Real-Time Panorama Stitching on Portable Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital cameras struggle to stitch high-resolution panoramic images in real-time due to computational intensity, requiring external computers and delaying user access to the final image.
Innovation Solution
A method for generating panoramic images on portable devices using low-resolution frames to determine optimal stitch lines, aligning and blending high-resolution frames with alpha blending maps, and performing sub-pixel registration to prevent pixel offsets, allowing for real-time processing and storage of high-resolution panoramas without user intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-resolution image frames are stitched together to create panoramic images, then the quality and resolution of the panorama is improved, but the computational time and processing power required increases dramatically
Solution Approach 1:
The patent divides the high-resolution image processing task into segments by first processing low-resolution versions to determine stitch lines and alignment parameters, then applying these parameters to the high-resolution images. This segmentation allows the computationally intensive alignment calculations to be performed on smaller data sets, reducing overall processing time while maintaining high output resolution.
Solution Approach 2:
The patent performs preliminary processing by determining optimal stitch lines and alignment transformations using low-resolution preview images before committing to the full high-resolution stitching process. This preliminary action identifies the correct geometric relationships and blend parameters in advance, so that when high-resolution images are stitched, the computationally heavy alignment calculations have already been completed on smaller data.
2Area of stationary object
If multiple high-resolution component images are stitched together, then the field of view and coverage area is improved, but the computational complexity and resource requirements worsen
Solution Approach 1:
The patent segments the stitching computation into two phases: first determining geometric relationships (transformations, stitch lines) from low-resolution images, then applying these pre-determined relationships to high-resolution images. This reduces the computational complexity of processing multiple large images by separating the geometric analysis from the final high-resolution composition.
Solution Approach 2:
The patent introduces low-resolution preview images as an intermediary step between capturing high-resolution component images and creating the final high-resolution panorama. These intermediary low-res images serve as proxies for determining alignment and stitch parameters, reducing the direct computational burden on the high-resolution data while still achieving accurate final results.
3Productivity
If real-time stitching is implemented on portable devices, then user accessibility and immediacy is improved, but the processing power and computational resources required worsen
Solution Approach 1:
The patent segments the stitching workflow into immediate low-resolution processing (for determining stitch lines and alignment) and deferred or optimized high-resolution processing. This allows portable devices to perform the critical path calculations quickly on smaller data, enabling real-time or near-real-time panorama generation without requiring sustained high processing power throughout the entire stitching process.
Solution Approach 2:
The patent performs preliminary determination of stitch lines and transformation parameters using low-resolution images before the final high-resolution stitching. This preliminary action on smaller data enables portable devices to prepare all necessary geometric and blending parameters in advance, reducing the real-time computational burden during actual high-resolution panorama creation and making real-time processing feasible on mobile hardware.
Data Source
Figure 1
Figure 2~3
Figure 4~5
AI summary
A technique of generating a panoramic image involves acquiring a set of at least two main image frames of overlapping portions of a scene. A map or other information is stored relating to the joining of the image frames. A main panorama image is formed by joining main image frames based on the map or information gained in a similar low-res process.