Array Image Stitching With Adaptive Homography Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image stitching methods suffer from false matching of feature points due to factors like camera shooting angle and image similarity, leading to suboptimal deformation and fusion results, with no effective correction mechanisms.
Innovation Solution
An image stitching method that evaluates stitching results using homography matrices and adaptively adjusts matrices by selecting from initial or previously-stitched image pairs or columns, reducing errors through timely corrections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional image stitching methods are used to process images with large similarity or varying shooting angles, then the stitching process can be completed, but false matching of feature points occurs leading to poor stitching quality
Solution Approach 1:
The patent implements a feedback mechanism by evaluating stitching results and using the evaluation outcomes to guide subsequent stitching decisions. The system calculates evaluation values based on extracted features and uses these values to determine whether to proceed with stitching or select alternative image pairs, thereby improving matching accuracy and reducing false matches.
Solution Approach 2:
The patent performs preliminary evaluation of potential stitching pairs before actual stitching occurs. By calculating evaluation values in advance and selecting image pairs with suitable evaluation values, the system prepares optimal matching candidates beforehand, preventing false matches from occurring in the first place.
2Productivity
If image stitching is performed without evaluation and correction mechanisms, then the process is simple and fast, but stitching errors cannot be detected or corrected
Solution Approach 1:
The patent introduces a feedback loop where stitching results are evaluated and the evaluation outcomes feed back into the stitching decision process. This allows the system to detect errors and make corrections while maintaining efficient processing through automated decision-making based on evaluation values.
Solution Approach 2:
The system performs self-evaluation and self-correction by automatically calculating evaluation values for stitching results and using these values to guide subsequent stitching operations. The system serves itself by identifying and correcting its own errors without external intervention, balancing speed and accuracy.
3Adaptability or versatility
If homography matrices are calculated for every image pair without selection, then all image pairs can be stitched, but computational resources are wasted on poor matching pairs
Solution Approach 1:
The patent calculates evaluation values in advance for potential stitching pairs and uses these pre-calculated values to select which image pairs warrant full homography matrix calculation. This preliminary filtering action prevents wasteful computation on poor matching pairs while ensuring suitable pairs are processed.
Solution Approach 2:
The patent applies different processing levels to different image pairs based on their evaluation values. Image pairs with high evaluation values receive full homography matrix calculation and stitching, while those with low evaluation values are rejected or given minimal processing. This local differentiation of quality ensures computational resources are focused where they are most needed.
Data Source
AI summary
An image stitching method includes obtaining images arranged in an array; acquiring, in a pair-by-pair stitching process performed on the images by rows along a first direction, a first stitching homography matrix corresponding to each image pair, wherein, the stitching result of the same image pair is evaluated based on the first stitching homography matrix corresponding to the same image pair, and if the stitching result does not meet a predetermined condition, the same image pair is stitched by utilizing the first stitching homography matrix corresponding to the image pair of the same column in a previously-stitched row, or a first initial homography matrix corresponding to the column, where the same image pair is located, is acquired and utilized to stitch the same image pair; and if the images contain multiple rows, further stitching the images by columns along a second direction to generate a stitched image.