360-Degree Image Encoding Using Projection Rotation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for encoding and decoding 360-degree images face challenges in efficiently processing and reconstructing these images, particularly in virtual reality applications where all surrounding directions need to be accounted for, leading to difficulties in compression and rendering.
Innovation Solution
A method and apparatus that decode a projection image of a 360-degree image from image data, acquire rotation information, and reconstruct the original image by rotating it based on this information, while also encoding the image by determining a predetermined angle for rotation, converting the image into a projection image, and generating a bitstream with the encoded projection image and rotation information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If 360-degree images are encoded using conventional compression standards, then image data can be stored and transmitted, but the compression efficiency and processing speed are insufficient for high-resolution VR content
Solution Approach 1:
The patent divides the 360-degree image into multiple projection images (e.g., equirectangular, cubic, or spherical projections) that can be independently encoded and processed. This segmentation allows parallel processing of different projection views, improving compression efficiency while managing computational complexity through distributed processing of image portions.
Solution Approach 2:
The patent transforms the 360-degree spherical image into multiple 2D projection images representing different viewing angles and orientations. By converting the 3D spherical coordinate system into multiple 2D projections, the system enables efficient conventional compression algorithms to process VR content while preserving the ability to reconstruct the original 360-degree view.
2Manufacturing precision
If rotation information is added to enable accurate reconstruction of original 360-degree images, then rendering accuracy improves, but data transmission and storage requirements increase
Solution Approach 1:
The patent encodes rotation information as compact parameters (rotation angles, orientation values) that describe the spatial relationship between projection images and the original 360-degree view. These parameter changes allow accurate reconstruction without storing redundant full-resolution image data, significantly reducing the quantity of transmitted data while maintaining high reconstruction accuracy.
3Productivity
If multiple projection images are used to represent 360-degree views, then compression efficiency improves, but the complexity of decoding and reconstructing the original image increases
Solution Approach 1:
The patent introduces rotation information as an intermediary element that mediates between multiple projection images and the original 360-degree image. This intermediary contains the necessary transformation parameters (rotation angles, projection types) that guide the decoding process, enabling the system to reconstruct the original image from multiple projections without requiring complex iterative algorithms.
Data Source
AI summary
Provided are a method and apparatus for encoding or decoding a 360-degree image. According to an image decoding method and apparatus according to an embodiment, an original 360-degree image is reconstructed by acquiring image data from a bitstream, decoding a projection image of a 360-degree image from the image data, converting the projection image into the 360-degree image, acquiring rotation information of the 360-degree image from the bitstream, and rotating the 360-degree image based on the rotation information.