Panoramic Image Generation Speed Optimization via Adjacency Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing panoramic image generation methods for user terminals suffer from low mosaicking speed due to inefficient feature matching and camera parameter optimization processes.
Innovation Solution
A method and apparatus for panoramic image generation that includes determining adjacency relationships among images, performing feature matching using SIFT points and RANSAC, optimizing camera parameters with bundle adjustment, and adjusting image colors to improve mosaicking speed and quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If general mosaicking method using SIFT feature and bundle adjustment optimization is used, then image mosaicking accuracy is improved, but mosaicking speed deteriorates
Solution Approach 1:
The patent segments the image set into adjacent image pairs based on adjacency relationships, performing feature matching only on paired images rather than all images simultaneously. This reduces the computational complexity from O(n²) to O(n) while maintaining matching accuracy through sequential processing of adjacent pairs.
Solution Approach 2:
The patent performs preliminary determination of adjacency relationships and preliminary feature matching on adjacent images before executing the full bundle adjustment optimization. This preliminary action provides initial parameter estimates that accelerate the convergence of the optimization algorithm, reducing overall processing time while maintaining final accuracy.
2Productivity
If mosaicking method for mobile phone with sensor tracking is used, then mosaicking speed is improved, but mosaicking accuracy deteriorates
Solution Approach 1:
The patent merges two approaches: it uses sensor information to quickly identify adjacent image pairs (speed improvement) while simultaneously performing feature matching and bundle adjustment optimization on these pairs (accuracy improvement). This combination leverages the strengths of both methods to achieve both speed and accuracy.
Solution Approach 2:
The patent introduces feature matching results as an intermediary between sensor tracking and final mosaicking. The sensor data provides initial adjacency information, feature matching refines this with visual verification, and the refined parameters are then used for final bundle adjustment, creating a multi-stage mediation process that ensures both speed and accuracy.
3Reliability
If feature matching is performed on all image pairs, then matching completeness is improved, but computational workload increases
Solution Approach 1:
The patent segments the comprehensive feature matching task into smaller sub-tasks by only matching adjacent image pairs. This segmentation reduces the number of comparisons from all possible pairs to just neighboring pairs, dramatically reducing computational workload while maintaining reliability through the transitive property of adjacency relationships in the panorama sequence.
Data Source
AI summary
The present disclosure provides a panoramic image generation method and apparatus for a user terminal. The panoramic image generation method for a user terminal includes: obtaining a plurality of images shot by a user terminal, determining an adjacence relationship among the plurality of images, and performing feature matching on adjacent images to obtain matched feature-point-pairs; obtaining an optimized camera parameter according to the matched feature-point-pairs and an initial camera parameter; performing a color adjustment on the adjacent images to obtain color adjusted adjacent images; and mosaicking the color adjusted adjacent images according to the optimized camera parameter to generate a panoramic image.