Terrain Elevation Data Resolution Enhancement via Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mobile devices face data storage limitations, making it challenging to store and display high-resolution terrain elevation data, which is essential for detailed terrain mapping and situational awareness.
Innovation Solution
An apparatus that obtains low-resolution terrain elevation data and applies it to a resolution enhancing model, such as a machine learning model, to produce higher-resolution terrain elevation data, allowing for more detailed terrain mapping without the need to store high-resolution data on the device.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-resolution terrain elevation data is stored on mobile devices, then terrain mapping detail and situational awareness are improved, but data storage requirements increase beyond device capabilities
Solution Approach 1:
The patent creates a copy relationship where a small low-resolution dataset is transformed through a machine learning model to generate high-resolution terrain data on-demand. The model learns the mapping from low-resolution to high-resolution data during training, then applies this learned transformation during inference to produce detailed terrain maps without storing them permanently
Solution Approach 2:
The machine learning model is trained in advance using pairs of low-resolution and high-resolution terrain elevation data. This preliminary training phase allows the model to learn complex terrain patterns and relationships, so that during actual use, the model can quickly generate high-resolution data from low-resolution inputs without requiring the expensive high-resolution data to be pre-stored on the device
2Quantity of substance
If low-resolution terrain elevation data is used, then data storage requirements are reduced, but terrain mapping detail and situational awareness deteriorate
Solution Approach 1:
The patent changes the resolution parameter of the terrain data dynamically. The system stores data at low resolution to save space, then uses a machine learning model to transform the resolution parameter upward when high-detail terrain mapping is needed. This allows the same data to serve multiple purposes at different quality levels
3Measurement precision
If high-resolution terrain data is generated and displayed, then situational awareness and terrain mapping quality are improved, but device memory and processing requirements increase
Solution Approach 1:
Instead of storing and processing large amounts of high-resolution terrain data, the system uses a compact machine learning model that copies the essential terrain information from low-resolution data and reconstructs it at high resolution only when needed for display or analysis
Solution Approach 2:
The system dynamically adjusts the resolution of terrain data based on computational needs and display requirements. The machine learning model provides on-demand upscaling, allowing the system to maintain low-resolution stored data while generating high-resolution output temporarily when detailed terrain mapping is required, then discarding the generated high-resolution data after use
Data Source
AI summary
A terrain elevation data generation method includes: obtaining, at an apparatus, a low-resolution data set corresponding to a geographic area and comprising a plurality of low-resolution terrain elevation values corresponding to first locations within the geographic area; and applying, at the apparatus, the plurality of low-resolution terrain elevation values to a resolution enhancing model to produce a plurality of higher-resolution terrain elevation values corresponding to second locations within the geographic area, a second quantity of the second locations within the geographic area being higher than a first quantity of the first locations within the geographic area.