Magnetic Map Artifact Detection Using Autoencoder and GAN Repair
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Geological and geophysical field maps often contain artifacts due to differences in map creation technologies and times, leading to manual efforts that are time-consuming and labor-intensive for experts to identify and fix.
Innovation Solution
Utilizing machine learning models, specifically autoencoders and generative adversarial networks, to identify and propose fixes for map artifacts, allowing experts to prioritize and refine the suggested corrections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual methods are used to identify and fix map artifacts, then experts can detect and correct artifacts, but the process is time-consuming and labor-intensive
Solution Approach 1:
The patent replaces manual mechanical inspection methods with machine learning models (autoencoders and GANs) that automatically detect and propose fixes for map artifacts. The system processes magnetic maps through neural networks to identify artifact regions and generate correction suggestions, eliminating the need for time-consuming manual review while maintaining or improving detection accuracy.
2Area of stationary object
If multiple regional maps are combined to create comprehensive maps, then coverage is improved, but artifacts from different creation technologies and times are introduced
Solution Approach 1:
The patent extracts and isolates artifact regions from combined maps using autoencoder-based anomaly detection. The machine learning model identifies regions that deviate from expected patterns caused by artifacts, separating these problematic areas from the valid map data. This allows the system to preserve the comprehensive coverage benefits while removing the reliability-damaging artifacts that result from combining multiple regional maps.
3Manufacturing precision
If experts manually review and fix each artifact, then map accuracy is improved, but the process requires significant labor and expertise
Solution Approach 1:
The patent implements a self-service system where machine learning models automatically detect map artifacts and generate correction proposals without requiring expert intervention for each artifact. The autoencoder and GAN-based system performs self-diagnosis of map quality issues and provides remediation suggestions, reducing both the labor requirements and the complexity of the correction process while maintaining high accuracy through iterative model training on expert-validated data.
Data Source
AI summary
Example computer-implemented methods and systems for detecting and fixing map artifacts are disclosed. One example computer-implemented method includes obtaining a magnetic map of a region. One or more areas in the magnetic map are determined based on a first machine learning model, where each of the one or more areas includes one or more artifacts in the magnetic map. The one or more areas are determined by generating, based on the first machine learning model and the magnetic map, a second map by the first machine learning model, and comparing the second map with the magnetic map. The one or more artifacts in each of the one or more areas are removed from the magnetic map.