Reference Image Selection for Projective Image Stitching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image stitching technologies face challenges in selecting an optimal reference image, leading to potential distortion and poor visual quality in composite images due to perspective distortion, especially when user input is not ideal.
Innovation Solution
A computer-implemented method that selects a reference image based on minimizing overall distortion by estimating the image with the minimum maximal distance to overlapping images or calculating distortions before and after projective transformations, ensuring a visually pleasant layout and warning users of excessive distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If a fixed ordering method is used to select the reference image (e.g., first image in the set), then the automation extent is improved, but the manufacturing precision of the composite image deteriorates due to perspective distortion
Solution Approach 1:
The system automatically evaluates each image's suitability as a reference image by computing distortion metrics and selecting the optimal one without user intervention. The algorithm serves itself by autonomously determining which image minimizes overall distortion across the composite, eliminating the need for manual reference image selection while achieving superior results compared to fixed ordering methods
Solution Approach 2:
The system changes the selection criterion from simple fixed ordering to distortion-based evaluation. By computing distortion parameters for each potential reference image and selecting based on these parameters, the system transforms the reference image selection process from a rigid automated approach to an adaptive one that optimizes for visual quality
2Manufacturing precision
If user input is required to establish a reference image, then the manufacturing precision of the composite image is improved, but the ease of operation deteriorates
Solution Approach 1:
The system performs reference image selection autonomously by evaluating distortion metrics for each candidate image and automatically choosing the optimal reference image. This eliminates the need for user input while maintaining high manufacturing precision, as the algorithm independently determines the best reference image based on objective distortion calculations
Solution Approach 2:
The system provides feedback to users about the selected reference image and the expected distortion levels. By computing and communicating distortion metrics, the system allows users to verify the quality of automatic selection without requiring them to manually specify the reference image, thus maintaining ease of operation while ensuring manufacturing precision
3Productivity
If projective transformations are applied to join images, then the productivity of image stitching is improved, but the object-affected harmful factors increase due to perspective distortion
Solution Approach 1:
The system performs preliminary evaluation of distortion metrics for each potential reference image before applying projective transformations. By pre-calculating distortion parameters and selecting the optimal reference image in advance, the system minimizes harmful distortion effects while maintaining efficient automated stitching throughput
Solution Approach 2:
The system changes the selection parameter from arbitrary or user-specified reference images to distortion-optimized reference images. By selecting the reference image that minimizes overall distortion and adjusting projective transformations accordingly, the system reduces harmful distortion effects while preserving the efficiency benefits of automated projective transformation methods
Data Source
AI summary
The present disclosure includes systems and techniques relating to selecting a reference image for images to be joined in accordance with projective transformations. In general, one aspect of the subject matter described in this specification can be embodied in a computer-implemented method that includes obtaining projective transformations corresponding to two dimensional images to be joined together in accordance with the projective transformations; selecting one of the two dimensional images to be a reference image for remaining ones of the two dimensional images, the selecting being based on a measure of overall distortion for the two dimensional images; setting a projective transformation of the one of the images according to a group transform; correcting remaining projective transformations of the remaining images in accordance with the setting the projective transformation of the one of the images; and making the two dimensional images and the projective transformations available for further processing and output.