Dynamic Panorama Image Stabilization and Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional panorama shooting methods struggle to generate videos with a wide angle of view, as they require the shooter to remain stationary, leading to artifacts when the device moves, and are unable to create videos by stitching still images.
Innovation Solution
An electronic device with a processor and camera that stabilizes images using an algorithm, generates basic extended images by overlaying stabilized images, and creates inference images using an inference model to produce a final video with a larger angle of view than the original images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of moving object
If conventional panorama shooting is used to create wide angle images, then the angle of view is extended, but artifacts occur when the device moves or the shooter moves
Solution Approach 1:
The patent transitions from static panorama shooting (requiring fixed position) to dynamic panorama shooting that adapts to device movement. The system dynamically adjusts the stitching process based on detected device displacement, allowing the shooter to move while maintaining image quality and avoiding artifacts through real-time correction of the stitching algorithm.
Solution Approach 2:
The patent implements feedback by detecting device movement (through sensors or image analysis) and using this information to correct the panorama stitching process. The system monitors displacement during shooting and adjusts the alignment and blending of individual images accordingly, preventing artifact formation while maintaining extended angle of view.
2Area of moving object
If still images are stitched to create panorama, then wide angle view is achieved, but video with wide angle of view cannot be created
Solution Approach 1:
The patent creates a universal panorama generation system that can produce both still images and videos with wide angle views using the same underlying technology. By processing sequences of images through the stabilized stitching algorithm, the system generates video outputs alongside traditional image panoramas, making the solution applicable to multiple output formats and use cases.
3Ease of operation
If dynamic panorama shooting is used to allow shooter movement, then ease of operation is improved, but artifacts occur due to device displacement
Solution Approach 1:
The patent implements dynamic panorama shooting that adapts to shooter movement in real-time. The system detects device displacement during capture and dynamically adjusts the stitching parameters, allowing the shooter to move freely while maintaining image quality. This dynamic adaptation resolves the contradiction between shooting flexibility and image quality by making the system responsive to actual shooting conditions.
Data Source
AI summary
An electronic device is provided. The electronic device includes at least one processor, and a memory functionally connected to the at least one processor. The memory may store instructions that, when executed, enable the electronic device to obtain a plurality of images, generate a first basic extended image based on first images among the plurality of images, identify at least one first masking area included in the first basic extended image, and generate a first inference image by modifying the at least one first masking area using at least one first inference result, based on the first images and the first basic extended image. An angle of view of the first inference image may be larger than an angle of view of each of the first images.