Image Stitching via Rigid and Non-Rigid Registration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image stitching technologies face content discontinuities due to deformation information from different acquisition parameters, leading to poor continuity in combined images.
Innovation Solution
A system and method that perform rigid and non-rigid registrations on overlapping image sections to generate corrected images, which are then combined using weight functions to improve content continuity and image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If multiple images with different acquisition parameters are combined using traditional stitching technology, then the coverage area of the combined image is expanded, but content discontinuities occur due to deformation information differences
Solution Approach 1:
The patent performs preliminary registration (both rigid and non-rigid) on the images before combining them. This preliminary alignment process corrects deformation information differences caused by varying acquisition parameters, ensuring that content continuity is maintained while still achieving expanded coverage area through the combination of multiple images.
Solution Approach 2:
The patent transforms the deformation information parameters from different acquisition conditions into a unified reference frame. By applying non-rigid registration that adjusts local deformation parameters, the system reconciles differences in acquisition parameters while preserving the expanded coverage benefit of combining multiple images.
2Productivity
If rigid registration is performed on overlapping image sections, then alignment efficiency is improved, but local deformation information is lost
Solution Approach 1:
The patent segments the registration process into two distinct stages: rigid registration for global alignment and non-rigid registration for local refinement. The rigid registration stage provides efficient overall positioning, while the subsequent non-rigid registration stage corrects local deformation information, thus achieving both high efficiency and high precision without compromising either aspect.
Solution Approach 2:
The patent introduces non-rigid registration as an intermediary step between rigid registration and final image combination. This intermediary process bridges the gap by taking the efficiently aligned result from rigid registration and refining it with detailed local deformation corrections, thereby preserving both efficiency gains and precision requirements.
3Reliability
If non-rigid registration is performed on all images, then content continuity is improved, but computational complexity increases
Solution Approach 1:
The patent segments the application of non-rigid registration to only the overlapping regions of images rather than processing entire images. This segmentation approach maintains content continuity in the critical overlapping areas where deformation differences matter most, while significantly reducing computational complexity by avoiding redundant processing of non-overlapping regions.
Solution Approach 2:
The patent applies non-rigid registration selectively to local overlapping regions rather than uniformly across all images. This local quality approach ensures that computational resources are concentrated where they are most needed (in overlapping areas requiring continuity) while reducing overall complexity by excluding non-overlapping regions from intensive processing.
4Manufacturing precision
If weight functions are used to combine corrected images, then image quality in overlapping regions is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary corrections on individual images before the combination stage, preparing them in advance for efficient merging. This preliminary action includes applying the weight functions to determine contribution levels of each image in overlapping regions beforehand, so that the actual combination process requires minimal additional computation, thus improving image quality while limiting time loss.
Data Source
AI summary
The present disclosure may provide a system. The system may obtain a first image of a first region of a subject and a second image of a second region of the subject. The first region and the second region may have an overlapping region. The system may generate a first corrected image by correcting the first image based on the second image and a second corrected image by correcting the second image based on the first image. The system may generate a combined image by combining the first corrected image and the second corrected image.