18516799. SPATIAL CALIBRATION METHOD simplified abstract (CANON KABUSHIKI KAISHA)
Contents
SPATIAL CALIBRATION METHOD
Organization Name
Inventor(s)
GÉRALD Kergourlay of CHEVAIGNE (FR)
ERIC Nassor of THORIGNE-FOUILLARD (FR)
SPATIAL CALIBRATION METHOD - A simplified explanation of the abstract
This abstract first appeared for US patent application 18516799 titled 'SPATIAL CALIBRATION METHOD
Simplified Explanation
The abstract describes a method for spatial calibration of a first image against a second image by obtaining pairs of points of interest and iteratively determining a projection model between the two images.
- Method for spatial calibration of images:
- Obtain pairs of points of interest in the first and second images - Determine a projection model iteratively - Calibrate the first image based on the determined projection model
Potential Applications: This technology could be applied in fields such as computer vision, augmented reality, and image processing for accurate alignment of images.
Problems Solved: This technology solves the problem of accurately calibrating images in spatial alignment tasks, which is crucial for various applications such as image stitching and 3D reconstruction.
Benefits: - Improved accuracy in aligning images - Enhanced performance in computer vision tasks - Increased efficiency in image processing workflows
Potential Commercial Applications: "Enhancing Image Alignment for Augmented Reality Applications"
Possible Prior Art: One possible prior art could be the use of feature-based image alignment techniques in computer vision and image processing.
Unanswered Questions: 1. How does this method compare to existing image calibration techniques in terms of accuracy and efficiency? 2. Are there any limitations or constraints in the application of this method to real-world scenarios?
Original Abstract Submitted
According to some embodiments of the disclosure, it is provided a method for spatial calibration of a first image against a second image. After obtaining a plurality of pairs of points of interest, each pair matching a point of interest in the first image to a corresponding point of interest in the second image, the method further comprises determining, iteratively, a projection model between the two images and calibrating the first image based on the determined projected model. An iteration of the determining comprises displacing a position of a point of interest, determining a projection model candidate using multiple pairs of points of interest and evaluating a reprojection error between the positions of the points of interest as projected using the projection model candidate and the corresponding points of interest.
