Grid-Based Plane Estimation for Augmented Reality Boundary Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for plane estimation in images, particularly in augmented reality applications, face challenges in accurately identifying and representing planar surfaces, leading to suboptimal interaction between virtual and real objects.
Innovation Solution
A processor-implemented method that divides an image into grid cells, classifies feature points as inliers or outliers, selects boundary grid cells based on their distribution, and estimates an expanded plane to accurately identify planar surfaces, enabling precise placement of virtual objects in augmented reality environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional plane estimation methods are used, then the process is simpler, but the accuracy of plane boundary identification deteriorates
Solution Approach 1:
The image is divided into multiple grid cells, and feature points are extracted and classified within each grid cell. This segmentation allows the system to process local regions independently, improving boundary identification accuracy while maintaining manageable computational complexity through localized analysis.
Solution Approach 2:
The patent introduces a grid cell dimension overlaying the image space, creating a two-level structure (image space + grid space). This dimensional transformation enables systematic classification of feature points and more accurate plane boundary detection by organizing data in a structured grid framework.
2Measurement precision
If feature points are densely extracted, then the plane estimation accuracy improves, but the computational load increases
Solution Approach 1:
By dividing the image into grid cells and processing feature points within each cell separately, the system can extract sufficient feature points for accurate estimation without needing to process every possible point in the entire image. This reduces overall computational load while maintaining estimation accuracy.
Solution Approach 2:
The patent extracts feature points selectively within grid cells rather than uniformly across the entire image. This partial action approach focuses computational resources on relevant regions, achieving accurate plane estimation without the excessive computational burden of processing all image points.
3Adaptability or versatility
If the plane estimation covers a larger area, then the utility in AR applications improves, but the difficulty of accurate boundary detection increases
Solution Approach 1:
The grid cell segmentation enables the system to handle large areas by breaking them into manageable local regions. Each grid cell's boundary detection is performed independently with high accuracy, and the results are combined to form the complete plane boundary, thus maintaining detection accuracy across large areas.
Solution Approach 2:
The introduction of grid cell coordinates as an additional dimensional layer provides a structured framework for detecting and measuring boundaries across large areas. This grid-based coordinate system simplifies the complexity of boundary detection in extended regions by providing regular sampling points and systematic organization.
Data Source
AI summary
A processor-implemented method of estimating a plane includes dividing a target image into a plurality of grid cells, estimating an initial plane based on reference feature points comprising a reference feature point selected from each of the plurality of grid cells, classifying each feature point of feature points extracted from the target image as being an inlier located on the same plane as the initial plane or an outlier located on a plane different from the initial plane, selecting boundary grid cells from the plurality of grid cells based on a distribution of the inliers and the outliers from among the feature points, and estimating an expanded plane associated with the initial plane based on the selected boundary grid cells for use in identifying boundaries of a planar surface presented in a scene corresponding to the target image.