Mixed Reality Position Estimation Using Segmented Search Ranges
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for measuring the position and orientation of an image sensing device in mixed reality systems face challenges, particularly when dealing with complex objects and environments where measurement line segments overlap, leading to incorrect edge correspondence and erroneous position and orientation estimation.
Innovation Solution
The technique involves projecting a virtual object onto the sensed image, setting search ranges for each side of the virtual object based on its positional relationship, searching for edges within these ranges, and calculating the position and orientation relationship between the virtual and physical objects using the correspondence between the projected sides and detected edges.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If measurement line segments with high density are used for arbitrary measurement target objects, then the application range is widened, but search areas overlap causing erroneous edge correspondence
Solution Approach 1:
The patent divides the search area into multiple sub-regions based on the spatial distribution of measurement line segments. By segmenting the search space and assigning specific sub-regions to individual line segments, the method prevents overlapping searches and ensures that each edge is correctly associated with its corresponding line segment, thereby maintaining reliability while handling high-density arbitrary objects.
Solution Approach 2:
The patent applies different search strategies to different regions of the image based on local characteristics. For areas where measurement line segments are densely distributed, restricted search areas are imposed, while other regions may use more flexible search methods. This local adaptation allows the system to maintain accuracy in critical overlapping regions while preserving versatility for various object types.
2Ease of manufacture
If search areas for all measurement line segments are processed independently, then the method is simple to implement, but process redundancy occurs reducing productivity
Solution Approach 1:
The patent merges the search processes for multiple measurement line segments by identifying and combining overlapping search areas. Instead of independently processing each line segment's search area, the method consolidates redundant search regions into single processing units, thereby reducing computational redundancy while maintaining the simplicity of the overall approach.
Solution Approach 2:
The patent introduces dynamic adjustment of search areas based on the spatial relationships between measurement line segments. The search region for each line segment is dynamically restricted to exclude areas already covered by other segments, creating an adaptive processing scheme that reduces redundancy while preserving implementation simplicity through automated region management.
3Quantity of substance
If the distance between measurement line segments is small, then more segments can be used for complex objects, but search areas overlap with high probability
Solution Approach 1:
The patent performs preliminary arrangement of search areas before the actual edge detection process. By pre-calculating and assigning distinct search regions to each measurement line segment based on their spatial relationships, the method ensures that even when line segments are densely distributed, their search areas remain distinct and non-overlapping, thereby maintaining reliability while allowing use of numerous segments.
Data Source
AI summary
A measurement line segment projection unit (400) projects, onto a sensed image, a three-dimensional model which is arranged at the position and orientation of a physical object (199). A search range is set for each side of the virtual object projected onto the sensed image, near the side of the virtual object in the sensed image based on a positional relationship between the side and other sides of the virtual object. A side of the physical object (199) on the sensed image is searched for within the search range for each side of the virtual object. The position and orientation relationship between the physical object (199) and an image sensing device (50) is calculated using the correspondence relationship, determined based on the search result, between each side of the three-dimensional model projected onto the sensed image and each side of the physical object (199) located on the sensed image.