Automated Boundary Detection in Low Contrast Remote Sensing Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining boundaries of features in low contrast remote sensing images are inefficient, requiring manual digitization by skilled analysts and semi-automated solutions that fail due to non-uniform contrast across object boundaries.
Innovation Solution
A boundary determination system that performs principal component analysis on spectral signatures of pixels in a selected area, generating a boundary based on the analysis, identifying additional pixels within an ellipsoid, and dilating the boundary to include these pixels without exceeding an inflection point, thereby reducing human intervention and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual digitizing by skilled analysts is used, then boundary identification accuracy is maintained, but time consumption and labor intensity increase significantly
Solution Approach 1:
The patent replaces the manual mechanical process of analyst digitizing with an automated computational system. The system uses principal component analysis on spectral signatures to automatically identify and delineate building boundaries, substituting human visual inspection and manual tracing with algorithmic image processing and spectral analysis.
Solution Approach 2:
The patent transforms the boundary identification problem from a visual contrast-based approach to a spectral signature-based approach. By analyzing spectral characteristics and using principal component analysis, the system identifies boundaries through spectral variations rather than relying on visual contrast, enabling automated processing while maintaining accuracy.
2Productivity
If semi-automated imagery analysis is used, then processing speed increases, but accuracy deteriorates in low contrast conditions
Solution Approach 1:
The patent changes the analysis parameter from visual contrast to spectral signature characteristics. By using principal component analysis on spectral data, the system can identify boundaries in low contrast conditions where traditional visual methods fail, maintaining both automation and accuracy.
Solution Approach 2:
The patent moves the analysis from the spatial domain (visual contrast in 2D images) to the spectral domain (multi-dimensional spectral signatures). This dimensional transformation allows the system to exploit spectral variations that are not visible in standard imagery, enabling accurate boundary detection in low contrast conditions.
3Device complexity
If traditional boundary detection methods are used, then simplicity is maintained, but they fail to handle non-uniform contrast across object boundaries
Solution Approach 1:
The patent replaces simple contrast-based detection with spectral signature analysis using principal component analysis. This parameter change enables the system to handle non-uniform contrast by relying on spectral characteristics that remain consistent even when visual contrast varies across different parts of an object boundary.
Data Source
AI summary
A boundary determination system for determining boundaries of features in a digital image is provided. The boundary determination system includes a processor coupled to a memory. The processor is configured to analyze a spectral content and a spatial structure of a feature in a digital image, characterize the feature in the digital image, and distinguish the feature from immediate surroundings of the feature in the digital image.