Homography Estimation via Multithreshold Edge Contours
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional augmented reality systems face challenges in consistently determining the location and orientation of real-world objects due to ambiguous object shapes, varying lighting conditions, and maintaining stable virtual object positioning in real-time video feeds.
Innovation Solution
The method involves capturing digital images of target objects and generating contours at multiple brightness threshold levels to estimate a homography, which reduces variability under changing lighting conditions and stabilizes the positioning of virtual augmentations by averaging contour estimates across different exposure levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If a single brightness threshold is used for contour detection, then the processing speed is fast, but the positioning stability deteriorates under varying lighting conditions
Solution Approach 1:
The patent segments the contour detection process by generating contours at multiple brightness threshold levels simultaneously. Instead of using a single threshold, the system divides the detection into multiple parallel operations at different threshold levels (e.g., first threshold, second threshold, third threshold), then combines the results to achieve stable positioning under varying lighting conditions.
Solution Approach 2:
The patent changes the brightness threshold parameter to multiple discrete levels. By detecting contours at different brightness threshold values and combining these detections, the system achieves robustness against lighting variations while maintaining processing efficiency through parallel computation.
2Reliability
If multiple brightness threshold levels are used for contour detection, then the positioning stability improves, but the processing complexity increases
Solution Approach 1:
The patent merges the contour detection results from multiple brightness threshold levels into a single unified homography estimation. By combining the contour instances detected at different threshold levels, the system achieves improved positioning stability while managing complexity through integrated processing.
Solution Approach 2:
The patent uses a plurality of brightness threshold levels (excessive action) to ensure robust detection under all lighting conditions. By detecting contours at more threshold levels than a single threshold would provide, the system guarantees stable homography estimation even when lighting conditions cause some thresholds to fail, with the complexity managed through efficient parallel processing.
3Use of energy by moving object
If conventional single-threshold contour detection is used, then the computational load is low, but the homography estimation accuracy deteriorates under varying lighting
Solution Approach 1:
The patent ensures continuous useful action by detecting contours at multiple brightness threshold levels simultaneously. This continuous multi-level detection approach maintains accurate homography estimation under varying lighting conditions, as at least some threshold levels will produce reliable contours regardless of lighting changes, with the computational load distributed across parallel operations.
4Ease of manufacture
If contours are generated at multiple brightness threshold levels, then the need for calibration is reduced, but the processing time increases
Solution Approach 1:
The patent performs preliminary action by generating contours at multiple brightness threshold levels during the image processing stage itself, rather than requiring separate calibration procedures. This preliminary multi-level contour generation is integrated into the real-time processing pipeline, eliminating the need for separate calibration steps while managing processing time through efficient parallel computation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of analysing a captured image, comprising the steps of capturing a digital image comprising an instance of a target object for each of a plurality of different brightness threshold levels, generating contours from the captured digital image that indicate where in the captured digital image the pixel values of the captured digital image cross the respective brightness threshold level identifying instances of a contour corresponding to a characteristic feature of said target object for at least two of the respective brightness threshold levels and estimating a homography of the characteristic feature of the target object of the captured image based upon the two or more instances of its corresponding contour.