Inverse Stereo Image Matching for Terrain Change Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for change detection in stereo image data often require constraining collection geometry or performing time-consuming and error-prone ortho-rectification, which introduces spatial and spectral distortions, making it difficult to accurately detect real terrain changes without manual editing.
Innovation Solution
The system performs epipolar rectification on stereo images to produce rectified data, followed by hybrid stereo matching and digital model generation, predicting areas where stereo image matching should fail due to underlying terrain, thereby identifying real terrain changes without ortho-image data or precise geometry constraints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If ortho-rectification is performed on stereo images, then geometric accuracy is improved, but spatial and spectral distortions are introduced that produce erroneous change results
Solution Approach 1:
The patent extracts only the essential rectification needed for change detection by using epipolar geometry-based matching without performing full ortho-rectification. This removes the harmful distortions while retaining sufficient geometric correction for accurate change detection between stereo images.
Solution Approach 2:
Instead of rectifying images to a common map coordinate system (traditional ortho-rectification), the patent inverts the approach by matching images directly in their original coordinate systems using epipolar geometry, thereby avoiding the introduction of spatial and spectral distortions while still achieving accurate change detection.
2Loss of time
If ortho-rectification is performed without compensation for topology, then processing time is reduced, but differential layover distortions persist that defeat change detection
Solution Approach 1:
The patent introduces epipolar geometry as an intermediary framework that enables accurate stereo matching without requiring full ortho-rectification or topology compensation. This intermediary approach achieves both efficiency and reliability by working directly with the stereo image pair's inherent geometric relationships.
3Measurement precision
If re-projection is performed to remove layover distortions, then change detection accuracy is improved, but pixel distortion fundamentally changes spatial and spectral character
Solution Approach 1:
The patent extracts only the necessary geometric correction for change detection by using epipolar matching, removing the harmful pixel distortions introduced by full re-projection while retaining sufficient accuracy for detecting real terrain changes.
4Productivity
If epipolar rectification is used instead of ortho-rectification, then computational intensity is reduced, but some geometric correction is still needed
Solution Approach 1:
The patent applies partial rectification through epipolar geometry-based matching, which provides sufficient geometric correction for change detection without the excessive computational burden of full ortho-rectification. This partial action achieves the necessary precision while maintaining high productivity.
Data Source
AI summary
A system and method for finding real terrain matches in a stereo image pair is presented. A method for finding differences of underlying terrain between a first stereo image and a second stereo image includes performing epipolar rectification on a stereo image pair to produce rectified image data. The method performs a hybrid stereo image matching on the rectified image data to produce image matching data. A digital surface model (DSM) is generated based on the image matching data. Next, the method identifies areas in the DSM where the stereo image matching should fail based on the image matching data and the DSM to generate predicted failures. The method can then determine real terrain changes based on the predicted failures and the image matching data.