Multi-View Video Atlas Patch Transforms for Lower Decoding Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing immersive video systems face challenges in optimizing the compression and decoding of multi-view videos, particularly in reducing the computational complexity and resource requirements for devices due to inefficient patch arrangement and processing in atlases, which affects the quality of immersive experiences.
Innovation Solution
A method for coding and decoding multi-view videos that involves applying transformations such as oversampling, sub-sampling, and pixel value modifications to patches within atlases to optimize their arrangement and reduce the number of pixels processed, using syntax elements to indicate transformations, and predicting transformation parameters to minimize signaling costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If patches are extracted from multiple views and arranged in atlases for intermediate view synthesis, then the quality of immersive experience is improved, but the computational complexity and resource requirements for decoding increase
Solution Approach 1:
The patent extracts only the necessary patches from original views and arranges them in atlases, rather than transmitting or processing complete views. This selective extraction reduces the amount of data that needs to be processed during decoding while maintaining the quality needed for intermediate view synthesis, thereby improving immersive experience without proportionally increasing computational complexity.
Solution Approach 2:
The patent segments the video content into discrete patches that are arranged in atlases. This segmentation allows the decoder to process only relevant patch regions rather than entire views, reducing computational load. The patches are organized in a structured atlas format that enables efficient access and processing during the view synthesis process.
2Reliability
If all original views are transmitted to the decoder, then the quality of intermediate view synthesis is improved, but the bandwidth and data transmission requirements increase
Solution Approach 1:
Instead of transmitting all original views, the patent extracts only the essential patches from each view and packages them into compact atlases. This extraction approach transmits only the minimum necessary data required for high-quality intermediate view synthesis, significantly reducing bandwidth requirements while maintaining synthesis quality.
Solution Approach 2:
The patches are pre-arranged in atlas structures during encoding, organizing the extracted patch data into an optimized format before transmission. This preliminary organization reduces the data volume that needs to be transmitted and prepares the data for efficient decoding and synthesis operations at the receiver端.
3Productivity
If patches are arranged in atlases with optimal configuration, then the compression efficiency is improved, but the device complexity for arranging and managing patches increases
Solution Approach 1:
The patent segments patch arrangements into standardized atlas structures with defined configurations. This segmentation creates a systematic approach to organizing patches that improves compression efficiency through regular patterns, while the standardized nature of the segmentation reduces the complexity of managing and arranging patches during encoding and decoding.
Data Source
AI summary
A method of decoding a coded data stream representative of a multi-view video. The stream includes coded data representative of at least one atlas corresponding to an image having at least one patch. The patch corresponds to a set of pixels extracted from at least one component of a view of the multi-view video, the view not being coded in the coded data. The decoding includes: decoding, from the coded data stream, the atlas, including decoding the patch; determining, for the decoded patch, whether a transformation, and which transformation must be applied to the decoded patch, the transformation including at least a modification of pixel values of the decoded patch, the modification including applying a mapping to the pixel values to reduce a dynamic range of the pixel values, and applying the determined transformation to the decoded patch.