Super-Resolution Network for Panoramic Image Sharpness
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies, particularly super-resolution deep learning networks, are ineffective in enhancing the resolution of panoramic videos or images with a 360-degree field of view, leading to blurred playback on virtual reality devices due to insufficient feature extraction and resolution adjustment.
Innovation Solution
A super-resolution deep learning network model, specifically a super-resolution generative adversarial network (SRGAN), is trained and modified to increase the resolution of low-resolution panoramic images from 4K to 16K by adjusting feature filters from 3*3 to 5*5 pixels, and then decreasing the high-resolution images to medium-resolution for optimal display on virtual reality devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a super-resolution deep learning network model with preset feature filters is used to process panoramic images, then the processing speed is maintained, but the resolution enhancement quality is insufficient and distortion remains high
Solution Approach 1:
The patent applies local quality by differentiating feature filter configurations for different regions of the panoramic image. Specifically, the system uses different filter sizes (3x3 for edge regions, 5x5 for center regions) and different downscaling ratios (0.25 for edges, 0.5 for center) to match the local resolution requirements and distortion characteristics of different panoramic zones, thereby improving overall resolution enhancement quality without uniform complexity increase
Solution Approach 2:
The patent segments the panoramic image processing into distinct regions (edge regions and center regions) with different processing parameters. This segmentation allows the system to apply optimized feature filter configurations tailored to each region's specific requirements, resolving the contradiction between maintaining processing speed and improving resolution quality through region-specific optimization
2Measurement precision
If the feature filter resolution is increased from preset value to established value, then the image sharpness increases by 6%, but the computational complexity and processing time increase
Solution Approach 1:
The patent implements partial action by applying the higher-resolution 5x5 feature filters only to the center region of the panoramic image where maximum sharpness enhancement is needed, while using the more efficient 3x3 filters for edge regions. This selective application achieves the 6% sharpness improvement in critical areas without proportionally increasing processing time across the entire image
Solution Approach 2:
The patent changes the feature filter parameter (size) dynamically based on spatial position within the panoramic image. By transitioning from a uniform filter size to a position-dependent filter size (3x3 at edges, 5x5 at center), the system optimizes the balance between image sharpness improvement and processing efficiency through parameter adaptation
Data Source
AI summary
A method for processing an image that can be played on a virtual device, including obtaining a super-resolution deep learning network model, which is trained to learn to reconstruct an image from low resolution to high resolution; wherein the super-resolution deep learning network model includes a plurality of feature filters to extract features of the image; modifying the resolution of the feature filters from a preset value to an established value, wherein the established value is higher than the preset value; inputting a low-resolution image into the super-resolution deep learning network model; and increasing the resolution of the low-resolution image to become a high-resolution image through the super-resolution deep learning network model.