Seismic Texture Classification Using Expert Feature Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The coherence of seismic lines in seismic reflection data often makes it difficult to determine the geological meaning of each seismic layer, hindering effective classification and interpretation of seismic textures.
Innovation Solution
A method and apparatus that utilize user input, such as sketches and natural language descriptions, to analyze seismic data, transforming this input into feature vectors that enhance the classification of seismic textures by embedding expert knowledge into computer vision algorithms, thereby improving the discrimination between texture classes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional seismic reflection analysis methods are used, then the process maintains simplicity and automation, but the classification accuracy and discrimination between seismic textures is insufficient due to unclear coherence of seismic lines
Solution Approach 1:
The patent introduces an intermediary system consisting of feature extraction modules and machine learning classifiers that bridge the gap between raw seismic data and geological interpretation. This intermediary layer processes the unclear seismic coherence by extracting multiple features (amplitude, frequency, wavelength, texture) and applying classification algorithms, thereby improving measurement precision without requiring direct manual interpretation of ambiguous seismic lines.
Solution Approach 2:
The patent segments the seismic analysis process into distinct computational stages: data preprocessing, feature extraction (amplitude, frequency, wavelength, texture), classification, and interpretation. By dividing the complex analysis into modular segments, the system achieves higher classification accuracy while managing complexity through organized, reusable components that can be independently optimized.
2Measurement precision
If manual interpretation of seismic coherence is attempted, then expert geological knowledge can be applied, but the process becomes time-consuming and less productive
Solution Approach 1:
The patent implements self-service through automated feature extraction and classification systems that perform seismic texture analysis without continuous human intervention. The machine learning models automatically process seismic data, extract relevant features, and generate classifications, enabling the system to serve itself in the analysis pipeline while maintaining high interpretation accuracy through algorithmic consistency and repeatability.
Solution Approach 2:
The patent applies preliminary action by pre-processing seismic data to enhance coherence and pre-extracting multiple features (amplitude, frequency, wavelength, texture) before classification. This preliminary preparation work is performed automatically on large datasets, reducing the time required for manual interpretation while maintaining or improving accuracy through comprehensive feature analysis.
3Measurement precision
If comprehensive feature extraction is performed to improve classification, then discrimination between seismic textures improves, but the computational complexity and processing time increases
Solution Approach 1:
The patent applies partial action by selectively extracting and analyzing the most discriminative features for each seismic texture classification task. Rather than uniformly processing all possible features across all data, the system identifies and focuses on the subset of features (amplitude, frequency, wavelength, texture) that provide the greatest discrimination value, thereby improving texture discrimination while reducing unnecessary computational overhead and processing time.
Data Source
AI summary
A method is provided, the method including: displaying an image on a display; detect a user input corresponding to one or more portions of the image; analyzing the user input to determine at least one feature vector corresponding to the user input; and determining a classification for the one or more portions of the image based at least on the at least one feature vector.