Shared Spherical Image Processing for Resource-Constrained Action Cameras
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies, such as multiband blending for spherical content, are resource-intensive and typically confined to post-processing workstations due to high memory and computational complexity, making them unsuitable for embedded devices like action cameras.
Innovation Solution
The method involves shared image processing among multiple devices, where an action camera performs a partial multiband blend on spherical content and transmits the partially processed data to another device, which can then complete the blending or share the content as-is, depending on capabilities and usage considerations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If sophisticated image processing techniques like multiband blending are applied to spherical content, then image quality is improved, but memory and computational complexity increase making it unsuitable for embedded devices
Solution Approach 1:
The patent divides the image processing task into multiple segments: the action camera performs initial processing (stitching, exposure blending) on spherical content, then transmits the partially processed data structure to a second device which performs additional processing. This segmentation allows complex processing to be distributed across devices with different capabilities, enabling high image quality without requiring all processing to occur on resource-constrained embedded devices.
Solution Approach 2:
The action camera performs preliminary image processing operations (stitching overlapping regions, exposure blending, tone mapping) before transmission. This preliminary action reduces the computational burden on the second device while ensuring that critical processing is completed on the capture device, balancing image quality requirements with embedded device limitations.
2Manufacturing precision
If sophisticated image processing techniques like multiband blending are applied to spherical content, then image quality is improved, but memory requirements increase making it unsuitable for embedded devices
Solution Approach 1:
The patent extracts only the necessary processed data from the action camera and transmits it to the second device. Instead of performing all processing on the embedded device (which would require large memory), the system extracts partially processed spherical content data structures and transmits them, reducing memory requirements on the embedded device while still enabling high-quality processing on the second device.
3Manufacturing precision
If the action camera performs complete image processing, then image quality is maximized, but processing time and power consumption increase
Solution Approach 1:
The action camera performs a partial image processing operation (initial stitching and blending) rather than complete processing. This partial action reduces processing time and power consumption on the embedded device, while the second device completes the remaining processing operations, achieving high image quality without the full time and power cost of complete processing on the constrained device.
Data Source
AI summary
Methods and apparatus for shared image processing among multiple devices. In one embodiment, an exemplary action camera performs a partial multiband blend. Even though the action camera may not have resources to handle the multiband blend of the entire action camera's footage, it can do a significant portion. The partially blended content can be used in ready-to-share applications, or completely blended by another device.