360-Degree Image Decoding with Selective Reference Expansion

The patent divides the 360-degree image into multiple partitioning units (e.g., cube faces, rectangular regions) and processes each unit separately with appropriate encoding strategies. This segmentation allows the system to handle the massive data of 360-degree images in manageable chunks, improving compression performance while maintaining organized processing complexity

The patent transforms 360-degree spherical images into different projection formats (e.g., cube map projection, equirectangular projection) before encoding. This dimensional transformation enables conventional 2D encoding techniques to be applied effectively to 3D spherical data, significantly improving compression performance while the projection format selection provides flexibility to manage processing complexity

The patent performs image expansion only on specific partitioning units or regions that require prediction, rather than expanding the entire reference picture. This selective expansion approach maintains prediction accuracy for critical regions while significantly reducing processing time and computational complexity

The patent applies different expansion strategies to different regions based on their importance and characteristics. High-priority regions with motion or prediction requirements receive full expansion treatment, while static or less important regions use simplified or no expansion, optimizing the balance between prediction accuracy and processing efficiency