horizon Skeletonization
US20050171700A1 (en) Device and system for calculating 3D seismic classification features and process for geoprospecting material seams
GB2444167A (en) Method for stratigraphic interpretation of seismic images
EP2496967A1 (fr) Procédé permettant de créer un modèle terrestre à strates hiérarchisées
WO2004008338A1 (fr) Construction d'un modele de reconnaissance de forme appliquee a l'exploration et a la production petrolieres
Lawal Seismic Faults Detection Using Saliency Maps

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

17P Request for examination filed

Effective date: 20050131

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IT LI LU MC NL PT RO SE SI SK TR

AX Request for extension of the european patent

Extension state: AL LT LV MK

DAX Request for extension of the european patent (deleted)
A4 Supplementary search report drawn up and despatched

Effective date: 20080605

RIC1 Information provided on ipc code assigned before grant

Ipc: G01V 11/00 20060101ALI20080530BHEP

Ipc: G06K 9/62 20060101AFI20080530BHEP

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: EXAMINATION IS IN PROGRESS

17Q First examination report despatched

Effective date: 20090206

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN

18D Application deemed to be withdrawn

Effective date: 20090203