Multi-Homography Parameter Determination for Non-Planar Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for determining camera parameters fail when the object includes non-planar surfaces, as they are based on the premise of a planar object.
Innovation Solution
Determine N first and second regions in images captured from different positions by known optical devices, calculate homographies between these regions, and decompose them to determine parameters related to the optical devices and the object, even when the object has non-planar surfaces.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If homography decomposition is performed based on the premise of a planar object, then parameters related to the camera can be determined for planar objects, but the technique cannot determine parameters when the object includes non-planar surfaces
Solution Approach 1:
The patent divides the object into multiple planar regions (first regions in the first image, second regions in the second image) and performs homography decomposition for each region separately. By segmenting the object into multiple planar patches, the technique can handle non-planar surfaces by treating each patch as approximately planar, thus resolving the contradiction between maintaining planar object premise and extending to non-planar objects
Solution Approach 2:
The patent extends the parameter determination from a single 2D image plane to multiple 3D spatial regions by capturing images from different positions and identifying corresponding regions across multiple images. This multi-dimensional approach allows the technique to handle objects with complex 3D structures including non-planar surfaces
2Measurement precision
If multiple homographies are determined and decomposed to handle non-planar surfaces, then parameter determination accuracy for non-planar objects is improved, but the processing complexity increases
Solution Approach 1:
The patent combines multiple homography decomposition results from different planar regions to determine the final camera parameters. By merging the information from N different homographies, the technique achieves more accurate parameter determination for non-planar objects while systematically managing the processing complexity through unified mathematical frameworks
Data Source
AI summary
A parameter determination method includes: determining N, N being a natural number greater than 1, first regions included in at least a part of a first image acquired by capturing a target with a first optical device positioned at a first position and whose internal parameters are known, or the first image projected from the first optical device; determining N second regions corresponding to the respective N first regions in a one-to-one manner in a second image acquired by capturing the target with a second optical device positioned at a second position different from the first and whose internal parameters are known; determining N homographies between the N first regions and the N second regions in a one-to-one manner; and determining at least one of a first parameter related to the second optical device and a second parameter related to the capturing target by decomposing each of the N homographies.