Electronic Device Image Stitching with Object-Aware Combination Lines
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electronic devices that capture and process images, such as those for virtual reality or 3D applications, often generate stitching images with default combination lines that can overlap with objects of interest, causing distortion and inconvenience to users, requiring manual adjustment through separate programs.
Innovation Solution
An electronic device equipped with multiple image sensors and a processor that recognizes objects, determines depth information, and automatically adjusts the combination line based on object recognition and depth data to minimize overlap with objects of interest, thereby improving the stitching process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a default combination line is used in image stitching, then the stitching process is simple and fast, but the combination line may overlap with objects of interest causing distortion and user inconvenience
Solution Approach 1:
The system performs preliminary object recognition and depth estimation before finalizing the stitching process. By identifying objects of interest and their depth information in advance, the system can pre-determine an optimized combination line that avoids overlapping with these objects, thus preventing distortion before it occurs while maintaining efficient processing
Solution Approach 2:
The combination line is made dynamic rather than fixed. The system adjusts the combination line position based on real-time analysis of object locations and depth information from multiple image sensors. This dynamic adjustment allows the stitching process to adapt to different scenes and object configurations, avoiding distortion while maintaining processing efficiency
2Manufacturing precision
If the combination line is moved manually using a separate program, then the stitching image quality is improved, but the operation complexity and time consumption increase
Solution Approach 1:
The system performs self-service by automatically recognizing objects, estimating their depth, and determining the optimal combination line position without requiring user intervention. The processor analyzes image data from multiple sensors, identifies objects of interest, and autonomously adjusts the stitching parameters to avoid distortion, eliminating the need for manual adjustment while maintaining high image quality
Solution Approach 2:
The system uses feedback from object recognition and depth estimation results to automatically adjust the combination line position. By continuously analyzing the relationship between objects and potential combination lines, the system provides real-time feedback to optimize stitching quality without requiring user input or manual program intervention
3Adaptability or versatility
If multiple image sensors with overlapping fields of view are used, then panoramic and 3D image capabilities are enhanced, but the complexity of image stitching increases
Solution Approach 1:
The system introduces an intermediary processing layer that includes object recognition and depth estimation modules. These intermediaries bridge the gap between multiple image sensors and the final stitching process by providing structured information about objects and their spatial relationships, simplifying the complexity of combining images from multiple sensors with overlapping fields of view
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An electronic device is provided. The electronic device includes image sensors configured to include a plurality of pairs of image sensors, each pair of image sensors facing in substantially the same direction and generating image data and each of which has a different field of view, a processor configured to be operatively connected to the image sensors, and a memory configured to be operatively connected to the processor. The memory stores instructions, when executed, causing the processor to receive at least one set of image data from the pairs of image sensors, recognize at least one object from the image data, determine depth information of the at least one object, and stitch the at least one set of image data using the at least one object and the depth information.